Information to Users
INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter fiice, while others may be from any type o f computer printer.
The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book.
Photographs included in the original manuscript have been reproduced xerographically in this copy. Ifigher quality 6” x 9” black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. UMI A Bell & Howell Iiifonnation Company 300 North Zeeb Road, Ann Arbor MI 48106-1346 USA 313/761-4700 800/521-0600
NATIONAL AND LOCAL MEDIA COVERAGE AND THE PRESIDENTIAL NOMINATION CAMPAIGN OF 1992
DISSERTATION
Presented in Partial Fulfillment of the Requirements for
the Degree Doctor of Philosophy in the Graduate
School of The Ohio State University
By
Audrey A. Haynes, M.A
*****
The Ohio State University 1996
Dissertation Committee: Approved by Professor John H. Kessel, Adviser
Professor Paul Allen Beck Adviser Professor Herbert Asher Department of Political Science
Professor Janet Box-Steffensmeier UMI Number: 9710578
UMI Microform 9710578 Copyright 1997, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized copying under Title 17, United States Code.
UMI 300 North Zeeb Road Ann Arbor, MI 48103 ABSTRACT
This study investigates three questions related to national and local news media: first, what was the volume, direction and focus of national and local campaign coverage in the 1992 presidential election campaign? Second, what explains the differences or similarities in the way the national and local news media cover the campaign? Finally, does national and local news media coverage influence primary and caucus outcomes?
Using content analysis, data was gathered from ABC World News Tonight broadcasts, the front pages of three national daily newspapers and 21 state/metropolitan daily newspapers from December 1, 1991 to June 3, 1992.
The major findings of the study are the following: local coverage is different from national coverage; the antecedents of local coverage are different than those of national coverage; and local coverage appears to have a greater impact on vote than national coverage. Moreover, there is extensive policy coverage for the in-party campaign. This dissertation is dedicated to two special individuals without whose help and encouragment this study would never have been completed:
James Vernon Chin, my husband and best friend and Professor John H. Kessel, whose guidance over the years has been greatly appreciated.
in ACKNOWLEDGMENTS
I wish to thank my advisor and chair. Professor John H. Kessel for all of his advice, guidance and scholarly inspiration.
I would also like to express my heartfelt gratitude to Professors Paul Allen Beck,
Janet Box-SteflFensmeier, Herb Asher and Herb Weisberg, all of whom have been exceptional teachers and professional role-models.
I wish to credit my peers in the Department of Political Science’s Graduate
Program, particularly Tanya Charlick-Paley, Steve Nichols, Rosie Clawson, Staci Rhine,
Laura Luehrmann, Jennifer Segal and Greg Strizek for their support during this process, and Retta Semones for all of her patient answers.
Finally, my mom, Mili Haynes, deserves my thanks for her support and words of encouragement.
IV VITA
February 12, 1965 ...... Bom, Colorado Springs, Colorado
1987 ...... Journalism, University of Georgia
1991...... M.A, Political Science, University of Georgia
1991-1994...... Graduate Teaching and Research Assistant, The Ohio State University
1994-Present...... Instructor, Georgia State University
PUBLICATIONS
Research Publications
1. A A Haynes, P.H. Gurian, and S. Nichols, “The Role of Candidate Spending in Presidential Nomination Campaigns.” Journal o fPolitics 59:1 (1997).
2. H. Weisberg, A A Haynes, J. Krosnick. 1995. “Social-Group Polarization in 1992.” In Feast for Democracy, ed. H. Weisberg (New Jersey: Chatham House).
3. P.H. Gurian and A.A. Haynes, “Campaign Strategy in Presidential Nominations, 1976-88.” American Journal o f Political Science 37:1 (1993)
FIELDS OF STUDY
Major Field: Political Science TABLE OF CONTENTS
Page
Abstract...... ü
Dedication ...... üi
Acknowledgments ...... iv
Vita...... V
List of Tables...... x
List of Figures...... -xvi
Chapters;
I. All Politics is Local... News Coverage and Presidential Nomination Campaigns...... 1
1.1 Researching Media Coverage and Presidential Election Campaigns...... 6 1.1.1 Early Studies of Media Coverage Content ...... 9 1.1.2 The Emergence of Television: Studies of the Mass Media Campaign...... 12 1.1.3 Expanding the Focus: Media Studies in the 1980s and 1990s ...... 14 1.2 General Theories of Media Coverage Content ...... 20 1.3 Generalizability of Media Study Findings: The Question of Mass Media Homogeneity ...... 24 1.4 The Nature of Presidential Nominations ...... 26 1.4.1 Structural Properties of the Nomination Campaign ...... 27 1.4.2 Candidate Performance of the Candidates...... 30 1.4.3 Candidate Resources ...... 32 1.4.4 Candidate Characteristics ...... 34 1.5 The News Media and the Presidential Nomination Campaign: Plan of the Study ...... 36 1.5.1 What Moves Media Coverage at the National and Local Level?... 38
VI 1.5.2 Does Media Coverage Move Outcomes?...... 41 1.6 Conclusion ...... 42
Data and Design ...... 45
2.1 Content Analysis as a Research Tool ...... 45 2.2 News Media Sources to be Analyzed ...... 47 2.3 Units of Analysis ...... 55 2.4 Coding Procedures ...... 57 2.5 Conclusion ...... 66
An Overview of the Events of the 1992 Democratic and Republican Presidential Nomination Campaigns ...... 66
3.1 Early Days ...... 66 3.2 Initial Contests: Setting the Campaign in Motion ...... 70 3.3 Winnowing and Mist-Clearing Stage ...... 80 3.4 Pre-convention and General Election Maneuvering Stage ...... 99 3.5 Conclusion: The Media and the Campaign ...... 101
National and Local Media Compared ...... 103
4.1 Media Homogeneity ...... 103 4.2 Media Coverage of the Campaign: “Attention” Differentiation Among Outlets...... 106 4.2.1 National Media Attention to the Campaign ...... 107 4.2.2 Non-national Media Attention to the Campaign ...... Ill 4.2.3 National and Non-national Campaign Attention Compared 115 4.3 Coverage of Candidates: Amount ...... 121 4.3.1 National Media Outlets ...... 121 4.3.2 Non-national Media Outlets ...... 124 4.3.3 National and Non-national Outlets Compared to the New York Times...... 126 4.4 Coverage of Candidates: Direction ...... 129 4.4.1 National and Non-national Direction of Coverage ...... 130 4.5 Cases of News Media Variation ...... 136 4.5.1 Paul Tsongas: Coverage of a Candidate’s Defeat ...... 136 4.5.2 Brown: Victory Coverage Across Outlets...... 139 4.6 Coverage of Candidates: Content ...... 142
4.6.1 As the Nomination Campaign Stages Change so Does the Coverage...... 148 4.6.2 Candidates and Their Issue Coverage ...... 151
VII 4.7 Summary and Implications ...... 168
5. What Moves News Media Coverage of the Presidential Nomination Campaign? ...... 172
5.1 The Changing Face of Local News Coverage of Presidential Campaigns...... 172 5.2 The Politics of Local and National Media Coverage of Candidates; Theoretical Questions ...... 175 5.3 Data...... 180 5.4 Predicting General Trends in Media Coverage ...... 181 5.4.1 Early Contests ...... 181 5.4.2 Critical Contest Weeks ...... 185 5.5 National and Local Pre-Event Coverage ...... 189 5.6 Possible Factors Influencing Local Media Pre-Event Coverage of Candidates ...... 191 5.6.1 Candidate Attention ...... 191 5.6.2 Candidate Relative and Absolute Spending ...... 194 5.6.3 Performance Factors ...... 195 5.6.4 Home State and Home Region Affiliation ...... 198 5.6.5 Competitive Field ...... 200 5.6.6 National Attention ...... 202 5.6.7 Local Competitive Status ...... 203 5.7 Toward a Model of Local Candidate Coverage ...... 204 5.8 Possible Factors Influencing the Direction of Local Media Coverage of the Candidates ...... 208 5.8.1 Incumbent Deference ...... 210 5.8.2 National Verdict ...... 211 5.8.3 Candidate Attention ...... 214 5.8.4 Home State and Region ...... 215 5.8.5 Performance Factors ...... 217 5.8.6 Candidate Field Size ...... 219 5.8.7 Scandal ...... 220 5.7.8 Local Competitive Status ...... 221 5.9 Toward a Model of Local Directional Coverage ...... 222 5.10 Factors Influencing National Coverage ...... 227 5.11 Factors Influencing the Direction of National Media Coverage.. 230 5.11.1 Incumbency ...... 231 5.11.2 Performance Factors ...... 232 5.11.3 Competitive Field ...... 234 5.11.4 National Polls ...... 235 5.11.5 Scandal ...... 237 5.12 Toward a Model of National Directional Coverage ...... 237
vm 5.13 Pre-event Substance Coverage by the National and Local Media ...... 243 5.14 Discussion and Implications ...... 247
6. The Political Consequences of National and Local Media Coverage...... 252 6.1 The Intermediary Model of Presidential Nomination Campaign Outcomes...... 255 6.2 Bivariate Hypothesis Testing...... 257 6.3 Media Coverage and Vote Outcome ...... 264 6.3.1 The Democratic and Republican Campaigns...... 270 6.4 Conclusion ...... 279
7. National and Local Media and the Nomination Process ...... 276 7.1 Summary of Findings ...... 276 7.1.1 National and Local New Media Coverage; Homogeneity Revisited ...... 276 7.1.2 What Moves National and Local News Media Coverage ...... 281 7.1.3 The Political Impact of Local and National News Media Campaign Coverage...... 284 7.2 The Question of Media “Bias” ...... 286 7.3 Normative Concerns: The Media and Presidential Nominations .291 7.4 Conclusion ...... 294
Appendices ...... 296 A Candidate Local and National Coverage Time Trends ...... 296 B Coverage Areas and Amounts by Party Over Time Periods ...... 311 C Variable L ist...... 325 Bibliography...... 329
IX LIST OF TABLES
Table Page
2.1 Circulation and Readership of National Newspapers ...... 53
2.2 Circulation and Readership of Core Newspapers ...... 53
2.3 Circulation and Readership of Additional Newspapers ...... 54
2.4 Content Category Breakdown of News Media Coverage ...... 61
4.1 National Newspaper’s “Attention” to the Campaign: Percentage and Number of Front Page and Lead Campaign Stories ...... 108
4.2 ABC’s World News Tonight's “Attention” to the Campaign Percentage and Number of Campaign Story Segments and Lead Segments...... 109
4.3 Local Newspaper’s “Attention” the Campaign Percentage and Number of Lead Campaign Stories ...... 110
4.4 Local Newspaper’s “Attention” to the Campaign Percentage and Number of Front Page Campaign Stories...... 112
4.5 Campaign Coverage Similarities Among National and Local News Outlets...... 118
4.6 Comparison of Media Coverage Measures Story Count Versus Word Count ...... 120
4.7 National News Media’s Candidate Attention Average Candidate Weekly National Coverage Amounts and Rankings ...... 123
4.8 Local News Media’s Candidate Attention Average Candidate Weekly Local Coverage Amounts and Rankings ...... 124
4.9 National Coverage Differences Relative to the New York Times ...... 127 4.10 Non-National Coverage Dififerences Relative to the New York Times...... 128
4.11 Directional Coverage of Democratic Candidates: Negative and Positive Coverage Percentages...... 131
4.12 Directional Coverage of Republican and Independent Candidates: Negative and Positive Coverage Percentages...... 132
4.13 The Directional Coverage Balance The Difference between Positive and Negative Coverage Percentages ...... 135
4.14 National and Non-National News Coverage of the Tsongas Suspension (March 19 and 20) ...... 138
4.15 National and Non-National Coverage of Brown after His Coimecticut Primary Victory (March 24/25)...... 141
4.16 Substance of the Coverage for the ABC World News Tonight...... 150
4.17 New York Times Issue Coverage of Candidates ...... 152
4.18 Washington Post Issue Coverage of Candidates ...... 153
4.19 Wall Street Journal Issue Coverage of Candidates ...... 154
4.20 ABC World News Tonight Issue Coverage of Candidates ...... 155
4.21 Des Moines Register Issue Coverage of Candidates ...... 156
4.22 Manchester Union Leader Issue Coverage of Candidates ...... 157
4.23 Bangor Daily News Issue Coverage of Candidates ...... 158
4.24 Salt Lake Tribune Issue Coverage of Candidates ...... 159
4.25 Atlanta Journal Constitution Issue Coverage of Candidates ...... 160
4.26 Miami Herald Issue Coverage of Candidates ...... 161
4.27 Chicago Tribune Issue Coverage of Candidates ...... 162
4.28 Milwaukee Sentinel Issue Coverage of Candidates ...... 163
XI 4.29 Newport News Daily Press Issue Coverage of Candidates ...... 164
5.1 The Relationship of News Media Outlet Coverage and Campaign Period ...... 184
5 .2 The Relationship of News Media Outlet Coverage and Critical Contest Weeks...... 187
5.3 Pre-event National Coverage Means by Candidate ...... 191
5.4 Pre-event Local Coverage Means by Candidate ...... 191
5.5 The Relationship of Candidate Attention and Spending with Local News Media Coverage ...... 194
5.6 The Relationship of Candidate Performance and Local News Media Coverage ..197
5.7 The Relationship of Home State and Home Region and Local News Media Coverage...... 200
5.8 The Relationship Between Candidate Field Size and Local News Media Coverage...... 202
5.9 The Relationship Between National News Media Coverage Levels and Local News Media Coverage ...... 203
5.10 Model of Local News Media Coverage: Democrats ...... 205
5.11 Model of Local News Media Coverage: Republicans ...... 206
5.12 Incumbent Deference by the Local News Media ...... 211
5.13 The Relationship Between National Directional Coverage and Local Directional Coverage...... 212
5.14 The Relationship Between Candidate Attention and Local News Outlet Positive Coverage ...... 215
5 .15 The Relationship of Home State and Home Region and Local News Outlet Positive Coverage...... 217
5.16 The Relationship Between Candidate Performance and Local Directional Coverage...... 218
XU 5.17 The Relationship Between Candidate Field Size and Local Directional Coverage...... 220
5.18 The Relationship Between Scandal and Local Directional Coverage ...... 221
5.19 The Relationship Between State-Level Competitive Status and Local Directional Coverage ...... 227
5.20 Model of Local Positive Coverage: Democratic and Republican Campaigns ...... 223
5.21 Model of Local Negative Coverage: Democratic and Republican Campaigns ....226
5.22 The Relationship Between Performance and National News Outlet Coverage .. 228
5.23 National Coverage Differences Between the Incumbent, Frontrunners and Challengers...... 228
5.24 Model of National News Media Coverage: Democratic and Republican Campaigns...... 230
5.25 National News Media Outlet Negative Coverage of Incumbent President Bush ...... 232
5.26 The Relationship Between Candidate Performance and National News Media Outlet Directional Coverage...... 233
5.27 The Relationship Between Candidate Field Size and National News Media Outlet Directional Coverage ...... 235
5.28 The Relationship Between National Poll Standing and National News Media Outlet Directional Coverage ...... 236
5.29 The Relationship Between Scandal and National News Media Outlet Directional Coverage: Democrats Only ...... 237
5.30 Model of National News Media Outlet Positive Coverage ...... 240
5.31 Model of National News Media Outlet Negative Coverage ...... 242
5.32 Pre-event Issue Coverage of Democratic Candidates: National and Local News Media Outlets Compared ...... 245
Xlll 5.33 Pre-event Issue coverage of Republican Candidates: National and Local News Media Outlets Compared ...... 246
6.1 The Relationship of Local News Media Outlet Coverage and Vote Share Controlling for National News Media Outlet Coverage ...... 258
6.2 The Relationship of National News Media Outlet Coverage and Vote Share Controlling for Local News News Media Outlet Coverage ...... 259
6.3 The Relationship Between Overall Direction of Coverage and Vote ...... 259
6.4 The Relationship Between Candidate Resource Investment and Vote Shares by Party and Candidate ...... 262
6.5 The Relationship Between State Ideology and Candidate Vote Shares ...... 263
6.6 The Relationship Between Home State and Home Region and Vote Shares 264
6.7 The Relationship Between National and Local News Media Coverage (Amount and Direction) and Candidate Vote Shares ...... 266
6.8 Local News Coverage (Amount and Direction) Variables and Their Relationship to Candidate Vote Share: Democratic and Republican Campaigns ...... 267
6.9 National News Coverage (Amount and Direction) Variables and Their Relationship to Candidate Vote Share: Democratic and Republican Campaigns...... 268
6.10 Model of Candidate Vote Share: Democratic Campaign ...... 270
6.11 Model of Candidate Vote Share: Republican Campaign ...... 273
B. 1 Substance of the Coverage for the New York Times...... 312
B.2 Substance of the Coverage for the Washington Post ...... 313
B.3 Substance of the Coverage for the Wall Street Journal ...... 314
B.4 Substance of the Coverage for the ABC World News Tonight ...... 315
B.5 Substance of the Coverage for the Des Moines Register ...... 316
B.6 Substance of the Coverage for the Manchester Union Leader...... 317
XIV B.7 Substance of the Coverage for the Bangor Daily News ...... 318
B.8 Substance of the Coverage for the Atlanta Journal Constitution ...... 319
B.9 Substance of the Coverage for the Salt Lake Tribune ...... 320
B. 10 Substance of the Coverage for the Miami Herald ...... 321
B.ll Substance of the Coverage for the Chicago Tribune ...... 322
B. 12 Substance of the Coverage for the Milwaukee Sentinel ...... 323
B. 13 Substance of the Coverage for the Newport News Daily Press ...... 324
XV LIST OF nOURES
Figure Eâgê
4.1 National News Media’s Campaign Attention Levels: Percentage of National Media Campaign Coverage Per Month ...... 110
4.2 Local News Media’s Campaign Attention Levels: Percentage of Local Media Campaign Coverage Per Month ...... 113
4.3 Combined Republican and Democratic Campaign Coverage by Outlet ...... 143
4.4 Republican Campaign: Percentage of Campaign Coverage Devoted to Coverage Area by News Media Outlet...... 144
4.5 Democratic Campaign: Percentage of Campaign Coverage Devoted to Coverage Area by News Media Outlet...... 146
7.1 Dynamic Model of Presidential Nomination Vote Outcomes ...... 286
A 1 Brown National Coverage Time Trends ...... 297
A.2 Brown Local Coverage Time Trends ...... 298
A3 Clinton National Coverage Time Trends ...... 299
A4 Clinton Local Coverage Time Trends ...... 300
A.5 Harkin National Coverage Time Trends ...... 301
A.6 Harkin Local Coverage Time Trends ...... 302
A.7 Kerrey National Coverage Time Trends ...... 303
AS Kerrey Local Coverage Time Trends ...... 304
A.9 Tsongas National Coverage Time Trends ...... 305
XVI A. 10 Tsongas Local Coverage Time Trends ...... 306
A. 11 Bush National Coverage Time Trends ...... 307
A 12 Bush Local Coverage Time Trends ...... 308
A 13 Buchanan National Coverage Time Trends ...... 309
A. 14 Buchanan Local Coverage Time Trends ...... 310
xvu CHAPTER I
ALL POLITICS IS LOCAL. . .
NEWS COVERAGE AND PRESIDENTIAL NOMINATION CAMPAIGNS
Everyone who follows politics is aware that the news media are important in presidential nominations in the United States where voters are the mainstays of decision making rather than party “bosses.” Without the information and analysis provided by the news media, Americans would probably have even less knowledge of the nomination campaign and the candidates seeking their party’s nomination. While much scholarly attention has been focused on assessing the potential bias of the news media’s coverage and determining the news media’s influence on presidential general elections, we know comparatively less about the media’s roles in presidential nominations and in sub- presidential electoral events. While our knowledge of the national media is still limited, we know significantly less about the local news media. Yet given their place as the main arbiters of information at the local level, these state and metropolitan news outlets may play potentially critical roles in an important electoral process.
Federalism is often discussed as a political concept, but the media system is federal as well. It incorporates a national media system that deals most often with the “great and aggregate” events, while the state and metropolitan news media deal with the “particular.”
Yet the boundaries between these two types of media are permeable, particularly when the object of coverage is both national and local in nature. The presidential nomination process exemplifies the situation of both a national and localized event.
Therefore, there are three reasons to explore local news coverage and its role in the nomination phase of presidential elections. First, local news is considered Americans’ most important source of information, yet local news coverage is relatively unexplored area, particularly with regard to political information and the implications of this information. A recent Gallup survey found that local television newscasts (57 percent) followed closely by newspapers (53 percent) were considered extremely or very important to them personally as a source of news, compared with 50 percent who said the same about network news. More than 55 percent of those surveyed said they got their news very regularly fi'om newspapers, 57 percent fi’om local television stations, and 59 percent from the nightly network news programs. Moreover, of those who said they followed the news closely, 57 percent said they considered newspapers as extremely or very important sources of news. Thus, any examination of the influence of the news media should attempt to account for the information available from both the national news media as well as the local news media as it is likely that the public utilizes each for news information, and potentially relies more on local news sources. (P.2-7, Gallup Monthly. September 1995)
Second, while local news may cover the same events as the national news, suggesting a high degree of homogeneity, the way local and national news organizations report, rank and interpret the events may differ. Local conditions may fi-ame the national events. The national news media often ignore the more specific of these conditions due to
the broad nature of their audiences. Moreover, if we examine the stories the news media
present to their audiences, we would expect there to be variation between national and
local outlets due to the a varied characteristics and informational needs of their audiences.
For example, while the defense budget might be important in Newport News, Virginia,
home to naval ship manufacturers, the Haitian refugee situation is more important in
Miami, Florida. What the incumbent president, his opponents, or the other party’s candidates has to say about an issue that is important to the local populace is likely to be printed in the prominent pages of the local press or broadcast on the local news. In other words, the reporting priorities of the two types of outlets may vary.
Third, state and metropolitan news coverage of the campaign will likely have a
“local” flavor as the primary and caucus events are more provincial in nature. Thus, stories may be centered toward state-level or regional issues and political cultures and contexts.
They may focus on state-level candidate activities rather than reporting on what the candidate is doing in another part of the country. This means that the infomiation presented may diverge firom the coverage presented by the national news organizations due to state-level characteristics and candidate-level responses to these characteristics.
These differences may have implications for the outcomes of individual primaries and caucuses and the choice of the eventual nominee.
Presumably, the news media system, represented by the network and local news, national and local newspapers, and national magazines are the mediators of information about the candidates, the events, the outcomes, and the voters’ responses. The news media also create their own “news information” in the form of verdict, expectations, and
general campaign analysis. Nimmo and Combs (1989) suggest that individual’s
knowledge of politics is rarely the product of “direct involvement.” Instead, the public’s
knowledge consists of perceptions, “focused,” “filtered,” and “fantasized” by a host of
mediators. Graber concedes this point as well. “However extensive their political
experiences and contacts may be, the bulk of political information to which they are
exposed and which they absorb is beyond their personal experiences. It comes from the
mass media either directly through personal exposure or indirectly through people with
whom they talk about current affairs.” (Graber 1994; 1)
One major assumption of conventional wisdom is that news coverage of candidates
is directly related to their success or failure. In other words, coverage of a candidate
before a particular event can have an influence on the outcome of the event. The Reagan
era left many with the impression that dominant images and themes could win support and
thus elections. The empirical base supporting this assumption, however, is still rather
weak. At the individual (voter) level, studies have had difiSculty determining whether the
media actually affect vote choice or attitude change. (See Bartels’ 1993 study for a
successful example of when this has been accomplished.) Aldrich’s (1980) treatise on strategies and choices in presidential nomination campaigns comes the closest to a comprehensive model of candidate vote outcomes. At the candidate level, Aldrich formalized the relationships between candidate behavior, the dynamics of the nomination campaign (which includes national media coverage of the campaign) and the primary and caucus outcomes. However, his theoretical model has yet to be fully tested. The evidence from those who actually take part in the campaign gives more support to the news media’s importance to candidate success. During the presidential nomination campaign, candidates indicate that they care about what the media say about the campaign. Party and candidate activists are often quoted as saying that editors and reporters are important players in the campaign process. Candidates believe that the focus and content of media coverage affect their chances of winning. Campaign operatives spend a substantial amount of time trying to manipulate their own or their competition’s
“expectations” or at least attempt some type of “spin control.” At minimum, the campaign attempts to focus the media’s attention on some issue or topic that benefits their candidate or hurts their competition. Journalists themselves often examine potential biases in their own coverage. All of this, at least at face value, suggests that there is the “perception” of media influence by candidates, their supporters, their operatives, and by the press itself.
Moreover, this influence occurs both ways. While the news media may influence the campaign, the other players in the process can influence the news media as well.
The nominating process provides us with a valuable institutional framework in which to explore the question of local and national news media influence. It offers multiple state-level contests that may be studied individually. Moreover, the nomination campaign allows us to assess change over time by examining particular variables over the length of the entire campaign or during particular periods within the campaign. Finally, the nomination campaign is such that the nature of the competition can change, the issues can change, the candidate field can alter dramatically, candidate resources and other variables can change. The questions of if how and when the local news may be associated with these changes in comparison to the national news can be pursued and the implications of these relationships can be explored.
While there has been relatively little research about local news in particular, studies of news media coverage of the presidential general election are available. These studies and the theories they have contributed to will be reviewed. Particular attention will be paid to those studies of presidential nomination campaigns and local news media.
Finally, the plan of the study will be presented.
1.1 Researching Media Coverage and Presidential Election Campaigns
Social scientists have explored the news media generally in two ways. First, scholars have focused on the message. What is the nature of the content? Is it biased in how much attention it gives to certain candidates? Does it focus on particular types of information more than others? Does the message vary between different media outlets?
Second, does the message have an effect? Is there a relationship between media message and subsequent behavior, preference or attitude either from an individual or group of individuals?
The first studies of media content and its potential effects were long on assumption and short on proof. During this period of study at the dawning of the 20th century, the research was very impressionistic since public opinion studies had not been developed.
Lippmarm’s work in the 1920s could be so classified (Lippman 1922). As new technologies arose in mass communication, scholars became interested in what effect such advances might have for “persuasion” or more specifically “manipulation.” Radio and then television were viewed at the outset as incredibly powerful tools that could be
misused for financial or political gain. This belief that mass communication could be used
for inordinate levels of influence was heightened even more by the use of propaganda by
Nazis in Germany and Fascists in Italy. The leaders of these movements were quite direct
in their use of mass communication as a conduit to the public. Social scientists during this
time approached the issue of mass media messages and influence with the basic
assumption that political propaganda was very influential, and that its influence stemmed from the actual rhetorical power of the message.
The second period of media research began when scholars systematically sought to uncover contingencies in media impact, often emphasizing public opinion change or voting behavior. When could media messages make a difference? When were they ineffective?
During this period the study of media effects produced a range of findings aided by more sophisticated data gathering and data analyzing techniques. Early work in the area included the “minimal effects” theory that argued that mass communications reinforced either prior attitudes or preferences or activated latent ones (Lazarsfeld et al. 1948; Katz and Lazarsfeld 1955; Klapper I960). The public that acted as the consumer of media messages was depicted as a relatively immune body — one that was not steered easily astray by political propaganda.
In more recent research efforts, scholars have proposed a range of limited and conditional effect views (Cook et al. 1983; Iyengar and Kinder 1987, 1982; McCombs and
Shaw 1972; McGuire 1985; McKuen 1984; Neuman 1986; Patterson and Davis 1985;
Patterson 1976; Robinson and Sheehan 1983; Weaver et al. 1981). Yet some work has appeared that suggests media messages are potential sources for influence in the political arena both at the individual and mass behavior level (Bartels 1993; Entman 1989; Iyengar and Ansolabehere 1995; Page, Shapiro and Dempsey 1987; Patterson 1980). Moreover, when media researchers have shifted their focus fi'om political persuasion and manipulation to cognitive effects they have found political learning effects ( Chaffee, Zhao and Leshner 1994; Weaver and Drew 1991; Weaver and Drew 1995; Zhao and Chaffee
1995). The generally accepted view, at least in the area of voting behavior, is that the media can matter, but their impact is dependent on the interaction between message and audience (Graber 1993).
In addition to the individual level and aggregate change models of media influence, there is a strain of research focused on investigating the content of the message. Often this type of research is geared toward determining if the news media show bias toward particular candidates (generally how much coverage volume has been distributed to the candidates) (Evarts and Stempel 1974; Graber 1976, 1971; Gurian and Barthlow 1991;
Malaney and Buss 1979). Other research has examined the foci and frameworks of media coverage (for example, whether the media concentrate primarily on a type of coverage - horse race, issue, personality and so on (Brady and Johnston 1987). Studies also have investigated media coverage to reveal some aspect of candidate-level behavior (Aldrich
1980) or media behavior toward particular types of candidates (Hagen 1992). 1.1.1 Early Studies of Media Coverage Content
The mapping work on presidential elections has often focused on presidential
general election coverage and pursued the question of news media “partisan” bias. For
example, Higbie’s study looked at the 1952 presidential election. He focused his attention
on the press treatment given to Adlai Stevenson, the Democratic party nominee, and
Dwight Eisenhower, the Republican nominee, by 14 Wisconsin daily newspapers from the
parties’ conventions to general election day. Higbie compared their coverage by
determining how much front page coverage they received and how this coverage was
presented. For example, was there a bold, large headline with accompanying pictures or
simply a short story near the bottom of the paper in small print? He measured headline
size and type, story size and placement, and photograph size and location. He found that
Stevenson and Eisenhower were treated almost equally with regard to the amount of news
space they were each given on the front page of the Wisconsin dailies. However,
Eisenhower’s name and countenance were more prevalent in the headlines and
photographs due in part to the popular appeal of the general.
There were two such studies of the 1956 presidential election coverage. Both were focused on potential biased reporting by the press. Markham and Stempel (1957) examined the level of coverage in 24 daily newspapers in Pennsylvania. They measured story space by colunm inches, frequency of story, the direction of the campaign issues, and the headline size and prominence. They coded each newspaper’s contents from October 1 to election day in November of 1956. While they found a certain degree of variation among individual newspapers on the amount and type of coverage given to the two parties’ nominees, Eisenhower and Stevenson again, the overall treatment was found to be fair.
Repass and Chaffee (1968) pursued the question of partisan press bias in their study of the 1956 presidential general election. In their examination of partisan newspapers in Wisconsin, one Democratic and seven Republican, they found that the
Republican presses appeared to favor the incumbent Eisenhower by a 16 percent margin.
The single Democratic leaning paper favored Stevenson by much less - a 5 percent margin.
Yet they found that if one looked only at strictly “campaign” coverage of Eisenhower rather than including coverage of his presidential administrative acts, then the Republican partisan press did not seem as favorable.*
Stempel’s 1961 study focused on coverage of the two party nominees by the
“prestige press.”" He analyzed the coverage from September 4 through November 7, 1960 by categorizing stories and headlines by size and partisan leanings. His results suggested that when taken individually, the elite press varied in their levels of coverage, but when
* Whether or not voters separated Eisenhower’s campaign coverage into pure campaign versus administrative act coverage complicates the issue. Often the press reports incumbent presidential activities but frames such events with the upcoming or ongoing campaign in mind.
^ In a survey conducted by the magazine Practical English, and reported on page 12 in the April 2,1960 edition of Editor & Publisher. 335 editors of the nation’s newspapers in cities with populations o f200,000 or more voted on the top dozen newspapers in the nation that in their opinion were considered superior. The list was expanded to fifteen newspapers. They were the following: New York Times. Christian Science Monitor. Milwaukee Journal. St Louis Post Dispatch. Washington Post Louisville Courier-Dispatch. Wall Street Journal. Atlanta Constitution. Chicago Tribune. Des Moines Register. Kansas Citv Star. Baltimore Sun. Miami Herald. Chicago Dailv News (no longer in existence) and the Los Angeles Times. These are the newspapers Stempel’s prestige or elite press are based upon.
10 taken as a whole, the coverage was markedly balanced. However, this idea of balance is practically irrelevant since most individuals do not read a “sample” of elite newspapers.
Danielson and Adams (1961) found that the national press is far more rigorous, or complete in their coverage of campaign events than local newspapers. They examined a nationally representative sample of 90 daily newspapers from September 1 through
November 7, 1960. Stempel (1965) using his sample of 15 elite newspapers found very similar coverage distributions in his analysis of coverage during the 1964 general election campaign. Hart (1965) examined four elite U.S. newspapers. He found that all four tended to give slightly more coverage to the Republican party. The New York Times, one of the papers in his analysis, stood out as the most equitable in their coverage of the Republican and Democratic candidates, Nixon and Kennedy, respectively.
Doris Graber (1971) extended the research by drawing a relatively large sample of twenty daily newspapers from a national pool. She coded campaign news stories and the opinion/editorial sections for four weeks before the end of the 1968 general election campaign. She categorized the coverage by the amount of candidate coverage, the percentage of coverage representing candidate quotes, and the number of times presidential “characteristics” and policy issues were mentioned. She found that during this election period, the newspapers in her sample favored the Republican candidate, Nixon, when they discussed the candidates. Yet they treated the candidates with more equality when they used their quotations. Finally, the newspapers focused most of the coverage on personalities and candidate style, although in the 1968 campaign, domestic issues were featured prominently as well. At the time questions still abounded on civil rights and the
11 Vietnam War. There was enough social discontent present for a third party challenge to arise, which may have contributed to the prominence of issue coverage.
Stempel’s analysis of the elite press during the general election of 1968 found that overall the major candidates again were treated evenly in terras of how much coverage they received contradicting Graber’s finding of more coverage for Nixon. The third party candidate, George Wallace, received 22 percent of the coverage on average. However,
Stempel noted that individual newspapers did vary somewhat in the distribution of coverage they allotted to the candidates. He suggested that editorial bias might affect the individual distribution of coverage.
1.1.2 The Emergence of Television: Studies of the Mass Media Campaign
Moving into the post-1968 campaign era, scholars began to focus their attention on the nature of television and campaigns. Evarts and Stempel (1974), for example, examined three television networks and compared them with coverage in three news magazines and the fifteen elite newspapers analyzed in an earlier study. They sampled days falling between early September and early November of 1972. They found that the CBS
{Columbia Broadcasting System) Evening News provided the most balanced (evenly distributed) coverage of McGovern and Nixon. Other outlets were not so evenhanded.
Hofstetter’s 1976 study of the Nixon versus McGovern races finds roughly equal treatment by the three major networks {The American Broadcasting Company, The
National Broadcasting Company, and CBS). He also found that the networks, at least in
1972, focused on issues in their extended coverage of the campaign. Specifically
12 Hofstetter found that when the news program did more than a bumper (a very short segment that provides a bridge from one report to another) on the campaign, the content often focused on policy issues. Graber complements Hofstetter’s study with an examination of coverage from a national sample of newspapers during the presidential campaign of 1972. In this study she focused on the role that Nixon’s incumbency might play on his coverage and the overall nature of the campaign’s coverage. She found that the press emphasis tended less to candidate style and personality than it had during the
1968 campaign, but focused more on experience, ability and candidate ideology. This may be due to the candidate field including an incumbent. Meadows (1973), also studying the election of 1972, found evidence in his analysis (of three network news programs and three national newspapers) that while there was equality of campaign-oriented coverage,
Nixon through his incumbency, could generate more coverage in various policy areas. In their examination of local newspaper coverage of congressional races Clarke and Evans
(1983) found that incumbents generated more issue and ideology coverage than did challengers.
In their 1979 study, Malaney and Buss try to explain the variation between print and broadcast news media coverage of campaigns during the 1972 presidential election.
While most previous studies had pointed out difierences between the two mass communication outlets, few attempted to explain these differences. Malaney and Buss’s results suggest that the inherent structural differences in the network news and newspaper organizations are responsible for what they focus on and how they report information generally, and political information specifically. Malaney and Buss posit that the print
13 media will emphasize issues more than the networks because they have the means to deliver such in-depth coverage. Television coverage demands more drama and less time.
Network news is a rundown of the day’s events. Newspaper coverage is an exploration of the day’s events and a continued evaluation of prior events.^
1.1.3 Expanding the Focus: Media Studies in the 1980s and 1990s
Much of the preceding news media scholarship was generated in response to the belief that the mass media did have an influence on voters and on candidates although empirical research had a diflBcult time demonstrating anything beyond limited efiects.
Thus, when scholars attempted to glean out whether bias toward particular candidates existed, they did so because if the media is somehow influential, any bias could erode our democratic processes. Yet again, one of the problems inherent in these generally useful studies was that the general election as the study context does not always provide a “fair” testing ground.
Many voters may not be influenced by media coverage during the presidential general election because a majority of them have already made up their minds. Flanigan and Zingale (1994) report that in 1992, 39 percent of the electorate had determined their vote choice before the conventions; 14 percent had done so during the conventions; and
45 percent did so during the general election campaign. The presence of a third party
should be noted, however, that they compared the Associated Press (A?) wire-service with the CBS Evening News. A more realistic comparison would have been to compare a daily newspaper with the network evening news. The wire service provides an avalanche of news every day, only some of which makes it to the printed page or newscast.
14 candidate can increase the amount of voter uncertainty. In 1984 and 1988 the percentage
of voters making up their minds before the general election campaign was 65 percent and
60 percent respectively, compared with 1992's 53 percent. Information, such as the news
media transmit, may have more of an effect during periods of uncertainty, i.e., less
information, more choices. Therefore the level and stage of the electoral event may
condition how influential the media may be. This may not occur, for example, only from
the presidential nominating stage to the general election stage, but within the nominating
phase. The ability of information to influence trends in candidate preferences should
decline as the choice becomes more limited (as other candidates are winnowed out) and as
the body of general information about the candidates increases. As Flanigan and Zingale
(1994) point out, “Since the impact of the media depends in part on the amount of prior
information the audience has about a subject, we would expect there to be considerable
variation in its impact depending on the state of the election campaign and the level of
public knowledge about the candidate.” (P. 155)
Patterson and McClure (1976) made the claim in their analysis of the 1972
presidential general election that television campaign coverage had no influence at all on
the audiences’ views of the candidates or on their understanding of the election issues.
They condemned television by suggesting that by presenting the campaign as
entertainment the media succeed in trivializing the event. Their results suggested that a candidate’s television commercials furnished voters with more serious information on the
15 issues than all the regular network news programs combined. This view has come under
repeated criticism by researchers in journalism and mass communication studies.*
Yet in Patterson’s (1980) later study of the mass media in which he focused on the
1976 presidential nomination campaign, the emphasis was on the informational role of the
media. In this study of three network news programs, two news magazines and four
newspapers analyzed from January 1 to November 2, 1976, he found that issue awareness
of individuals did increase during the nomination campaign, but that the amount of
learning varied considerably by candidate (pg. 155) and that much of the focus of the
coverage was on the horse race.
The fact that voters’ recognition of candidates and their issue positions can vary
has been observed repeatedly, but whether this variation depends on the candidate’s
behavior as opposed to the media’s, or an interaction of both, has been explored little. For example, much of the literature has been critical of the media’s attention to the game and the lack of issue coverage. Yet candidates, when they do focus on issues, do so in a difruse way believing that this gives them flexibility when addressing different parts of their partisan coalitions, preferring not to offend any potential supporters (Graber 1993).
Patterson (1980) and Keeter and Zukin (1983) found little substantive coverage in the nomination campaign. They suggest that substance may be more visible in the general election when partisan differences are emphasized by the press. Moreover, the media may
*Part of the criticism stems from use of exposure measures without controlling for attention. Also, the analysis is based on “raw correlations that represent individual differences, not controlled tests of a causal model” (Zhao and Bleske 1995). Finally, other studies using alternative measures and methods have failed to produce consistent results (Hofstetter, Zukin and Buss 1978).
16 be better able to report on substantive differences between the candidates when there are
only two sets of differences to dredge out/ Yet the presence or absence of significant
amounts of policy/issue coverage may be conditional.
Moreover, candidates themselves are guilty of “pitching” or “spinning” the game.
In their quest for competitive advantage, candidates set expectations for themselves and
their opponents. They evaluate the impact of the loss on their campaign. They define the
issue debate in simple terms — often for strategic purposes (Keeter and Zukin 1983) —
and focus on their polls.
Robinson and Sheehan (1983) analyzed the coverage of CBS television news and
the United Press International wire service during the 1980 campaign. One of the main
purposes of the study was to investigate the question of “quality” of media coverage by
setting up five criteria for evaluation: objectivity, fairness, access, seriousness and
comprehensiveness. Their work provides a basis for the evaluation of coverage quality.
As to their findings, they counter the postulate of liberal media bias by revealing that bias
is not particular to party, but rather to incumbents and fi’ontrunners. These candidates
receive more coverage than their opponents, and sometimes it is more critical.
^ In their study of voter learning in the 1992 general election campaign. Weaver and Drew ( 1995) found that increased interest in the campaign (which also correlates highly with newspaper usage) and more exposure to local and national television news was significantly associated with more knowledge of the issue stands of George Bush and Bill Clinton. There results counter assertions made during the 1992 campaign that the “new news” such as Larry King Live, Good Morning America, KTTV and other “nontraditional” news media were responsible for greater voter attention to issues and interest in the campaign. The 1992 coverage was considered difierent that previous years due to the major television networks commitment to responsible reporting. The Report of the Twentieth Century Fund cites five positive developments that took place in television news: 1. Less reliance on photo opportunities; 2. Longer sound bites; 3. Increased issue coverage; 4. Critiques of campaign commercials; 5. Benefits of the “new news” such as more direct communication between voters and candidates produced by the ninnerous call-in opportunities on shows such as Larry King Live.
17 Stempel and Winhauser (1984) examined “elite” press coverage again during the
1980 presidential general election. They found that in this election incumbent president
Carter received more editorial coverage than Reagan. Yet Reagan received more
directional (or evaluative) coverage. They also found that independent candidate
Anderson received far less coverage than did George Wallace in 1968. Overall, aggregate
campaign coverage (how much attention the media devoted to the campaign) declined in
1980.
Brady and Johnston (1987) examined whether the content of media coverage
during the 1984 Democratic party nomination campaign was composed of primarily
“horse race” coverage or “issue” coverage. What they found was that there is a
substantial amount of serious issue-oriented and candidate-oriented coverage during the
campaign. They found that people do learn about candidates during the primaries, but that
they probably learn much too late about some candidates. They also found evidence that
media coverage of candidates is linked to the voters’ perceptions of candidate electability
and viability. While they did not directly estimate the strength of the relationship, they did
examine trends in both variables that support their position. Bartels’ (1988) study of the
1984 presidential nomination campaign found support for media effects in candidate
awareness levels for voters who were both high consumers of news and low consumers of
news.
Patterson and Davis (1985) focused their attention on the New York Times and on the Syracuse Post-Standard. They found that from January to February 1984, the different
Democratic candidates, Mondale, Hart and Cranston were perceived differently by the
18 media examined, which in turn earned these candidates differential degrees of coverage.
In another study of the 1984 nomination campaign, specifically the Iowa and New
Hampshire contests, Buell (1987) looked at local, regional, and national newspaper coverage during the time period fi'om October 1, 1983 to March 2, 1984. Specifically he examined the Boston Globe. Concord Monitor. Manchester Union Leader. New York
Times and the Wasiiington Post. His research found that the press places a heavy emphasis on these early contests and uses a horse race/game frame most frequently in their presentation. He also found that there was coverage variation between the national, regional and local press.®
Marc Ross (1992) examined how three potential effects of television news affect the public’s perceptions of candidates. Using the National Election Study’s Continuous
Monitoring Survey conducted during the nomination campaign of 1984, he found support for three different types of effects (candidate-specific attention, horse race and tone effects) on the public’s assessment of a candidate’s attractiveness, their willingness to vote for the candidate and their judgements of the candidate’s probability of winning the nomination. He suggests that television coverage of candidates does matter but in different ways and to different degrees across candidates. Moreover, he suggests that there is evidence for the power of strong indirect media effects.
®Other studies have confirmed the press’s emphasis on the game, as well as its focus on “visible” activity and its preference for hard facts (Graber 1993; Marshall 1980). Graber ( 1993) points out in her work that the media tend to “cast” the candidates as winners, losers, frontruimers, likely nominees and has-beens. She believes that given this penchant for setting expectations and interpreting results, the news media can often set the standard for winning a contest. Even if a candidate wins, yet does not meet the standard regarding vote percentage, that candidate can be rendered “bounceless” or even “winless.”
19 Hagen (1992) focused on the variation in media coverage of the candidates paying special attention to the coverage given to frontrunners. His examination of the 1984
Democratic campaign produced evidence that a high degree of coverage difference exists between frontrunners and their challengers. However, within the frontrunner’s campaign coverage, he found a great deal of variation as well. Other studies have also found that the media confront the dilemma of covering many candidates by giving uniformly “thin” treatment to all candidates but those designated as frontrunners.^ Hagen’s major finding, however, is that frontrunners, while getting uniformly more coverage, must deal with increases in “critical” press coverage. In his study he attempts to explain what causes this increase in media scrutiny focusing generally on the media’s role and less so on the role of opponents’ whose rhetoric gets increasingly sharp as their chances at the nomination decrease. Gurian and Barthlow (1991) also found that the national media tend to favor winners with more coverage during the nomination campaign.
1.2 General Theories of Media Coverage Content
There are four theories that provide an explanation of the way the media cover political events. First, the campaign model of reporting suggests that the tempo of the campaign generated by the candidates, their staffs, and in the case of the presidential nomination, the voters in each state, determines what is covered by the media. Reporters
^In the 1996 Republican nomination campaign, the Center for Media has already released data suggesting that Dole received at least 55% of the press coverage of the Republican candidates. The remaining 45% is shared among the other seven candidates.
2 0 take their cues from actions of the campaign players and dutifully report the process and the results.
An alternative model is the incentive model. This model suggests that the media have economic/material needs and must respond to the tastes of their audience.
Therefore, those events considered interesting and exciting are covered since a general audience, like that of a news broadcast or daily newspaper, are more likely to want entertainment along with their information. The needs of the candidates (exposure) or the ideal voting public (candidate and issue information) are ignored unless, according to
Graber, “they manage to generate the kinds of stories and pictures that journalists find irresistible. This is why most campaigns now employ professionals who know how to create events that attract desirable media coverage” (pg. 262).
The Wall Street Journal, for example, is more likely to focus on the candidates’ economic and business policies in order to respond to the tastes and needs of its more specialized audience. However, most national newspapers and television networks are not as focused as the Wall Street Journal. So what is their incentive — to present exciting news to sell papers?
Under a third model, which I shall call the responsible intermediary model, it is assumed that the news media sometimes they see their role as simplifiers and interpreters protecting the American public from being manipulated by political chicanery. The news media may derive a purposive return for their political coverage. Editors and reporters may view themselves as part of the process of democratic government — as a linkage between the public and the government.
21 One other possible model, is the partisan intermediary model. Here a news outlet’s
motive for covering the campaign might be partisanship. Newspapers and other news
outlets such as the national network news organizations have been accused of liberal bias.
Conceptually, an incentive for news outlets might be to slant the news in a partisan feshion
to support their own ideological or partisan biases. The Manchester Union Leader’s editor, Nacky Loeb, claimed that the her newspaper’s coverage of Pat Buchanan and
George Bush influenced the voters of Manchester in particular, and New Hampshire, in general, to support Buchanan.
If the economic incentive model holds in presidential nomination campaigns, one might expect more horse race coverage as an incentive would be sales from “exciting” news, but in combination with the other models, the responsible intermediary and partisan intermediary model, it is not simply selling papers but serving the public or serving one’s own partisan interests. Here we would expect not only exciting coverage, but also insightful coverage, analysis and investigative reporting and in some cases slanted coverage.* Under the campaign model, we might expect straight reporting on the campaign as it progresses with less analytical scrutiny and more factual description.
What I expect to find is that to a certain extent each of these models may apply.
News media outlets do cover the process of the campaign. Some newspapers cover the process aspect more. These are the news outlets that generally rely on wire service reports
*In 1996, the Atlanta Constitution’s focus during the summer was the Olympics. Coverage of the Dole and Clinton campaign was put aside until the Fall due to the necessity of providing information on and utilizing resources for 1996 Olympic coverage. (Inteiview-A.L. May, Eiditor, Atlanta Constitution.) Thus, other events unrelated to the campaign can motivate editors to ignore the campaign for pecuniary reasons.
2 2 and have no office in Washington, D C. or New York. Their stories cover who, what,
where, how and sometimes the why. Other newspapers do both; they provide event-
oriented coverage and investigative reporting, analysis, and poll information that they have
generated. The ratio of the incentive-based coverage, responsible and partisan
intermediary types of coverage, and the campaign event/activity-based coverage is
determined by the news media outlet’s editors and stafi^ their resource base, and their
audience. Most news media believe they have a responsibility to cover events of such
political importance. It is a critical part of their credibility as news organizations. Yet those
events that are interesting, exciting or involve some type of conflict may receive more
coverage, even when the news outlet has high purposive returns. The degree of
partisanship in reporting has lessened in relation to the partisan presses of the 1800s and
the early and mid-1900s. But there are still some newspapers that have visible partisan
attachments. While these theories are difficult to test directly, examining the content of
various news media outlets’ coverage can shed some light on which theory or theories
come closest to matching the reality of campaign news coverage.
1.3 Generalizability of Media Study Findings: The Question of Mass Media Homogeneity
One reason there has been little in the way of a movement to define mass news media in more definitive terms has been the assumption of media homogeneity (Graber
1993; Marshall 1981; Patterson 1980; Leroy and Sterling 1973). This assumption is based
23 on findings that suggest that media coverage of the presidential nomination and general election phases is homogenous as well as observations regarding the mass news bureaucracy - the role of the wire services, prefab editorials, and syndication services.
While few studies have concentrated on this question exclusively, it has been discussed peripherally because researchers have been concerned with questions of data source and measurement within studies that have focused on more substantive questions. However, the evidence in the literature is sometimes contradictory - some studies have found differences among outlets, others have found little or none of great significance.
Patterson’s study of the 1976 presidential nomination campaign stands out as one of the most comprehensive studies of the nature of media coverage and its individual-level effects. He compared the network’s evening newscasts. Time and Newsweek, the Erie
News/Times, the Los Angeles Herald/Examiner and the Los Angeles Times. However,
Patterson counsels that his study does “not provide a very precise indication of what was reported about the presidential candidates” by these outlets but rather concentrates on the news themes and news makers that dominate in the coverage. (P. 12) Thus, his study’s finding of homogeneity reflects larger similarities of coverage in dominant themes and foci.
Marshall (1980) also suggests that media coverage is relatively homogenous and thus in his analysis of media verdicts, he relies solely on fi-ont page coverage of the Washington
Post and New York Times as media data sources. He suggests that relative to their space, news outlets are similar in how much coverage they give to various individuals and events.
This is explained by the “rules” the news media use to assess campaigns, particularly regarding judging candidate success and the institution of “pack journalism.” While this
24 may be true of the mass news media (defined as the national news wire services, the three television networks with news organizations, and the large metropolitan newspapers syndication services, such as the New York Times and the Washington Post) of two decades ago, it may not be as true in this new expansive communication age. Moreover, underlying the assumption of mass news homogeneity is the assumption that the news generated by national organizations is the only “news that matters” regarding national events.
Danielson and Adams (1961), Patterson and Davis (1985) and later Buell (1987) found differences in media outlet coverage of the candidates that could be explained by variation in whether the outlet was considered national, local or regional in nature.
Problematically, it is likely that at times media coverage, between national outlets and local outlets, is similar. When crises occur, scandal arises or a concrete event of importance occurs (such as the New Hampshire primary) the news media are likely to carry stories on the same topic. Whether or not the coverage is identical is another question. Moreover, since some days produce an overabundance of “news,” the media must often choose what to cover. Other days there is little news and yet the “news hole” must be filled. While the assumption of media homogeneity allows researchers to more efficiently represent national media coverage in their models, the question of appropriateness of the assumption depends on the nature of the study. If a study is attempting to predict or explain an attitude or behavior, can we assume that all national news media outlets have presented the same information, in the same manner and with the same attention level? Or can we assume that the local media have not contributed to
25 development of the attitude or behavior? Unfortunately, too little research explores these points, particularly in areas where national and local factors can contribute to attitudes and behavior, such as presidential nomination campaigns.
1.4 The Nature of Presidential Nominations
The presidential nomination campaign is a complex series of events that involve multiple candidates, a variety of issues, and a vast array of voters across the country.
Candidates are extremely active during the campaign, moving from state to state on a daily, sometimes hourly, basis. The information potentially available to the public is immense, and the range of possible interpretations of this information is also rather wide.
One of the major players in the process of determining what information gets out to the public is the news media (Entman 1989).
The news media, mainly at the national level and potentially at the state level
(although little is known regarding this relationship), are the mediators of the nomination campaign. Their role of presentation and simplification of “news” information for mass consumption is one of paramount importance. Kinder and Sears (1985) suggest that the events and individuals that succeed in dominating the headlines in the papers and over the airwaves for a substantial period of time become known to almost everyone. The media, through their ability to select the focus and content of the coverage presented to the public, may affect a number of campaign factors. However, the campaigns for the parties’ nominations have a number of interesting properties that may in turn have some influence on the focus and content the media outlets select. The structure of the process itself, the
26 performances of the candidates, the candidates' resources and the candidates’ characteristics — these factors may influence the news media’s interpretation of what should be focused upon and what should be said.
1.4.1 Structural Properties of the Nomination Campaign
There are a number of structural properties (norms, rules, timing, laws, etc.) that may aflFect the absolute amount of coverage given to an event and the candidates active in that event. Presidential nominations are determined through a parade of state primary elections and state party caucus/conventions in which the outcome of a particular contest
(particularly delegate-laden and early primaries) is likely to have a major impact on those that follow; news coverage, then, like many other factors in the nomination campaign, leans strongly toward the early and the “big payoff” contests. These are the initial indicators that set the stage for the rest of the campaign, and have often structured the competition for the subsequent contests. Candidates who do well in Iowa and New
Hampshire gamer a substantial amount of news coverage (Robinson and Sheehan 1983;
Keeter and Zukin 1983) particularly if they do better than they were expected to do.
Those covered by the press are raised in stature, while those left out of the spotlight are often condemned to the margins. In addition, contests deemed “important” or “crucial” are also given greater attention by the news media. For example, in the 1992 Democratic campaign. South Dakota, a state usually relegated to the margins, became the early
“make-or-break” contest between Bob Kerrey and Tom Harkin. Florida became the show down state for Tsongas after he declared it to be the “neutral” territory where he would
27 challenge Clinton. In 1980, John Connally pegged his hopes on South Carolina making it
a much discussed state contest.
Delegate selection rules are also expected to have some influence on the way in
which the media cover campaign events. Primaries, by virtue of the simplicity and ready availability of their results, are expected to receive more overall coverage than caucuses
(Gurian 1993). In addition, when a candidate wins a primary, this has implications for momentum and viability. Caucuses, excepting a competitive Iowa caucus, generally are not given significant attention before, during or after by most of the news media. Thus winning a caucus may be beneficial in terms of accruing delegates, but the “momentum boost” is much smaller than that which an important primary win provides (Marshall
1981). Moreover, caucus results are often tentative, open to interpretation, and sometimes unfathomable. For example, in 1992 the Virginia Republicans used caucuses to select delegates to go to the national convention. When caucus goers voted for delegates there was no presidential preference attached to the delegate names. Therefore, the news media could not discern what percentage of the caucus participants supported Bush or
Buchanan.
Finally, the number of same day contests may limit the overall coverage levels given to candidates, while actually increasing the overall coverage of the event. For example, southern Super Tuesday, as it has become known, has a large number of states conducting their primaries and caucuses on the same day. The media, with its limited time and space, must somehow cover these events. What generally occurs is that the campaign is simplified. Winners and big losers may receive the most coverage. However, many
28 candidates may be lost or partially ignored due to the need for the news media to report a
complex event with many participants and simplify the results and implications.
Structural properties of the campaign can influence the overall amount of
coverage. Early contests get more coverage. States with many delegates at stake receive
more coverage. This translates to higher coverage in absolute terms for all the active
participants. In addition, at the candidate level we see structural properties affecting the
coverage of frontrunners, strong challengers and big losers by increasing coverage their
coverage. Simultaneously, weak challengers or the so-called also-rans are ignored or not
covered by the media during important contests. Also, the level of attention given to
candidates may vary by event. For example, candidates may see a relatively large portion
of news coverage during the New Hampshire primary, whereas all candidates may receive
relatively little coverage (compared with the other news of the day rather than to each
other) during the Hawaii caucus, which is covered negligibly, if at all, by the major news
media and other state and local media outside Hawaii.
1.4.2 Candidate Performance
In presidential contests the substantive deficiencies of media coverage are most
noticeable during the primary period when a large slate of party candidates is running in
each primary. The media confront the dilemma of covering a multitude of candidates by giving uniformly “thin” treatment to all candidates but those designated as frontrunners or close-running “viable” candidates (Graber 1993; Aldrich 1980; Patterson 1980). By virtue of the amount and direction of coverage, the media may be “cuing” the public as to which
29 candidates are “serious” contenders. Thus, a candidate who does well is more likely to be
covered. Those who do not may not be covered, or perhaps may only receive some
limited coverage of their poor showings.
Nor is it as simple as winning and losing. Even candidates who do well may have
to meet a standard. Winning may not be enough. One may have to win by a certain margin
in order to be considered successful (Marshall 1981). For Harkin it was 60 percent in
Iowa. In New Hampshire, Buchanan had set 40 percent as the magical number, but his 37 percent finish was sufBcient to make George Bush a loser and create the standard by which his performance would be measured in subsequent contests. Therefore the standards or expectations to which candidates are held and their success or failure in meeting these expectations may have an effect on the candidate’s subsequent media coverage.
As I have mentioned, we know a fair amount about the coverage of candidates who do well because they are generally the ones receiving coverage. We know less about those who do not do well or who falter after some success. Moreover, we know relatively less about over time coverage variation for candidates in general. Only a few studies have tracked media coverage of a candidate from the start to finish of the nomination campaign process. Nor do we know whether short term success (a burst of momentum) or long term success (viability) has difierential relationships to media coverage. Does momentum guarantee coverage? Does viability ensure coverage? For example, what is the nature of the media coverage a candidate receives if he wins a state, but is not considered a fi-ontrunner or strong challenger? Do the media negate his result and concentrate on the
30 finishes of the frontnmner and runner up? Does this depend on the stage of the campaign?
For example, the early contests are considered more dynamic. There can be dual frontmnners, as there were in 1992 with Clinton and Tsongas occupying those roles when the campaign outcomes were fluid and rather unpredictable. Kerrey’s win in South Dakota made him a “player” as it was early yet in the campaign. Had South Dakota been positioned later in the campaign, Kerrey would probably have been out of the running even earlier. In addition, a non-fi-ontrunner win may be treated more seriously by the media early and less so later in the campaign, as when Brown won the Connecticut primary. At this point, Clinton had amassed many of the delegates needed to win the nomination by that making Brown’s win unimportant for his own success, but rather a damper on the Clinton campaign.
1.4.3 Candidate Resources
Candidates are competing among a field of opponents for votes and for the delegates that are allocated based on these votes. Much as in any political battle, candidates are required to invest their resources in order to gain whatever advantage they can. Several resources are critical to presidential nomination candidates. Money, time, personnel, organizational strength — all of these are considered a necessary (albeit often not sufiicient) part of a successful campaign.
Resources are required to maintain the campaign infrastructure and to generate voter recognition and support. Financial resources generally are considered fundamental - the foundation upon which other resources are built. The indication that a candidate has a
31 strong financial base may be viewed and interpreted by a number of critical players in the campaign especially during the early days or the “invisible” primary. Early in the process political consultants look to the potential candidates and assess their financial resource levels. This gives consultants who are determining for whom they will work some indication of the “durability,” “viability,” or “support-level” of the candidate. Even before the campaign gets under way, the news media (mainly at the national level), as well as other potential candidates and so on, look at the financial resources that have been raised in each camp.
Again, financial resources act as a “cue” or indicator for potential support. Often ignored is the fact that generally much of a candidate’s seed money comes from his home state or region. But if an individual cannot raise enough money to the meet the Federal
Election Commission’s requirements for qualification for matching fimds ($5,000 in contributions of $250 or less in twenty states), there is little potential for that candidate to gamer enough cross-state support to make it through the nomination campaign.
Additionally, the money requirement for matching eligibility is so low that if a candidate cannot raise $100,000, that candidate really is out of the running.
Lack of money might also constrain a candidate’s ability to travel, to set up state- level organizations (often Tsongas had only individuals’ homes as a headquarters), to send out direct mail, to hold rallies and town meetings, to transport media and supporters, to place ads, and so on. Thus, financial resources are used as a surrogate for candidate strength and potential success.
32 One other resource a candidate has is their own presence or that of their surrogates. A visit to a state before its primary or caucus can generate attention from the national and local press, party activists, and the competition. This is very true of the state news media. While the national press may not report on Jerry Brown’s trip to Las Vegas the week before the caucuses, generally, the local news media do cover the visit. In this instance, the very presence of the candidate makes the “event” newsworthy. According to
Arterton (1993) campaign managers are critically aware of how important a candidate’s visit to a state can be. In 1992, George Bush experienced a great deal of reproach from
New Hampshirites who were unhappy with his scant attention to their state. Brown, relatively disadvantaged financially, used his presence in both Connecticut, Maine, Nevada and Colorado, often spending little on accommodations and support staff, to win votes away from his better-financed opponents.
Candidate activity in a state, by virtue of the candidate’s or that of surrogates’ presence, is expected to be associated with candidate coverage levels in that state. Yet the strength of this association has not been explored. Brown’s example suggests that the relationship is a positive one, but there are instances where the relationship is not as fi-uitfiil for the candidate. For example Richard Gephardt spent literally himdreds of hours driving across Iowa in 1987 and 1988 visiting every location on the Iowa map. Yet the local and national press foimd little to report, and of what they did find to report, little was positive. So not only must a candidate be active and visible in a state, the press must determine that there is a response from the audience the candidate is working.
33 1.4.4 Candidate Characteristics
Candidate characteristics, principally, home state aflSliation, incumbency, name recognition, are all probably related to candidate media coverage at both the state and national levels. To a certain extent these qualities represent the preexisting levels of recognition a candidate has achieved. Home state affiliation can give a candidate an advantage in terms of the state media. For the Des Moines Register. Tom Harkin was the
“local” running for president. For the Newport News Daily Press or the Richmond Times.
Douglas Wilder was a sitting Governor trying to balance his administrative duties with a run for president. Being the home state candidate can have its disadvantages as well as advantages when it comes to the local press. As to the national news media. Harkin’s home state status in Iowa dampened the national media’s post-Iowa coverage, and thus robbed Harkin of much need exposure as a “winner.”
Incumbent status (if the candidate is a sitting president, or a vice-president, potentially) and high levels of name recognition (such as Bob Dole, Newt Gingrich or
Mario Cuomo have achieved) can give a candidate an advantage both in the national and state news media. Incumbents may be favored with greater levels of media coverage simply by virtue of the way they can generate newsworthy events from the White House.
In addition, those candidates with high levels of national recognition can probably gamer more coverage earlier. In this sense, these types of candidates are often called the
“serious” or “heavyweight” candidates. They are well known to the press and the public.
An example of such a candidate can be found in John Glenn when he competed in 1984's
Democratic nomination campaign. However, well-known candidates may be advantaged
34 early on, before many contests have taken place. If a well-known candidate does not show “winning” results, even these candidates may be left behind. Indeed, such candidates may receive far more negative coverage due to the fact that they are so nationally recognized yet unsuccessful at the polls. This is particularly true when one of the first contests is in a relatively low population state in which candidates become intimately known to the voters. John Glenn may have been a national hero, but when he sat in a
Manchester, New Hampshire diner, he was just another candidate plying his wares. Bob
Kerrey may well have been another example of this phenomenon in 1992.
Incumbents suffer somewhat the same tribulations. Being an incumbent has its advantages, yet at the same time, should demands of the White House place the incumbent candidate in an awkward or unflattering position, the coverage can become unflattering as well. Thus, while advantages may accrue to candidates having particular characteristics, the ability to turn these characteristics into favorable press coverage may be conditional.
Bush’s strategy to use his position to help amplify his campaign themes appears to have backfired repeatedly; his use of federal “pork” was viewed as politics at its worst and was reported in that maimer in some media outlets; his trip to Japan was deemed a failure although he was trying to highlight his “jobs, jobs, jobs” theme; his State of the Union address fell flat amid inflated expectations of his “cures” for the country’s economic ills.
Even Jay Leno critiqued the president stating, “Last week George Bush was photographed reading a fairy tale to some preschooler in Maiyland, and if you watched his State of the
Union Address tonight you heard him reading a fairy tale to the entire nation.”(“Tonight,”
NBC, 1/28/92)
35 Having discussed the nature of presidential nominations in the modem era by focusing on four areas, structural features, performance-oriented, resource-oriented, and candidate-oriented characteristics, I will now outline the basic plan for this study of the nature and role of the news media in presidential nomination campaigns.
1.5 The News Media and the Presidential Nomination Campaign; Plan of the Study
In this study, I attempt to marry the two branches of research on news coverage of political events. I focus on both mapping media coverage and assessing the potential influence of media coverage on presidential nomination outcomes as well as the potential effect of various political factors on generating this media coverage.
I focus on a series of questions: First, was coverage of the 1992 presidential nomination campaign of a similar nature when we compare national with local media and compare within these two categories as well? If there is variation, what does it look like?
Do national media outlets cover the candidates more and local outlets less? Are the stories that appear in the media outlets similar in content and direction of coverage? Are local media outlets less critical of candidates? Does local coverage focus on issues to a greater extent than the national media? Are there times when the local and national media coverage are more likely to be similar? Dissimilar?
While I expect there to be a substantial degree of similarity between national and local coverage of the campaign and its participants, I also expect that local coverage will vary somewhat from the national media at various times of the campaign. This variation.
36 in theory, is driven by state media focusing on state-specific issues driven of course by the information needs of the local audience and state-specific candidate activity.
In my attempt to reveal and understand the nature of news coverage of the campaign, I test several hypotheses at the candidate level. I test the hypothesis that media coverage (both national and local) varies substantially across candidates. Thus, a confirmation of this hypothesis would require the visible and significant differences both at particular time points and in overtime patterns. The cross-sectional (one time point) variation among candidates should be evident as candidates have different characteristics, resources, strategies, ideologies and so forth that lead the press to treat them differently in response to their own assumption that the voting public will treat the candidates differently.
Over time variation, on the other hand, may be evident as candidates pass through various stages of the campaign and as the field of candidates alters. Note that early in
1992 Jerry Brown received relatively little coverage compared to his peers, even after his close finish in Maine. But when the field diminished to a two-person race and Brown had shown some strength against Clinton, his media coverage levels increased. Some candidates, however, remain in similar coverage levels over their entire participation in the race. For example, in 1992, Tom Harkin's percentage of coverage (outside of his home state) rarely moved above the low levels he generated at the outset. In fact, his media coverage actually increased before his dropping out since that event had more implications for the campaign than his staying in did.
37 Campaign coverage among candidates can vary beyond simple volume, although that may be one of the most important dimensions of media coverage. The focus of the coverage content and the direction of the coverage may vary as well.
1.5.1 What Moves Media Coverage at the National and Local level?
Knowing that media coverage varies for candidates both within and among news media outlets is interesting, but important only if there is an effect associated with this variation, and fathomable only if we understand what is moving the variation in the first place. I assume that coverage can vary in two categories: the first relating to the campaigns in general and the second relating to individual candidates.
The structure of the nominating campaign itself may have a relationship to the amount of “campaign” coverage generated by the news media and be related to the amount of coverage given to candidates in absolute rather than relative terms. Specifically, campaign coverage is generally higher when there are the following present: early contests, events with large numbers of delegates at stake, and extremely competitive contests. This is true of both the national and local media outlets. However, the local media outlets are more likely to diverge fi’om national patterns of absolute campaign coverage when their event nears. One of Maine’s prominent daily newspapers, the Bangor
Daily News, gave extensive coverage to their caucuses, while the national press concentrated on candidate strategies going into South Dakota and Junior Tuesday or took a break from politics to cover other national and international news. In Maine the campaign was prominent news.
38 Another structural parameter is the size of the candidate field. If the field is large,
we should expect wider gaps in the amount of coverage given to candidates. Since news
outlets deal with a finite amount of time and space, some candidates (fi'ontrunners, big
winners or big losers) may receive more while others (the also-rans, fiinge candidates) may receive much less. As the competitive field narrows, we may begin to see more parity between candidates, but only if there is real competition. Once a candidate goes beyond a certain delegate tally, challengers or not, the news media generally move onto another story or the next stage of the process, the general election campaign.
Candidate performance is expected to be positively related to the amount and direction of the mass news media coverage. I expect that candidates who perform well will receive proportionately more coverage. The direction of this coverage can vary, however.
Initially candidates doing well can expect positive coverage. However, scholarly and journalistic observation have shown that fi'ontrunners quickly become the target of more detailed and consequently more critical reporting. Therefore, subsequent coverage of fi'ontrunners may tend to be negative, or a least of mix of positive and negative, with the greater balance being negative. As to challengers, their coverage may actually increase, particularly if the field has fallen to a three- or two-person race, and become somewhat more positive.
I propose two major dimensions of performance: momentum, a short term function of winning; and viability, a long term function of accruing delegates. Candidate momentum and viability indicators should be positively related to media coverage in general, although increasing viability (in other words — being the fi'ontrunner for an extended period of time)
39 can have a negative effect as well. The expectation is that they have a stronger association with national news coverage than with local news coverage which may be more affected by candidate characteristics.
Candidate resources can affect the national and local news media in different ways.
State specific resource allocation is expected to influence local press coverage. As to the national media, their coverage is influenced not so much by how much a candidate is spending state by state, but rather how much the candidate has in receipts. When a candidate spends a great deal of money in Wisconsin that is an indicator of the attention that he is giving the state. This may be of interest to the national media as a small part of a much larger story, but in Wisconsin, candidate attention generates local press attention.
When candidates begin to show signs of not having very much money left, the national news media is more likely to pay attention, just as they are more likely to pay attention to the candidate who has much to spend, as this is considered an indicator of ftmd raising success and thus support.
Candidate qualities and characteristics are expected to be associated with the amount and direction of the coverage both at the local and national level. For example, do incumbents receive higher levels of coverage? Is there more variability in the tone of this coverage? Do they receive more coverage of substantive issues than other candidates? Do incumbents and fi'ontrunners generate less anti-opponent coverage than other candidates?
One might expect those doing poorly to have more critical comments attributed to them
40 by the news media since they are more likely to be on the attack. Finally, home state status is expected to increase news media coverage for the candidate in his or her state as is home region status.
1.5.2 Does Media Coverage Move Outcomes?
Media coverage (both at the national and the state level) is expected to be associated with candidate vote shares. I also expect that there may be significant differences in the relationships of the two different media sources to vote outcomes. State media coverage may be more strongly related to that state’s vote outcome than national media coverage. The assumption is that the national media can only pay so much attention to the activities in a particular state, especially when so much other activity is taking place.
Therefore, the state press, more familiar with the political territory and closer to the grass roots, is more likely to be a better predictor of the eventual outcome. Nevertheless, it could be that in electing a president, voters are more concerned with national issues and take their national election cues fi-om the national media rather than their more localized media. However, it is also likely that voters in states will generally receive most of their information fi-om local news outlets, although local news outlet campaign coverage may be influenced by the national news media - both television and print.
Media coverage should also be associated with the winnowing of the candidate pool. Candidates who receive little coverage, or continued negative coverage with little positive coverage, should be expected to drop out or suspend their campaigns.
41 1.6 Conclusion
Media studies began with the examination of manipulative messages - propaganda.
Since these were intended to move the public, the question of whether and, if so, how much they affected the public was a logical pursuit. With the advent of survey research, scholars began to focus systematically on voting behavior and public opinion as the variables of interest. Initially, there was failure to show an effect other than reinforcing existing or bringing forth latent beliefs and values. Still later studies found some evidence for media effects, although the conditions under which the effects occur are still under exploration.
Concurrent to the studies of “effect” were studies of content. Knowing that the mass media could influence behavior is meaningless unless we know how - and an integral part of that question is understanding the nature of the message. Systematic studies of the nature of the mass media in the area of political coverage began in the 1940s. They began with an important political process - the presidential general election. Of course at that time the presidential nomination process was not the participatory event for most states that it has become today. These studies began with print and moved on to include television coverage as well. Often the focus was on whether or not the media are biased, whether they are fair and complete in their coverage of political electoral events. Generally these studies focused on the general election and found relative equality in the coverage of the major candidates. But later studies that moved on to explore the more complex presidential nomination campaign found differences, particularly when national, regional and local media outlets were compared. At the same time, a number of scholars (like
42 Patterson 1980 and Marshall 1981) found similarities in the way the media, both national
and non-national, covered the nomination campaign.
While these and other studies provided important revelations about media
coverage, unfortunately, few have revisited the particular question of media homogeneity
even though close to two decades have passed since their introduction.’ The exploration
of local media coverage of presidential nomination campaigns has not been revisited for a
decade. Moreover, the question of differentiated impact on vote outcomes by national and
local news media outlets has not been addressed in the area of presidential nomination
campaigns.
The nature of this study is to some degree descriptive and exploratory, but its
contributions are important steps in advancing our understanding of media coverage in the
presidential nomination phase of presidential politics. First I set out to establish the degree
to which media coverage varies over candidates and over time. This requires the collection
of information on media coverage itself and the exploration of media coverage in terms of
its univariate qualities. Then the various hypotheses relating media coverage to various
properties of the campaign process and the candidates must be tested. These will help
provide some foundation for candidate-level/campaign level explanations of what moves
media coverage and the influence of national and local news media coverage on contest
outcomes. Finally, I will use the analyses to focus on the greater theoretical questions of
what model media coverage of the campaign follows and whether the process of
’Although I plan to examine the homogeneity of the news media in Chapter 5, until that point I will assume that it is homogenous for purposes of simplicity.
43 nomination is hindered or helped by the involvement of our national and state media. The next chapter outlines in more detail the necessary steps required to carry out this study.
44 CHAPTER 2
DATA AND DESIGN
2.1 Content Analysis as a Research Tool
Content analysis is the primary research method to be used in this study.
According to Bernard Berelson (1952), content analysis is the reading, summarizing and interpreting of a body of communication content. At the time of Berelson’s writing, much of the use of content analysis had been focused on; studying the content of American newspapers in the 1920s, analyzing the political content of radio in the 1930s, examining political propaganda generated by the war effort and examining the degree of partisan coverage of presidential campaigns in newspapers, magazines and on the radio in the
1940s.
Richard Hofstetter (1981) writes of the contemporary use of content analysis suggesting that it remains a useful tool. Regarding the mass media, he suggests that content analysis can be used to make inferences about the media that include the type of campaign coverage a media outlet focuses on: personalities, issues, localized topics, horse race, entertainment and so on. He also states that content analysis can be used to measure
45 the accuracy in the reporting of political news information and assessing levels of biased coverage of political campaigns.
Content analysis, however, has sometimes been considered a problematic research method. The method invites criticism concerning reliability and validity. Repeated tests, however, of the method have brought some balm to such worries. Guido Stempel (1955), in Increasing Reliability in Content Analysis, allays fears regarding reliability or as he put it, the “consistency of measurement.” Markham and Stempel (1957), Analysis o f
Techniques in Measuring Press Performance, compared three different measurement techniques used in content analysis to learn which was the most useful, valid and reliable.
Their study revealed little difference in measuring media coverage by using either column inches, headline size and direction, or the frequency of the mention of issues. Multiple measures produced valid representations of media coverage. Media coverage can be measured in a way that accurately reflects the concept and this measurement can be replicated.
To increase reliability, Stempel proposes that extra care be taken in the clarification, simplification and completeness of coding categories. As an example he cites the use of the “neutral” category as a means to negate problems that arise due to the forced inclusion of coverage in negative or positive categories.
In his book. Content Analysis: An Introduction to its Methodology. Krippendorf
(1980) suggests that researchers can avoid problems by specifying all limitations, inclusions, definitions and other analytical steps so that the study can be reproduced without the threat of vagueness.
46 The design this study will follow consists of a three-part examination of media.
First, using the media coverage data generated by a content analysis, the question of national and local media coverage homogeneity is addressed. Second, given the anticipated variation between national and local coverage, the factors that drive this variation, both over time and at a particular time point before a primary or caucus, are explored. Third, I will determine what, if any, impact the national and local media’s coverage has on candidates’ vote shares. Each of these questions uses data garnered using the same method; however, the analytical design for investigating these questions varies. Since the foundation of the study is the data generated from the content analysis, the process of collecting and coding the data will be discussed at length.
2.2 News Media Sources to be Analyzed
This study required the collection of media coverage data from two source groups.
First, data were collected daily from both national and local newspapers over both pre primary and formal nomination campaign periods. A daily record of media coverage from
December 1, 1991 to June 3, 1992 (approximately 25 weeks or 175 days of coverage) provides information on media coverage before the official start of the nomination campaign, during the height of nomination campaign activity, and during the less-active tail-end of the campaign. While random sampling might be more efficient, it opens the process to questions of “cyclical variation” (Riffe, Aust and Lacy 1993) associated with
47 newspaper sampling.* Daily assessments allow me to track the rise and
fall of attention paid to the campaign and its participants. More important, it allows for the
comparison of media coverage as to volume, direction and substance at many points in
time. Having this information not only provides insights into the nature of news coverage,
but also provides insights into the process of the nomination campaign itself.
To generate the data, sources of the data had to be selected. First “national news
media” were defined as mass media news outlets geared to serving a national audience as
evidenced by a large national readership or viewership. Moreover, the national outlet
should deal primarily with national political issues and act as a news leader. Thus, I
examine the largest of the national media with the greatest reach or influence nationwide,
excluding, however, national magazines due to their weekly format."
Of the national news network organizations, the ABC World News Tonight was selected because it had the highest viewership (2.1 million) of all the national network evening news programs in 1992. Given the importance of television, the study would not
* According to RifFe et al. (1995) “Regularities associated with government and policy activities suggest possible sources of cyclic variation.” They cite, for example, that Monday’s news follows two days of government and business inactivity. In some months government institutions may be comparatively more inactive than in other months and seasonal differences may affect content While their example is more relevant to pohcy coverage, political campaign coverage also falls prey to such cyclical variation. In the nomination campaign, sampling during March only (when Junior and Super Tuesday took place) would have made it seem that a great deal of media coverage was prominent, but when we compare it to January or April, we would see differences. Thus, nomination campaign coverage falls prey to cyclical variation as well.
^This weekly format does not allow the examination of what occurs the day after an event Weekly magazines such as Time. Newsweek or U.S. News and World Report are often less timely than daily newspapers or news broadcasts due to their production constraints. Nontraditional media such as morning talk shows or political radio shows were not included in the study based on reported low usage. Weaver and Drew’s ( 1995) study found that while those they surveyed in the Indiana Poll were likely to read a daily newspaper 4.4 days of the week and watch television news 5.1 days a week, the figures for nontraditional media were much lower. Respondents watched morning news on the networks an average of 1.6 days a week and tuned in to talk shows an average of 1.4 days per week. In addition, more than half (51 percent) said thqf paid little or no attention to the candidate ads on television.
48 be complete without including some measure of television news coverage. Optimally,
inclusion of each of the three major broadcast news networks and the Cable News
Network would have been desirable, however the unavailability or cost of transcripts from
NBC and CBS precluded their inclusion. The Cable News Network’s news format created
problems for data collection and coding as well as comparability. Moreover, transcripts
from their newscasts were not available for 1992.
Including AgC s World News Tonight, allows for comparisons among network
news and national and local newspaper coverage. In addition, there is no doubt that the
nationally broadcast news is by far the most focused on broad nationally relevant
information.^ Moreover, of all the national news outlets, a network’s evening news show
reaches the largest portion of the mass audience. As to the national daily newspapers, the
Washington Post. New York Times (National Edition/, and Wall Street Journal, each has
a significant national circulation with the Wall Street Journal having the highest.
According to Bagdikian (1971) The Wall Street Journal and the New York Times come the closest to being national newspapers. The Wall Street Journal is published
simultaneously in many different regional locations and is readily available in more cities than any other daily. The Washington Post, while not having as large a national circulation as the New York Times and the Wall Street Journal, is considered the political newspaper
^Ansolabehere, Behr and Iyengar (1993) state that the network news is nearly uniform across the three networks.
‘*The New York Times national edition is slightly different than the New York Times state edition. The state edition contains Metro, Real Estate, and Classified sections which the national edition does not Thus, most of the “local” news does not make it to the national edition, according to Elaine Daley of the New York Times Circulation department (Interview, August 1996)
49 of record according to journalism and communication scholars. Don Graham, the
publisher of the Washington Post refers to it as a “local” paper although analysis of its front page stories over the six month period examined foimd relatively few “local” stories.
There were several District stories involving municipal funding and the Los Angeles riots effect on the area. There was also a story about Virginia’s gim control law and a story about Maryland’s governor. Nevertheless, it would be remiss to suggest that the
Washington Post does not have a somewhat dual role as a local and national newspaper.
While there are other daily newspapers with high circulations such as the Los
Angeles Times and the Boston Globe, these can be thought of more as regional newspapers rather than national or international newspapers.* The national data sources included in the study include most of the large news organizations present in that media universe.® Moreover, they allow for the representation of several dimensions of national news media; the broader based network news (ABC), the government and political news
(Washington Post! economic news (Wall Street JournalV and the dominant national print news (New York TimesV Whether analyzed individually or combined, they afford a relatively comprehensive indicator of national news media coverage of the campaign and its participants.
*USA Today is one daily newspaper that could be included as a national newspaper. However, due to its recent creation, it is excluded from the study. News outlets were chosen so that coverage data from the 1992 campaign could be compared with coverage generated by these same outlets in prior election years as far back as 1972.
®The Christian Science Monitor of Boston is distributed nationally but its coimtrywide circulation is relatively small.
50 For the local coverage data, two purposive samples were created - one, to be called the core sample and the other the augmented sample/ The core sample newspapers were selected based on geographic location, temporal placement in the sequence of nomination events (early, middle or late In the process), traditional importance of the state contest, circulation and reach within the area, method of delegate selection used by the state in which the newspaper resided, and availability. The papers comprising the sample are; the Des Moines Register (Iowa), the Manchester Union Leader (New Hampshire), the Bangor Daily News (Maine), the Atlanta Constitution (Georgia), the Salt Lake
Tribune (Utah), the Miami Herald (Florida), the Chicago Tribune (Illinois), the Milwaukee
Sentinel (Wisconsin) and the Newport News Daily Press (Virginia).*
An additional sample of local newspapers used to supplement the data include the following: The Phoenix Republic (Arizona), the Denver Post (Colorado), the Idaho
Statesman (Idaho), the Times-Picavune (Louisiana), the Baltimore Sun (Maryland), the
Boston Globe (Massachusetts), the Clarion-Ledger (Mississippi), the Las Vegas Review-
Joumal (Nevada), the News and Observer (North Carolina), the Oregonian (Oregon), the
Argus Leader (South Dakota), The Knoxville News-Sentinel (Tennessee). The augmented sample includes data derived from the core sample of 9 states in addition to data from 12 additional local daily newspapers coded for a shorter time period for a total of 21 dailies.
’This study does not investigate local television news coverage of the campaign. For an analysis of local tele\Tsion news coverage of the campaign see Just, Crigler, Alger, Cook, Kean, and West (1996). They investigate the 1992 campaign from February until the general election in November and find, among other things, that local television news has significantly less campaign coverage than the network news organizations and local newspapers.
*The Des Moines Register (Iowa), for example, fits the criteria of an early contest in a caucus state while the Milwaukee Sentinel (Wisconsin) fits the criteria of a late contest in a primary state.
51 Thus, our pool of state media outlets increases to 21. Almost half of all states are
represented. If we consider the nomination campaign to have ended by mid-April, the
sample includes a large portion of the states holding contests during the competitive
portion of the Democratic campaign. For each of the additional outlets, a two week (14
days) period of coverage before that state’s primary or caucus (including the day of the
event) is examined and coded. These data will be referred to as pre-event coverage.
The additional newspapers were selected based on purposive and practical
reasoning. Primarily, early contests were given priority to generate as many cases that
included the greatest number of active candidates. For example. Bob Kerrey participated
in only ten primaries and caucuses before dropping out. Random sampling of all states
would not have generated as many early contest states. However, purposive selection
allows for the examination of Kerrey’s national and local coverage before nine of those ten
contests. Finally, availability was a concern. Some local newspapers were not available on
microfilm (thus past editions are difiBcult to obtain), had missing editions, or they were on
microfilm but could not be obtained fi"om the library holding them. Some had short periods
of release that excluded them fi-om analysis due to time constraints. In most cases, the
newspaper of choice was available.
Tables 2.1 through 2.3 provide data on the reach of these news media outlets.
They provide a perspective on how dependent voters might be on these sources of
information which is important in assaying the overall importance of the news media.’
’while most of the outlets selected reach a substantial public, a few — particularly in states with multiple metropolitan areas with distinct media markets or strong regional markets — only capture modest portion.
52 Table 2.1: Circulation and Readership of National Newspapers
Newspaper/State Total Estimated Out of State Estimated Out of State Circulation Readership Circulation Readership
New Yoric Times 1,039,304 2,806,121 520,019 1,404,051
Washington Post 788,752 2,129,630 27,179 73,383
Wall Street Journal 1,994,415 5,384,920 1,816,903 No data
Data Sources: Editor & Publisher Market Guide, 1992; Newspaper Circulation Analysis (NCA) 1990/91- Newspaper Rates and Data Part R, Standard Rate and Data Service. Estimated readership is determined by using a 2.7 family multiplier.
Table 2.2: Circulation and Readership of Core Newspapers
Newspaper/State Total Estimated No. State Households Circulation Readership/% and Proportion State In State/AM Registered Voters Market (estimated)
Des Moines Register/IA 212,748 574,420 37% 1,291,330 17.0%
Manchester Union Leader/NH 70,307 189,829 33% 469,733 15.0%
Bangor Daily News/ME 80,337 216,910 25% 480,081 17.0%
Atlanta Journal Constitution/GA 273,883 739,481 27% 2,444,925 11.0%
Salt Lake Tribune/UT 169,035 456,395 59% 600,276 28.0%
Miami Herald/FL 415,500 1,121,850 20% 5,575,611 8.0%
Chicago Tribune/EL 765,702 2,067,395 35% 4,321,222 18.0%
Milwaukee Sentinel/WI 285,888 771,898 NA 1,898,297 15.0%
Daily Press/VA 107,032 288,986 11% 2,434,742 8.0%
Data Sources: Editor & Publisher Market Guide, 1992; Newspaper Circulation Analysis (NCA) 1990/91- Newspaper Rates and Data Part EL, Standard Rate and Data Service; Presidential Primaries and Caucuses 1992: A Handbook of Election Statistics (McGillivray 1992)
Note: Total circulation is for daily morning editions for each newspaper. Out-of-state and less than 5 percent penetration rate county circulations are not included in these numbers. Estimated readership is determined by using a 2.7 family multiplier. Percentage registered voters is simply an estimate of what portion of registered voters might be reached by the newspaper.
53 Table 2.3: Circulation and Readership of Additional Newspapers
Newspaper/State Total Estimated No. State Households Circulation Readership/% and Proportion State Market In State/AM Registered (estimated) Voters
Phoenix Republic/AZ 332,367 897,391 47% 1,306,950 25.4%
Denver Posi/CO 225,692 609,368 34% 1,371,357 16.5%
Idaho Statesman/ID 55,538 149,952 30% 347,750 16.0%
Times-Picayune/LA 283,294 764,894 34% 1,570,747 18.0%
Baltimore Sun/MD 230,417 622,126 30% 1,721,120 13.5%
Boston Globe/MA 446,578 1,205,760 2,227,740 20.0% 38%
Clarion-Ledger/MS 112,300 302,400 20% 1,511,760 8.0%
Las Vegas Review- 121,849 328,992 75% 426,560 29.0% Joumal/NV
News and Observer/NC 200,907 542,449 15% 3,467,820 6.0%
Oregonian/OR 295,446 797,704 51% 1,204,800 25.0%
Argus Leader/SD 41,418 111,829 28% 347,980 12.0%
Knoxville News-Sentinel/TN 104,545 282,272 11% 1,670,080 6.0%
Data Sources: Editor & Publisher Market Guide, 1992; Newspaper Circulation Analysis (NCA) 1990/91- Newspaper Rates and Data Part IL Standard Rate and Data Service. Presidential Primaries and Caucuses 1992: A Handbook o fElection Statistics (McGiWivmy 1992)
Note: Total circulation is tor daily morning editions for each newspaper. Out-of-state and less than S percent penetration rate county circulations are not included in these numbers. Estimated readership is determined by using a 2.7 family multiplier. Percentage registered voters is simply an estimate of what portion of registered voters might be reached by the newspaper.
54 2^ Units of Analysis
While Stempel and Markham have suggested that several media measures can capture the concept of media coverage without great differences appearing among them, there is still an argument to be made about the quality of information extracted being dependent on the unit of analysis. While there may not be a “best” unit of analysis, as selection is dependent on the purpose of the analysis, when the analysis requires distinct measurement that can be aggregated, decomposed or divided easily, meaningful groups of words are clearly preferable. For example, a story headline, while representing the major theme or direction of the story, may capture only a fraction of the substance of the story.
The headline may mention only one or two candidates, whereas more are discussed in the body of the story. Using headlines alone is a less accurate representation of story content.
While headlines are often the most read portions of a newspaper, readers often read the rest of the story, particularly if it is of interest to them (Graber 1988).
Using paragraphs or column inches or even “mentions” may also lead to loss of information. For example, if the data gatherer uses paragraphs as the unit of analysis there is room for error if the goal is a precise representation of the volume, tone and content of news media coverage. Often there are multiple candidates or issues discussed in one paragraph. How does one divide a paragraph if three candidates are mentioned in it, but one merely fleetingly?
By focusing on groups of words (comprising meaningful phrases or sentences) we can portray the coverage a candidate or party receives with greater accuracy. Moreover, by using groups of words we can decompose a candidate’s aggregate coverage for a time
55 period into other variables, such as direction and content. This study uses “groups of
words,” as the unit of analysis, for rarely can any representation of a candidate (or his or
her actions) be captured in one word. Groups of words allow for easier parceling out of
coverage. For example, a sentence can easily be divided in half if words are used as the
unit of analysis. A sentence of fifteen words can be split into 7.5 words each whereas a
paragraph cannot. Half a paragraph could mean a great deal of coverage or little at all
depending on the size of the paragraph.
Again, this does not mean to imply that other measures of media coverage are
inappropriate. For example, capturing overall campaign coverage (inclusive of both
parties) or overall political coverage can be done efiRciently by counting stories or column
inches. If we are simply interested in measuring how much attention the media give an
event, using the simpler, more efficient measure is appropriate. It simply depends on how media coverage is defined conceptually and what the analysis is designed to determine.
In this study, news media coverage of the campaign and of the candidates means
news coverage that is the most readily available to information consumers. The conceptual
starting point is news media that is most prominent, therefore most likely to be absorbed by consumers. Consequently, the data collected fi"om the news outlets examined comes only from their fi"ont pages, although for the ABC Evening News, the entire 22 minute broadcast is coded since in many respects the news presented is similar to front page coverage in newspapers. Both concentrate on “breaking” stories of importance dealing with national and international topics and human interest stories although the local newspapers include state-related news. Short segments (“bumpers”) on the evening news
56 are similar to the “hooks” (short rundowns of inside stories) that appear on the newspapers front page.
2.4 Coding Procedures
For the print outlets, each day’s front page was analyzed and the following data recorded in response to these directions/questions:
1. Record how many campaign related stories appeared on the front page that day.
The term “campaign-related” means having any reference to the Democratic or Republican nomination campaign whether specific or general.
2. Enter the following information on each story;
a. Type of story - a news report, news analysis, editorial, abstract/caption.
b. The source of a story - stafi^ news service (note which one). If an author is given, record the name.
c. For state newspapers, was the story “local” in orientation? This requires that the activity or primary focus of the story (be it an issue, candidate or a general campaign oriented reference) directly relate or take place in the state community. One should recognize that some wire service stories could be local even if not written by a newspaper’s own staff member. Associated Press reporters may operate out of the same area. The dateline location (if available) will indicate the locality.
d. Did a photograph or photographs accompany the story? If so, who was pictured?
e. What was the full title of the story?
57 f. Did the story appear in the top half or bottom half (above/below the fold) of the newspaper?
3. Read the content of the story before proceeding to allocate coverage to candidates. This allows the coder some familiarity with the story before attempting to count words and allocate them accordingly.
4. Read each sentence. Determine if the sentence is candidate-specific or general.
If general, place number of words in the general campaign category. If candidate-specific, determine the following;
a. What category or mix of categories does the content fall under? The content categories are organized into five areas: policy/issues, candidate substance, campaign, horse race, and anti-opponent comment. The various subcategories falling under these five general ones are displayed in Table 2.4.
An actual example from the Wall Street Journal allows for demonstration: “Bush officials invited undecided Republicans to a town meeting with the president in Exeter,
New Hampshire. They weren’t happy when one who showed up was the town chairman of the Bush-Quayle Committee.”( 1/17/92)
This paragraph would be recorded as 16 for the campaign general activity category
(representing the activity of holding a meeting) and -16 for the support category. The support category includes coverage regarding public support of the candidate, poll numbers, endorsements, and individual support like that of a town chair. The words are given a negative sign due to the obvious indication that what occurred was bad news for the Bush campaign - “They weren’t happy” applies to the action discussed in the sentence.
58 If this were ail the coverage given to Bush that day one would tally up a score of 32 words
for Bush - 16 neutral, 16 negative and zero positive. Often sentences will not be as simple
as these. That is one reason that sentences are not used as the unit of analysis. Moreover,
particularly with newspaper coverage, sentence length varies tremendously - fi'om four to
40 words sometimes.
A more complex example fi’om ABC World News Tonight illustrates this point.
Brit Hume states: At a minimum, Buchanan hasforced the president to campaign here
(New Hampshire) more and move to the right more than Mr. Bush may have intetuied.
(2/12/92) The 23 words were categorized in the following way. This sentence involves interaction. Buchanan is being assessed for having used a strategy that forced Bush to do something he did not want to do. The negative efiect is implied rather than direct. Hume does not say that this has hurt Bush, but rather that he was ""forced" ... to do something he did not want to do. Thus, Buchanan receives 6+ for "At a minimum, Buchanan has fo rced...” and Bush receives a 16- for "the president to campaign here more and move to the right than Mr. Bush may have intended."
Sometimes the coverage is totally neutral. For example, “Bill Clinton was
Arkansas’ youngest Governor.” Or “Paul Tsongas’ wife Nildd campaigned in Maryland for her husband today.” In other instances it is patently negative or positive. “Clinton swept the remaining primaries with ease.” “Brown received a loud roar of approval from the culinary union members present as he talked about the need to protect workers.”
“Bush’s audience remained unmoved and unresponsive as he tried to talk about Desert
Storm and his military success in Kuwait.” “Harkin placed a poor fifth with less than 3
59 percent of the vote.” Had the sentence read “Harkin placed fifth” it would not be considered negative, but neutral.
Positive coverage was defined as any written word or group of words that have positive denotations for the candidate. For example, “He won New Hampshire with ease, generating massive supportfi’om across the state.” This entire sentence would be coded as positive (6+ for “outcome” and 7+ for “support” categories). The phrase, “Tsongas received 49 percent of the vote in New Hampshire” would be coded as 10 neutral in the outcome category. Other examples of positive coverage are the following: “Perot's ability to command so much loyalty fi’om volunteers ...” “Brown’s come from behind success in Connecticut. . . , ” “Clinton’s dominance of Super Tuesday has made him the definitive frontrunner. . . , ” “Tsongas’ pro-business message gained him vocal support from New Hampshire’s business community ...”
Negative coverage is defined as any written word of a group of words that have negative denotations for the candidates. For example, “Clinton’s campaign was sidetracked today by the barrage of questions concerning his alleged extramarital aflfair.”
This sentence would be coded 15- and placed in the character category. Another example can be found in coverage of the Bush campaign: “...playing to lackluster crowds in a half filled hall” or the “campaign is showing alarm and nervousness.” This latter statement would have been coded 6- and entered into the general campaign category.
After the information is entered on the code sheet, anyone looking at it could surmise who was getting coverage that day; what the coverage was about; and how it reflected on the candidate - whether it was favorable, unfavorable or neutral in tone.
60 Moreover, the reader could assess where the coverage was favorable or unfavorable.
Table 2.4: Content Category Breakdowns of News Media Coverage
Policy/I»u« Candidate Campaign Horserace Anti-opponent
Health carc tsaict Personal health Strategy Competitive standing D irect candidate Economy/Jbba Experience Organization Momentum quotation criticizing Federal budget (governmental or Advertising Viability/Electability an opponent Can Social Security and career) Finances Expectations deal with any topic. Welfare policy Character/Ethics Endorsements/ Taxes Background Voter Support Environmental policy (personal) Outcomes Labor policy Ideology General campaign International relations Campaign activities Urban policy Message Vice-presidential Gun control Relationship w/ choice Energy policy political party Spin Agricultural policy Gay Rights Domestic policy in general .Abortion Immigration Trade AIDS
the placement of the political story, the author and source (newspaper staff or wire service), and the coverage the other candidates’ received. In addition, for each newspaper analyzed, seven days were selected at random to be counted word for word. These word counts were totaled and averaged over the seven days to produce an estimate of the general number of words that appeared on that newspaper’s front page. Using this information, an indicator for a candidate’s coverage compared with other candidates (i.e., how much coverage that candidate received compared with the sum of all candidate coverage called a candidate coverage share) plus other front page coverage (i.e., how much coverage the candidate received relative to front page coverage overall) can be
61 created.'” The average number of stories appearing on the front page was also determined
during this process.
A similar process was undertaken in the coding of the ABC World News Tonight.
Transcripts of the evening news show were analyzed for the same period. These transcripts included indications of the visuals although the comparability of the visual aspects of television and newspapers is somewhat questionable and is not addressed in this study. The number of news segments/stories was determined (on average about 11 per broadcast). The exact word count for each broadcast was included in the transcript.
The following variables were created based on these data:
General campaign coverage variables include the following:
1. Number of campaign stories/segments relating to the campaign relative to average number of stories/segments. This measure was aggregated to create a weekly and monthly measure that captures the concept of news media attention to the campaign in general.
2. Total campaign coverage in words compared with total front page.
Candidate-specific coverage variables include the following:
1. Absolute number of words per candidate per day and week. (Range = 0-n)
'°For example, the New York Times contained an average of 2,376 words on its front page. If Tsongas received 250 words of front page coverage, one could say he occupied 11 percent of the front page content. In order to determine a newspaper’s average front page coverage in words, I conducted a random sample stratified by day. I selected one fix>nt page firom each of the day strata (Monday thru Sunday) so that one seven day period would be selected for each paper. The fiont page from these seven days was assessed for number of total words and then the figures were average to produce an estimate of the average number of words that appeared on that newspaper’s fiont page.
62 2. Relative number of words per candidate relative to average total front page coverage in words per day and per week. (Range = 0-100)
3. Relative number of words in coverage for candidate compared with competitors per day and per week. (Range = 0-100).
4. Positive and negative coverage share relative to within party competitors. This variable is computed by determining the total within party negative and positive coverage and determining what share a particular candidate contributed. (Range = 0-100)
These variables are explored in Chapter Five for each of the national and local news media outlets. They are explored as individual news outlets and as news outlet groups, i.e., national news outlets versus local news outlets. The are further utilized as independent and dependent variables in chapters six and seven, respectively.
To answer another question — what drives the differences as well as the similarities between national and local coverage? — the data must be organized differently and expanded. The focus of the second question requires that a particular time point be examined - a cross section of media coverage before state primaries and caucuses.
Theoretically, the nomination process can be viewed as a series of steps each with an outcome point (the vote). Before event one, a set of rules exists and certain information is available to those who must act (candidates, the media and primarily, voters). Once they act, the players move onto the next point. The rules may or may not be different, but the information changes since the prior play alters the game. This is how media coverage can be understood to act in the nomination process. Information before
New Hampshire could affect the vote outcome in New Hampshire. However, after New
63 Hampshire, this information becomes a part of the new set of rules and information. For
example, the results of New Hampshire reach the next play in that the candidate who won
New Hampshire has momentum. Those winning delegates have accrued some viability
status. Thus, the information is not lost and is consequential.
Voters, however, may remember very little, perhaps the name of the winner, and maybe the name of the runner-up. If they have no recollection of winners or losers in previous contests, this information is provided to them before their own contest by the news media when they handicap the race. To estimate what drives a candidate’s coverage level, direction and focus, we must consider this process. A candidate’s coverage level the week before the event (and events -primaries and caucuses - are often a week or so apart) is the focus of inquiry. We can ask ourselves, what happened in the previous contest?
Does that information affect how much attention the national or local media pay to the candidate? — to the candidate’s positive and negative coverage levels? This prior event coverage is also particularly relevant because it is the information provided at this time that is believed to have the most influence on the voters who will participate in that state’s primary or caucus.
The national and local news coverage generated for the first part of the study is used in the second part but only partially. The second portion of the study examines the national and local coverage that comes before a state’s contest. In the third part of the study, where local versus national media influence on vote outcomes is examined, these variables in turn become the independent variables. In these two latter portions of the study, the relationship of national and local coverage to other pertinent variables is
64 examined in the framework of bivariate and multivariate statistical analysis.
Operationalizations for the control variables used are given in subsequent chapter’s text.
2.5 Conclusion
The core and augmented samples are good reflections of dominant large to mid sized state newspapers during the early and middle portions of the 1992 Republican and
Democratic nomination campaigns. The national news outlets examined represent a substantial portion of the available daily nationally-oriented sources of news coverage.
Moreover, from these data sources a consecutive day non-probability sample provides a daily record of the candidates’ local prior event coverage to be aggregated, summarized or decomposed and used as dependent and independent variables.
Rather than attempting to identify typical coverage during this period, I am interested in determining amount, direction and content of the daily prominent news coverage of the candidates presented by national and local media outlets to the public during the 1992 presidential nomination season, determining what influences the variance of this news coverage, and finally, exploring the influence this news coverage may have on candidate vote shares in primaries and caucuses. While the findings of this study may not be generalizable to other campaign years, the results may help to confirm or question the results of previous studies made in previous campaigns, provide a base to which future findings can be compared, and suggest potential directions for future research in this area.
65 CHAPTERS
AN OVERVIEW OF THE EVENTS OF THE 1992 DEMOCRATIC AND
REPUBLICAN PRESIDENTIAL NOMINATION CAMPAIGNS
3.1 Early Days
The earliest days of the 1992 presidential campaign were not flashy and fast-paced.
Rather the stage seemed set for a very unexciting drama. Widespread belief that George
Bush, reaching new heights in presidential popularity after the Gulf War, was invincible
led many top Democratic candidates to stay out of the competing pool and leave the
challenge to the “other guys.” Among the Republicans, it looked as though none of the
usual lower-level, single-issue combatants that showed up on the New Hampshire ballot
every four years would seriously challenge their incumbent.
Obviously, rational-minded Democrats such as Richard Gephardt, Jay Rockefeller,
Bill Bradley and the much speculated about Mario Cuomo, did not want to waste a great
deal of capital fighting what seemed like a doomed battle. A second-tier of Democratic
-candidates, led by the early entry of former Massachusetts senator Paul Tongas, declared
as candidates for the Democratic party’s nomination. This was the year of the outsider:
Later, Arkansas governor Bill Clinton declared as an outsider ready to bring “change” to
6 6 America; Jerry Brown declared and said it was time to wrest Washington back from the special interests and their big money; Iowa’s senator Tom Harkin, Nebraska’s senator Bob
Kerrey, and Virginia’s governor, Douglas Wilder, all jumped in the race ready to display their outsider credentials and their ability to win the votes of the Democratic faithful. At this point in the campaign, some Washington insiders had their bets on Bob Kerrey as the
Democratic nominee. Kerrey seemed an ideal candidate. Handsome, intelligent, and he had a war-hero record that stirred the American patriot’s heart.
Toward the end of 1991 as the campaign was beginning to move into second gear, economic indicators began to decline. Now the invincible incumbent was saddled with a weak national economy and a peaceful post-Cold War world. While this led to an opening for the Democratic contenders — they now had an issue with which to attack President
Bush — it also encouraged a challenge from within. On December 10, 1991, Pat
Buchanan, a Washington, D.C. columnist and television talk show host, announced he would challenge Bush for the Republican nomination. Buchanan attacked Bush with vigor, criticizing his trade policy, his policy regarding U.S. participation in United Nations activities, his New World Order, and finally he attacked President Bush for breaking his famous pledge of “Read my lips: No new taxes!” Bush had broken the pledge during the
1990 budget negotiations with congressional Democrats. He had agreed to a tax increase to reduce what he felt was a greater threat to Americans, the increasingly large national debt.
With New Hampshire only a few months away, the contests for the parties’ nominations were in high gear. One of the most important activities taking place at this
67 point was fundraising. Governor Bill Clinton led the way. By the end of 1991 he had raised more money than any of his challengers. Clinton benefited from several factors:
First, his association and chairmanship of the Democratic Leadership Council; Second, an endorsement from the Managing Partners, a quasi-organization made up of Democrats willing to contribute money to the Democratic party to help get a Democrat elected president; Third, his positioning as a moderate (New Democrat) with a strong economic message. Still, even Clinton did not have much national support or even name recognition then. Most of his campaign money came from Arkansas. Tom Harkin and Bob Kerrey followed Clinton in the amount of money raised. Both Midwestern candidates, Harkin and
Kerrey, had raised similar levels of funds by the end of 1991, somewhere between 1.5 and two million dollars. Clinton had raised more than three million by the end of this time.
(Washington Post. January 3, 1992, pg. A4)
One indicator that the media uses to gauge viability is money. Due to his first place showing in fundraising, his competitors and the media treated Clinton as the frontrunner before the formal start of the primaries and caucuses.
During this period, the candidates toured the country trying to gain recognition and raise money to qualify for federal matching funds. They gave speeches, held campaign rallies, made the tours of radio and television stations, and sat with journalists. The media at this stage is very important because during this process they establish the starting block positions for the contenders. Being in the very front or in the very back of the starting line can have its dangers, but most candidates prefer the former to the latter. Governor
Douglas Wilder got the latter. Unable to generate much support for his candidacy and
68 citing problems that required his attention in the state of Virginia, Wilder dropped out of
contention in January. Clinton, on the other hand, got the former. His campaign strategy
contributed to his placement. For example, between mid-November and the end of the
year, Clinton had raised nearly $2.5 million. His campaign organization then released the
fund-raising reports that proved Clinton’s campaign was taking oflf This move forced his
opponents to release their reports. When they made the comparisons, the media reported that Clinton was outdistancing his rivals, thus creating a strong image of a front-runner.
One other factor used quite effectively by the Clinton campaign was the Florida
Democratic straw poll. They had begun early to create an expectation that the straw poll would be a very competitive, if not tough, fight for Clinton to win. In reality, they believed
Clinton would have no trouble, and he did win. While the event did not make the front pages of every newspaper, it did generate some momentum for Clinton.
As Robert Loevy writes, “There was no question that Bill Clinton had proven to be the big attention-getter during the news media candidate evaluation and promotion period of the 1992 race for the Democratic nomination. Many observers were amazed at how little was required for the news media to declare Clinton the leading candidate. Susan
Feeny of the Washington bureau of the Dallas Morning News pointed out: ‘The press designated Bill Clinton as the pre-primary and pre-caucus front-runner based on little more than three speeches at Georgetown, a successful fund-raising effort and a victory in the
Florida straw poll.”
With the frontrunner established before Iowa and New Hampshire, the only question left was who would become the runner up and who would be left among the
69 also-rans: Tom Harkin, Bob Kerrey, Jerry Brown or Paul Tsongas? Another question was- did any of these candidates have the support to beat Bush? The public already raised some questions about the quality of the Democratic contenders and the concerns of
American voters.
3.2 Initial Contests: Setting the Campaign in Motion
In 1992 the campaigns encountered a considerable amount of front-loading. States choose the time at which their delegates to either party’s national convention will be chosen. The Democratic party, however, has established a “window” for these contests that begins on the first Tuesday in March and ends on the second Tuesday in June. The
Republican party has no such window, but tends generally to follow the Democratic calendar due to state laws that govern the primaries and caucuses. There are a few exceptions allowed by the Democrats. Iowa, New Hampshire, Maine and South Dakota have traditionally conducted their delegate selection events earlier.
Iowa, traditionally an important event since the 1976 surprise showing of Jimmy
Carter, was not important in 1992. No Republican contest took place in Iowa. Buchanan had declared too late to run in Iowa, and Bush won the state. As to the Democrats, Tom
Harkin, as the home state favorite, deterred any real competition in the state. Without competition, the media did not cover the story of Iowa’s caucus to any great extent. The results (Harkin finished with almost 80 percent of the vote, followed by uncommitted,
Tsongas and Clinton, respectively) really did not affect expectations for the upcoming
70 primary in New Hampshire. Harkin received no momentum “bounce” for the subsequent race. Clinton had been leading in the national polls, and he led in the polls in New
Hampshire. This lead was soon to falter when information relayed through the news media reached the public.
Clinton became the focus of scandal. Broadcast and print outlets disseminated stories alleging that the governor had an extramarital affair with Gennifer Flowers while he had been in oflBce. The story appearing first in The Star, a weekly tabloid, received much publicity. Unfortunately for Clinton, the story found its way to other more credible news outlets as well. Contrary to popular belief, it did not receive the largest headlines or front page prominence. The story was often buried on the inside pages if present at all. When the New York Times finally produced a story, it did make the front page, but appeared on the bottom. The story had a smallish headline and did not mention Flowers’ name, instead it referred to a “cabaret singer.” Some newspapers did not run the story. Many, however, did and the Clinton campaign was worried. Rather than letting the media and his opponents take control of the issue, Clinton and his wife, Hillary, decided to go to Sixty
Mhmtes and thus preempt fiirther inquiry. Clinton believed that by going directly to the people, he could get his own side of the story out.
Before the fire died out on marital infidelity, the Wall Street Journal printed a story charging that Clinton had avoided the draft during the Vietnam War. The result of this charge, which some believe was the more damaging, was that Clinton began to drop in the polls. This decline left openings for other candidates to benefit. Positioned as a moderate liberal on fiscal issues, and from a state that bordered New Hampshire (in fact his home in
71 Lowell, Massachusetts was only 17 miles away from the New Hampshire border),
Tsongas began to run first in the New Hampshire polls.
The campaign in New Hampshire was a frenzy of candidate activity and news media activity. Major networks set up shop in Manchester the week before the big event.
Cable News Network (CAW) in particular, set up extensive satellite operations that other reporters used as well. Use of satellites allowed live reports on the New Hampshire campaign to be disseminated across the country. According to Loevy, hometown reporters analyzed what was going on in New Hampshire in terms that would interest their particular hometown audience. This was especially true when a candidate was from that particular state or if the hometown state had an upcoming presidential primary.
Each campaign developed media strategies: Tsongas set up his rallies so that newspaper and television reporters were set up to get shots of his many supporters, and he had a visual aid - his book - A Call to Economic Arms. While few newspapers or television stories gave the voter a review or critique of the book, they showed Tsongas holding it and they mentioned the book in the text. This helped promulgate an image of a thoughtful individual with definite ideas on how to solve America’s economic problems.
Buchanan had a busload of supporters who followed him everywhere with a ready-made crowd, and they wired him for sound throughout the day so the reporters who covered him could always hear what he was saying, even if they were a block away. Clinton had videos, and lots of them, that he distributed by door-to-door drops. Brown, the candidate on a budget, spent most of his time working college campuses.
72 While the Democratic race may have been the more competitive of the two, news organizations gave the Republican race far more media attention at this time, particularly by the local media. The Manchester Union Leader railed George Bush. And Nacky Loeb, editor of the newspaper, was on first name terms with Pat Buchanan. It showed. Loeb extolled his virtues while fi'ont page editorials hammered Bush daily. Buchanan, as well, worked hard to foment negative feelings toward the president among the New
Hampshirites. He used New Hampshire’s recession to focus on Bush’s economic failures.
He rallied the New Hampshirites “to send a message” to Bush that they were unhappy that he broke his tax pledge. Front page stories and evening news lead stories that spoke of unemployment and other economic downturns complimented his attack. In the end, no one thought that Buchanan could win the nomination. The question was how much he would embarrass the incumbent president. In New Hampshire Buchanan was able to do just that. Buchanan came away with 37.4 percent of the vote to Bush’s 53 percent. This was several points shy of the 40 percent Buchanan had set as a win, but it was enough.
This result prompted a disturbed Bush to issue a memorandum from the White House saying that he understood what the people of New Hampshire were telling him.
While most of the Democrats were pleased over the Republican primary results, they had their own issues to sort out. Tsongas won. He received 33.2 percent of the vote compared with Clinton’s 24.7 percent. The rest of the pack was farther behind. Reporters were stationed at each candidate’s headquarters ready for their statements. New
Hampshire had given Tsongas a real boost. Suddenly he was a player and potentially
73 viable. The media surrounded him, and he responded that he won because his message
was taking hold.
The media reported the Clinton “spin” on his second place finish in New
Hampshire. Clinton’s line calling himself the “comeback kid” appeared in stories across the nation. Clinton’s campaign was very eflfective at making the result a “better than
expected” one. He turned his scandals into political advantage citing that even with charges of draft dodging and marital infidelity leveled against him the people of New
Hampshire gave him their support. He also called Tsongas a regional candidate who would disappear when the campaign moved to the West and South. This line too was picked up by reporters. Their analysis of the subsequent outcomes reflected their acceptance of Tsongas and other candidates as regionally based yet lacking national appeal.
The also-rans, Harkin, Kerrey and Brown, received little attention post-New
Hampshire, and what they received was generally negative. In a story appearing in the
Denver Post on February 23, 1992, Kerrey said, “You cannot understand what happens when Paul Tsongas and Bill Clinton are on the nightly news for three days in a row.” They have dominated the news after New Hampshire, he said.
After New Hampshire the two M’s o f‘92 played a large role. Not momentum as has generally been the case, but money and media. With frontloading momentum does not appear to bring the benefits it can when there is adequate time between contests to collect and use funds that pour into the campaign. After New Hampshire, Tsongas was not able to benefit from the rush of money that did come his way. His organization’s
74 unpreparedness hindered not only the use of that money but the conversion of the
donations into federal matching funds. Clinton’s organization, which had more money to
start with, was better able to generate and facilitate the use of money. Not only did
Clinton have the most money before New Hampshire, but when the contest was over, he
was still in better shape financially, and able to tap a huge credit line he had established.
Moreover, in 1992, one of the little mentioned facts was a new twist on the federal
campaign limits. While the legal spending limit set in New Hampshire was $552,400, a
new Federal Election Commission regulation allowed campaigns to count only half the
money spent in a state within 28 days of the primary toward the state limit. The other half
is considered fund raising expenses. Bush spent two million dollars in New Hampshire
($1.5 million on television) and Clinton spent $1.5 million. Tsongas spent much less since by the end of January he only had $80,000, Kerrey had $115,000, Harkin had $250,000
and Buchanan had a paltry $12,000.
In some respects, Tsongas was lucky that his regional roots helped him overcome the lack of money. At the least, the fact that Tsongas was from Massachusetts allowed for
some name recognition in surrounding states. Part of this is due to the Boston Globe that had covered his career as a Massachusetts senator and is considered the regional paper for
New England. Moreover, southern New Hampshire (where most of the population is found) gets most of their television from Boston. Nevertheless, while Tsongas benefitted from his regional ties to New England in some ways, in others, he did not. His campaign organization, trying to prove he was not a regional candidate, decided to concentrate less time in Maine, where they felt they were strong. Jerry Brown’s constant attention,
75 beginning the day after New Hampshire, ended up helping Brown gain support there, but hurt Tsongas who did not contest the state until it was too late.
Most of the attention in Maine was geared toward the Democratic caucuses rather than the Republican caucuses which were a much more involved and drawn out process.
Moreover, there was no contest in Maine. Buchanan did not challenge Bush in one of his home states (Texas being his other).
The Democratic party’s outcome in Maine was a surprise to many. The candidate historically guaranteed to win, Paul Tsongas, did win, but only by two-tenths of a percent with Jerry Brown close at his heels. Clinton finished a poor fourth in a state where he was supposed to have had a strong organization. Both Tsongas and Clinton had been spending time in the South, particularly Florida, which had already been distinguished as a showdown site. Yet a few days before Maine’s Sunday caucuses, each flew back to Maine having heard that Brown was gaining on them. Brown’s personal campaigning had worked, much as Harkin’s would in caucuses to come. Now the race was uncertain.
Tsongas’ momentum seemed fi'ail, and it was now a three-person race, soon to be even more confused by the last of the individual (as opposed to multi-event) initial contests.
South Dakota.
In 1992, two of the five Democratic candidates were from states that bordered
South Dakota (Harkin from Iowa and Kerrey from Nebraska). South Dakota had moved up its primary in 1988 to play a more important role. It had previously been a part of the
June Super Tuesday that included California.
76 Kerrey was the favorite. As an ex-govemcr and current senator of Nebraska, he
knew the issues of the farmers, the cattlemen, and the environmentalists. He had a war record that was an authentic tale of heroism. South Dakota was Kerrey’s proving ground.
If he did not win there, he could not continue.
The Democratic campaign in South Dakota was a very well publicized event, and it included a debate. Although none of the major networks carried the program, it did appear on South Dakota public television, and C-Span ran the debate live. Many newspapers around the nation focused on the debate. Had South Dakota not become the
“do or die” contest of the Kerrey and Harkin campaigns, it is doubtful that it would have gained as much prominence.
Tsongas came under attack from his rival candidates since he was technically the frontrunner. Kerrey and Harkin both bit into him for not knowing anything about agricultural issues. Tsongas tried to deflect the criticism with honesty, by owning up to his ignorance, but also pointing up his strengths. Moreover, he said with some humor, he was the only candidate that had South Dakota roots in his family, which he did. Clinton and
Brown also ganged up on Tsongas.
The informational context prior to the primary is notable. First, leading daily newspapers such as the Argus Leader, a Gannett Corporation newspaper published out of
Sioux Falls, South Dakota, had run a story entitled “Tsongas surprised by front-runner role” and one entitled, “Clinton puts career first, war second. Admits he dodged Vietnam, but won’t let it stop campaign.” During the two weeks before the primary, Harkin and
Kerrey were featured in stories that criticized Clinton’s patriotism and his electability. The
77 Argus Leader also covered all the candidate activities including printing their schedules.
When the contest is important, the local press responds. When the contest is non
competitive, such as the Republican caucus in Idaho, held on May 26, there is not one
mention of the campaign. As to policy, the Argus Leader covered health issues and the
environment. Kerrey’s statement regarding a nuclear dump in Nebraska made the front
page. Their editorial pages gave a thumbs down to Clinton and a thumbs up for Buchanan
for his contribution of adding excitement to the Republican race as opposed to his policies
and so on. As the event got closer, the front page coverage increased with Kerrey
receiving a substantial portion, and Harkin following close behind. One day before the
South Dakota primaries, the paper ran stories on Brown’s success in Maine. The paper
endorsed Kerrey as its pick.
Their coverage the day before the primary was extraordinary. The list of stories
included the following: No more Mr. Nice Guys: Candidates trade jabs at South Dakota
forum (Cecil-Staff), With no loser in sight, Kerrey camp banks on poll advantage (Cecil
and Walker - Staff); Kerrey curtly quips at Pressler (Staff); Brown fired up by Maine
Caucuses (Staff); Brown Ties Tsongas in Maine (Staff); Candidates inject campaign themes into forum (Staff); South Dakota Showdown - A Primary Primer: Where to vote;
forum discussion. (Staff); Voters angry and involved (Gannett); It’s all economics (AP);
NH Makes it tough for More Candidates to enter (Gannett); Harkin flies on emotion
(Staff); Brown Happy on the Outside (Staff); Kerrey emphasis is on health care (Staff)
Tsongas - jobs first priority (Staff); Clinton wants to talk issues (Staff); Issues and
Answers Breakdown - Economy, environment, health care, family and defense (Compiled
78 by Staff); Bush win a mixed blessing (AP); South Dakota not always kind to neighbors
(Commentary); Buchanan always a fighter (AP); Rattled in New Hampshire Bush hits the
campaign trail (AP); Kiddies crunch com flakes with Kerrey (Staff); Clinton airs ad to counter Kerrey commercial (Staff); Tsongas struts liberal agenda (Staff). The day after, the headline read, “South Dakota gets Kerrey-Away; Republicans show Dissatisfaction”
(Staff).
The two neighbor candidates did well. Kerrey won with 40 percent of the vote, followed by Harkin with 25 percent. Clinton earned a third place finish, which again was better than expected, and gave support to his assertion that Tsongas was a regional candidate who could not pull national support. Clinton became the story once again as exit polls suggested that the character issue was hurting Clinton. Had the news media not reported his alleged infidelity and draft evasion, Clinton may have already been on his way to the nomination.
The post-event analysis across the news media called the campaign
“unpredictable,” “murky,” and “without a front-runner.” All the top vote-getters looked like regional candidates as the mixture of contests and candidates created this perception.
Moreover, it seemed that each candidate was claiming to be a winner. And they each were in some way. All the candidates had won or by virtue of doing better than expected, like
Brown in Maine, had some trophy to show. Still the frontrunner in finances and expectations was Bill Clinton, and he had not yet won an event. However, he was expected to hit his stride in the upcoming plethora of southern primaries.
79 Moreover, the Republican campaign would get heated again. Maine and South
Dakota lacked the colorful atmosphere that Buchanan brought to his campaign stops with
his rhetoric and his portable audiences. Yet his challenge in Georgia during Junior
Tuesday, would rock the Republican boat once again.
3.3 Winnowing and Mist Clearing Stage
During the winnowing and mist clearing stage, which in this campaign year were simultaneous due to the front-loaded contests, candidates began to drop out, and the campaign’s competitive pattern began to crystallize. Winnowing can begin right after Iowa and New Hampshire, but in 1992, due to the non-event nature of Iowa and the contests that came close on New Hampshire’s heels, wirmowing did not occur until the beginning of March. This was also the period of time in which the news media covered the campaign extensively. Although the top and bottom tiers were established, they were somewhat fluid because no single candidate dominated completely. Moreover, candidates Brown and
Buchanan were waging low-budget, message-oriented campaigns that allowed them to stay in the contest.
Winnowing is most influenced by the lack of the two M’s - money and media.
Both of the these are influenced by performance. When performance verdict is muddied by multiple candidates winning different events, the news media and potential givers may have difiBculty determining which candidate is the frontrunner or likely nominee. In 1992 momentum may have been replaced by the notion of “traction.” Candidates could dig in a little with each contest they did well or better than expected in. Others would slip, but then
80 gain a hold during their turn to claim some small victory. Because so much happened
during this period of activity, it will be discussed in detail.
There were seven contests taking place simultaneously on March 3,1992 — three
primaries and four caucuses. Harkin won Minnesota and Idaho (both caucus states).
Brown won Utah’s caucus. Clinton won the Georgia primary. Tsongas won the Maryland
primary and Washington’s caucus, and Brown won the Colorado primary. Of all the
candidates, an evaluator without prejudice might have said that Harkin or Tsongas had
done the best. Both had displayed an ability to win in both a caucus and a primary setting.
The only “total” loser during Junior Tuesday was Bob Kerrey who had poor showings
across the board and nothing to show for all of his efforts. Two days later on March 5,
1992, he formally withdrew from the Democratic campaign. While each of the other
candidates did have something to tout as a victory, their bank accounts were nearly empty,
except for Clinton.
Let us begin our examination of these contests by looking at Colorado. Colorado
is a good example of how Clinton’s superior financial status and organization helped him
maintain his frontrurmer status and outlive his competitors even under tremendous stress.
While Tsongas was the first to visit Colorado, mid-May 1991, Clinton spent the most
money there in the days before the event. While Tsongas visited there frequently, Clinton bought more ad time. While Tsongas had one headquarters in a fiiend’s home in Denver,
Clinton had multiple headquarters across the state. Clinton hammered at the issues and
spoke about his “New Covenant” with America emphasizing the forgotten middle class and successfully differentiating himself from Tsongas. Moreover, the Clinton campaign
81 staged campaign events that lent themselves to good pictures for both the broadcast and
print news outlets. Clinton was second only to George Bush in terms of the number of pictures that appeared on the pages of the national and local newspapers during this period.* Tsongas attempted to challenge Clinton, and spent money on campaign ads.
But while Clinton and Tsongas hammered at each other in Colorado with television ads and newspaper quips at each other, Jerry Brown pressed the flesh. Brown drew support from young people - college students, mostly. As to Harkin, the Denver
Post ran a story on February 26 entitled "Harkin, fading in Colorado, plans no visit before the primary”. The Denver Post also paid close attention to their local polls, publishing lead stories on Brown’s growing support, while they also ran stories such as “Tsongas; Tax cuts nonsense” and “Clinton calls Tsongas plan cold-blooded.” The race in Colorado was a negative one between the two frontrunners, Tsongas and Clinton. The day before the primary the Denver Post’s lead stories were “Colorado looks like a win for Tsongas,
Pollster: Clinton to come pretty close.” Even with all the media attention paid to Brown, their predictions still favored conventional wisdom’s picks - Tsongas and Clinton.
While the New York Times was running “Tsongas gaining momentum” on March
3, 1992, Brown stole the show in Colorado. One of Kerrey’s campaign staff was quoted in the Denver Post the day after: “Kerrey would have done much better if he would have had enough money to buy local television time a week ago or earlier. He wasn’t able to define who he was until the last few days. So Colorado voters never got to know him!"
And pressing the flesh was not Kerrey’s forte. Brown, on the other hand, spent nearly
'The Denver Post ran more pictures of Jerry Brown during the two weeks prior to the Colorado primary.
82 100 percent of his campaign time doing just that — speaking to relatively small crowds and individuals. The national press was indeed stunned. At this point the some of the press gave Brown greater credit than before. While he was not considered a “viable” candidate,
Brown was considered more “credible,” “legitimate” and potentially dangerous to Clinton.
Brown had done well in a primary state demonstrating that he was not simply a caucus candidate, like Harkin. And by his strong finish, he was able to qualify once again for federal matching funds.
While Colorado was a boon for Brown, it was a bust for Tsongas. He was expected for finish with a win. He had been leading the state polls the week before. And he had committed himself to Colorado early as a part of his strategy. But he came in third putting a damper on any momentum he hoped to take into Super Tuesday. And it was a local issue, and local coverage of this issue that made a great deal of difference. Nuclear power was Brown’s issue. He gained a great deal of support for his anti-nuclear stance.
During the course of the Colorado campaign, particularly at the debate held in Denver,
Tsongas’ opponents, including Brown, defined and attacked Tsongas’ position on nuclear energy. Moreover, Clinton seeing Tsongas as his main competition and wishing to be rid of him, launched a very negative ad campaign against Tsongas in Colorado casting
Tsongas as pro-nuclear energy, anti-social security and anti-labor union.
The Republican campaign was less interesting. Pat Buchanan had not been able to generate much interest in Colorado and strategically chose to bypass the state. George
Bush generally sent surrogates such as his wife and Jack Kemp, to campaign in the state.
Bush beat Buchanan 67.6 percent to 29.9 percent. It is no surprise that Buchanan received
83 very little press coverage from well-known papers such as the Denver Post. He was not there competing or creating news; therefore, he was not news.
Maryland and Georgia were also key states. Maryland was deemed important because it was considered neutral territory. There were no candidates who could claim a regional afiiliation. So candidates, particularly Tsongas, were suggesting this would be a good test of a “national” candidacy. Moreover, his campaign thought he could do well there. Georgia was key for both the Republicans and the Democrats. For Clinton, Georgia was expected to be his first win. If he did not do well, his candidacy would be put in question. For the Republicans, Buchanan had put most of his eggs in the Georgia basket.
This was his stand. If he did not do well, as he admitted himself^ his campaign might stall.
When the polls opened and the voters voted, Maryland belonged to Tsongas, without much surprise. Clinton came in second. But Maryland was helpful to Tsongas because it counterbalanced a loss to Brown in Colorado, and it demonstrated that he could win outside of New England. In Georgia, the gateway to the South, Clinton was all but the winner before the primary even took place, except that some believed the character issue would hurt him. Already in other southern Super Tuesday states like Tennessee, the pollsters were finding evidence that character issues could turn voters away from Clinton.
Indeed, the Knoxville News Sentinel endorsed Tsongas for reasons of “principled leadership” and “courage.” Even though Clinton was the home-region candidate, there were still a number of southern Democrats feeling a bit uncomfortable with Clinton. In the end, Clinton won the state of Georgia’s Democratic primary giving himself a bump into
84 Super Tuesday. He demonstrated that he could win votes in the south regardless of his character baggage.
Buchanan began an all out assault on Bush in Georgia. He used the relatively little money he had to purchase time for attack ads against the President. These ads focused on moral values and accused Bush of using federal money to subsidize pornography that featured homosexuality, showed no reverence for God, and exploited children. Buchanan also tagged Bush with supporting quotas and hurting the national economy. The Bush campaign responded with quick answers to these ads and a few attacks of their own. The
Bush campaign also used surrogates such as Drug Czar William Bennett, South Carolina
Governor Carroll Campbell, and Dan Quayle to counterattack with criticisms against
Buchanan. Bush even made a personal appearance in the state of Georgia - a two day swing at that. He had only visited New Hampshire three times prior to their primary. On
March 3, 1992, Buchanan had failed to land a knockout blow to Bush. The outcome was bad tidings for Buchanan. He had garnered 36 percent of the vote compared to Bush’s 64 percent. If he could not win in a place that was supposed to respond to his message more than any other state, his campaign was all but finished.
Junior Tuesday’s results created heated competition and somewhat vicious rhetoric that comes with desperation. Tsongas accused Clinton of using “code words” that pointed to Tsongas’ Greek heritage. He called Clinton a “pander bear” who would promise anything to anyone. Clinton responded with by calling Tsongas’ economic policies “cold blooded,” and calling him a “closet Republican.” The anti-opponent rhetoric heated up.
85 Kerrey, who seemed to be on his last legs, also turned negative, focusing on Clinton’s unpatriotic behavior and unelectability.
South Carolina was a stepping stone from Junior Tuesday to Super Tuesday.
Clinton was expected to do very well there, and he did. On March 7, 1992, a Saturday,
Clinton won 63 percent of the South Carolina Democratic primary vote. The competition lagged behind significantly with Tsongas garnering 19 percent and the rest claiming single digit results. While Georgia had usurped South Carolina’s place as the Southern bellwether contest. South Carolina did have an impact on the campaign nonetheless. After his disappointing fourth place finish, Tom Harkin dropped out of the race. He had campaigned hard in South Carolina and lost. Harkin’s traditional Democratic ideals had not taken hold with the voters in 1992.
South Carolina was not the only event held that day. Arizona, Nevada and
Wyoming all held caucuses. But in 1992 as in campaign years prior, caucuses did not appear to yield momentum". In Arizona, Tsongas narrowly beat Clinton. Yet this result was overshadowed by the South Carolina primary in the national press. And state coverage in a caucus state is sometimes quite limited. For example, during the two weeks prior to the Arizona caucus. The Phoenix Republic ran only one front page staft-written article related to the campaign. The rest were from wire or news service sources. That single front-page story was generated on March 7 by a stop made by Tsongas in the state and the speech he gave. Brown had visited twice earlier in the week and his visit did make the back pages. In all there were three localized stories on the caucus. One relating to
^Although generally Iowa has been known to give candidates a boost
86 Brown’s visit, one to Tsongas’ and one on the caucus activities. Clinton did not visit
Arizona during that time period. Thus, Arizona caucus voters were not given a great deal of attention nor were they provided much information regarding candidates and their stances on issues of importance to Arizona. While Tsongas beat Clinton in Arizona,
Clinton won in Wyoming’s caucuses.
On the next day, Sunday, Brown won the Nevada caucuses even though its favorite son, Andrew Moreau, was on the ticket. Prior to Kerrey’s dropping out, the local press in Nevada saw the race as one between Clinton, Tsongas and Kerrey. Kerrey had recently raised close to $20,000 at a Las Vegas fundraiser. Kerrey was also winning points on the nuclear waste issue. Organizationally, Kerrey and Clinton were seen to have the largest contingents, while Brown did have a volunteer staff, but primarily in the Las Vegas area. Tsongas had no formal statewide organization. On March 5, the Las Vegas Review-
Joumal ran this story, “Kerrey’s hints worry state’s top Democrats.” Brown was visiting that day. When the news came that Kerrey had indeed dropped out the race, Nevada
Democrats had to scramble to pick a candidate to endorse.
The day of the Nevada caucuses one of the Junior Tuesday stories in the Las
Vegas Review-Joumal was “Primary wins give Clinton, Bush boosts.” Within this story was a prominent quote by Tsongas saying that caucuses were not his strength. That statement foreshadowed his performance in Nevada. The issues played a role in the outcome of this state. Clinton had attacked Tsongas on nuclear energy and environmentalism and on his position on workers’ right to strike. While Clinton had not visited Nevada during this two week period prior to the event, he did spend money there
87 on ads and literature. While Clinton attacked Tsongas on these issues. Brown benefitted from the attack as well. Nevadans were familiar with their next door neighbor Brown, and he had issue stances that were greatly supported in the state such as a flat tax, opposition to nuclear repositories, and pro-labor credentials. The local press credited Brown’s win in
Nevada to the vast support he had from labor unions, particularly the culinary union. In some ways Brown hurt Tsongas more than Clinton because his ability to do better than
Tsongas made Tsongas’ challenge to Clinton appear weak. Tsongas’ status of runner up was diminished.
At this time as well. Bush got some very good news. The Associated Press put out a story detailing two reports that indicated that the economy was reviving. His campaign felt more secure as they now believed that Buchanan was a lessened if not forgotten threat, and now they could use some good news to get ready for the challenge in
November.
On March 10, 1992 — Super Tuesday— Florida, Louisiana, Massachusetts,
Mississippi, Oklahoma, Rhode Island, Tennessee, and Texas would all hold their primaries. Delaware, Hawaii, Missouri held caucuses.^ Even though there were many contests taking place that day, only one became critical - Florida. Part of the reason for this was the element of competition. Four of the southern states in the Super Tuesday mix had borders with Arkansas, Clinton’s home state. Brown and Tsongas’s strategy was to spend less time in these states since they were considered Clinton strongholds.
Massachusetts was Tsongas’ home state, thus Brown and Clinton swore off competition
^Texas held a beauty contest primar}' and a caucus for delegate selection purposes.
8 8 there. Texas had been thought to be a more competitive state. Tsongas and Brown both
campaigned there. But the polls showed Clinton pulling ahead there as March 10
approached, causing both Tsongas and Brown to abandon hope in a big, expensive state.
Clinton’s strong organization skills and his prior affiliation with a number of Texas
Democrats helped him win many endorsements there. So after a process of elimination, the only state left standing was Florida. Tsongas had earlier known this to be his best chance and had made statements that Florida was important and would serve as his testing place. Tsongas needed a win in a southern state desperately. Clinton wanted desperately to shut Tsongas out of the South. His campaign believed that by doing so, they could enter the northern contests to come with greater confidence.
Florida’s retirement community was feted by all three remaining candidates. Issues like social security, economic security, and support for Israel were discussed. In other communities, Cuban policy and the situation in Haiti took center stage. Clinton’s campaign strategy was to raise the specter of Tsongas’ anti-social security policy as well as his position of raising the gasoline tax.
But Tsongas’ losses were not all due to Clinton’s adept campaign. Part of
Tsongas’ troubles were explained by his own organizational weaknesses which hurt him most cruelly in his media coverage. The Boston Globe. Tsongas’ home state prestige paper, ran a story entitled, “A passive Tsongas runs into a Clinton sledgehammer.” They labeled his campaign “Rolling Blunder.” Already, his campaign had been criticized for its inability to process campaign contributions quickly enough to get federal matching funds in the bank, but now it seemed that disorganization was keeping Tsongas fi’om capitalizing
89 on media exposure. He garnered endorsements from many local and regional newspapers in Florida, but did not use the information for his advantage. He did not make himself available, or his campaign organization did not coordinate the day to day events well enough to make for good, timely pictures. Nor did they provide scheduling information that allowed the press to know where to find him. Moreover, he often missed the deadline to get time on the evening news. Tsongas now led in the category of poor organization that had once been dominated by Bob Kerrey whose campaign was even less adept.
Moreover, Tsongas was not good at attacking back. fBs response ads were sometimes confusing, lacking in directness. When scandal raised its head again for Clinton on the issue of a questionable land deal, Tsongas did not move to take the advantage.
Florida ended up going to Clinton by a substantial amount. Now the spin, generated by the
Clinton campaign and parroted by the press, was that Tsongas was just a regional candidate who had no chance of winning a national election - he had won his home state of Massachusetts, Rhode Island and Delaware. Clinton won everywhere else and he won by large margins.
The Republican campaign was a rout. Bush was finally able to deal the knockout punch to Buchanan. Moreover, the press was beginning to lose interest in Buchanan;
Perot was steadily becoming more visible. It was strongly believed that Buchanan could not win, and he was unable to generate votes consistently across the various states. Also, he could not amass delegates since most of the Republican primaries and caucuses are winner-take-all. While Buchanan vowed to stay in the race until the convention, the media paid markedly less attention to him. As the campaign moved into April and May,
90 Buchanan was briefly mentioned, if at all, by reporters in their stories. Instead they focused on Perot and the likely battle between the independent candidate. Bush and
Clinton.
Only one week after Super Tuesday, the remaining candidates in the Republican and Democratic contests would face each other in Illinois and Michigan. These two states were key parts of the Clinton campaign strategy. He began campaigning in Illinois early in 1991. He chose a national campaign manager who was fi'om Illinois and very well connected there. The Arkansas governor sought endorsements and support fi'om the
Democratic party officials and office holders. It did not hurt that Hillaiy Clinton had
Chicago roots. By the time the contest took place, Clinton had a superb organization and a great deal of “official” support. At this point in the campaign, Clinton was seeking to finally be rid of his competition and secure claim the title of frontmnner and nominee apparent for good.
For Tsongas, the situation was critical. He badly needed a win in Michigan or
Illinois. His reasons: first, he needed to generate more financial contributions, and second, he needed more fi-ee and positive press. Tsongas did something he did not generally like to do — he fought very aggressively for votes. He spent some of his money on ads that criticized Clinton’s handling of the Arkansas’ economy, the state’s poverty level and poor educational rankings. And he defended himself against the “lies” that he claimed Clinton told about him. But Clinton continued to attack. Clinton had enough money and was expecting more. He could respond quickly with his own ads. And, as the clear frontrunner.
91 he could generate a great deal of “free” media coverage. Thus, Tsongas did not really have
a chance in Illinois.
In Michigan, it was not so much Clinton’s organization, but the very nature of the
state and its interaction with candidate Tsongas that hurt him. Michigan’s population is
composed of many working class individuals, most of them members of some time of
labor union. It was a place that his pro-business message did not resonate. Nor was
Tsongas the candidate of choice for most African-American voters. They found more of a
representative in Clinton. On the positive side for Tsongas, Brown attacked Clinton at
every point. This involved the now infamous attack by Jerry Brown on Hillary Clinton and
her law firm’s role in a real estate scandal. But Clinton persevered by using Brown’s
attacks and the media’s investigative stories (the Washington Post introduced Hillary
Clinton’s potential role in Whitewater) to demonstrate two things; first, how popular his policy agenda was since voters continued to support him; second, that America really wanted a candidate who would beat George Bush in the general election. In Chicago, his popularity was high. The Detroit News reported a poll of likely Democratic primary goers that suggested that Clinton was the favorite, and it was a close call between Tsongas and
Brown. This was a devastating blow to Tsongas. He ended up getting third place in
Michigan and a rather distant second in Illinois. He was not even close to Clinton’s delegate totals and was nearing the end of his money. He suspended his campaign on
March 19, 1992.
Buchanan was also in a position of falling expectations. He campaigned in
Michigan, but was so far behind in the polls that even Buchanan’s bullishness was
92 subdued. The same could be said of Illinois. The Republican nomination campaign would
now be conducted in form only. From this point on, Buchanan’s name virtually
disappeared from the broadcast and print stories of the campaign.
With Tsongas out of the race, Clinton and Brown were left as competitors. No one
believed that Brown could win the nomination, but he had done reasonably well for
someone who had started off trailing behind all the other candidates in viability
assessments. Brown had actually started ahead in name recognition. Since he was the only
standing challenger to Clinton, he was given coverage by the press. Yet it was to continue
only as long as he was news, as long as he could continue to pull off the surprise. He did
in Connecticut. This was the first contest in which the contest was truly a two-person
race. The framework of the coverage changed. Brown became the anti-Clinton. Brown
won in Connecticut’s Democratic primary by 3 percent of the vote. He had received 38
percent — one of his better showings. Clinton followed behind with 35 percent. Tsongas,
suspended, but remained on the ballot. He received 20 percent. In an article from the
Denver Post (March 26, 1992, P. 4A) Brown suggested that while the media picked
Clinton as the wiimer, the people of the state of Connecticut chose him.
The ever increasing phenomenon of the fickle public raised its head. Now that it looked like Clinton would be the nominee, the public decided that they were not satisfied with the candidate the process produced. Exit polls in Connecticut suggested that
Clinton’s support was soft, and this finding was repeated in subsequent exit polls in later primaries. Moreover, the voters’ choice was now limited to two candidates. It would be expected that Brown, at this point, could probably gamer more votes, particularly from
93 ex-Haiidn supporters. Then another victory took place. One week later. Brown won the
Vermont caucuses. He won by a very wide margin (although the Vermont caucus
outcome was not as widely reported by the media as the Connecticut primary outcome).
Clinton was third behind uncommitted while Brown took 46 percent of the vote.
This boost came at a crucial time for Brown. These wins gave him momentum going into the next state. New York. Anything positive going into the New York primary was considered a bonus for Brown. His boost was somewhat diminished by Clinton’s win in the Kansas primary on April 7, 1992. Clinton won with 51 percent of the vote.
Minnesota held caucuses the same day, but also held a beauty contest primary that was virtually a tie between Clinton and Brown, but with Clinton squeaking by with the win.
New York’s primary was compared to the running of a gauntlet. Brown and
Clinton, but particularly Clinton, were probably concerned about the tough media market they were entering. Not only was it difBcult to purchase time given the high costs (not as much an issue for Brown who had little money, but at this point even Clinton was borrowing heavily on a line of credit), but New York reporters were considered some of the toughest critics on politicians..
Entering a state with an ethnically diverse ethnic population, Clinton had made a strategic error by playing golf at an all-white country club in Little Rock before the New
York campaign outing. At this point the New York Times, who had started their investigative reporters on assignments earlier in the campaign, published a story that charged that Clinton had exempted his ofiBce of the Governor from ethical standards that he imposed on the statehouse. These scandals required Clinton again to go on the
94 defensive. From infidelity, draft dodging, questionable land deals, and now the charge of not applying ethics to himself and his office while governor — Clinton seemed to have an infinite supply of scandals in his past. Many pundits and professional campaigners were practically astonished that a candidate could survive one scandalous hit after another. The media watchdogs were doing their job, but the nomination process seemed immune.
Clinton’s campaign continued to forge ahead into New York. Brown and Clinton went at each other forcefully while they both took attacks from the media.
The New York tabloids boldly printed headlines that focused on Clinton’s character problems. A new story arose that dealt with Clinton’s now infamous experimentation with marijuana and his claim of “not inhaling.” Reporters closed in on
Clinton asking tough questions that he answered, in the media’s view, obtusely. The issue became not the drug use but rather Clinton’s “tendency” to elude giving straight answers.
His opponents and critics came up with a different name for his elusiveness. They called it prevarication. When they were less polite, they called it lying. A few days later, a stoiy was released by the Associated Press that documented more details on Clinton’s alleged draft evasion. These details had not been mentioned by the Clinton campaign during their defense of the original charge.
New York voters responded with derision. Clinton could no longer count on cheering supporters at rallies. The nominee-apparent was confronted at campaign events by accusations of racism, disloyalty to country, and immorality. Moreover, the media began sending out signals that Clinton might be cracking under the pressure of intense media scrutiny - lashing out at his staff and protestors on several occasions. And while
95 hostile voters and a combative media maricet interacted with the Clinton campaign, Jerry
Brown attacked Clinton on issues. He attached Clinton’s record on worker safety in
Arkansas and his lack of support for civil rights. Arkansas did not have its own civil rights act. And it was at this point that many Democratic leaders, realizing that Clinton was going to be the nominee and that Bush could lose, rallied around Clinton. Ron Brown,
National Democratic Party Chair criticized Jerry Brown for his negative campaigning in
New York, actually suggesting that the trailing candidate should get out of the race.
Brown not only was chastised by the party leadership, but was also under attack by
Clinton. Clinton attacked Brown’s 13 percent flat tax proposal as a boon for the wealthy.
Other prominent New York Democrats joined with Clinton in criticizing Brown’s tax proposal. However, Brown hurt himself as well when he began talking about Jesse
Jackson as a vice-presidential candidate. This was not a new idea for Brown, but in the context of New York, particularly in the city, Jesse Jackson had different connotations. He was not only viewed as a black civil rights leader, but the Jewish community perceived
Jackson as anti-Semitic. His reference to “Hymie town” had not been forgotten, and
Brown alienated a group of potential supporters. Brown magnified the problem by using the term “money lenders” during a diatribe at a Wall Street political rally. Now the New
York press turned on Brown. These two strategic errors, touting Jackson as vice- presidential material and using the words “money lenders,” put his campaign in jeopardy.
Had these comments not been reported widely in the New York press, the damage might not have been so costly. Moreover, Tsongas was now hinting, sensing Brown’s weakness, that he might reenter the race depending on how well he did in New York.
96 Brown’s message throughout the campaign dealt with anti-Washington ideas -
taking back the government, getting rid of special interest influence, putting limits on
campaign contributions and so on. But Brown came under scrutiny about his campaign
finances when CNN broadcast a tough story on Brown’s 1-800 number. A reporter,
calling the number, said the operator at the other end told him how he could get around
the limit by having family members also contribute $100 each. Brown reacted with vigor
to the accusation. And he would do the same when the Los Angeles Times published an
article discussing Brown’s patronage of judgeships to campaign donors while he had been
governor of California. So New York evolved into potentially the most “critical” contest
of all. Both candidates were heavily scrutinized. When Brown became the only challenger
to Clinton, he found himself in a new media situation. With the complexity of a truly
multi-candidate race gone, the press focused in on the two remaining candidates for a
number of reasons; First, Clinton, the fi'ont-runner, wanted to get rid of Brown and shift
to the general election campaign, so he began his assault on Brown. Second, with only
two candidates in the game, the press had to focus on Brown, by virtue of position of
being the only other candidate left standing coupled with the fact that he had recently
begun to show signs that he might gamer support enough to have some role at the
convention.
Clinton, at this point in the campaign, tried the bypass strategy. He began to bypass the traditional news media and focus on alternative media, using them to go
directly to the people to counteract the negative images created in the news and Brown’s
97 television ads^. And in the end, this strategy, as well as Clinton’s ties to the local party organizations, as well as other campaign efforts, paid off. Clinton won the New York primary with 41 percent of the vote.
The Brown campaign, however, was crucially hurt when Tsongas, the non candidate, came in second with 29 percent of the vote followed by Brown’s 26 percent.
The draft Tsongas campaign had taken votes away from Brown. But Tsongas did not end up reentering the campaign. The results had only met one of Tsongas’ requirements. He knew he could not beat Bill Clinton. At this point, Clinton had amassed 1,300 delegates.
And again, he had come out of a tough situation amid numerous allegations with a win.
Clinton’s post-New Hampshire spin of the “comeback” Idd seemed applicable once again.
While he won, his poll numbers suggested that his supporters were doubtful. Now Clinton had a window of opportunity. With his competition nullified, he could begin to remake his image during the next stage of the campaign.
It should be noted that Wisconsin and Kansas were also holding primaries.
Minnesota held a non-binding beauty contest primary. Its delegates were chosen by caucus. Clinton and Brown paid little attention to Kansas. New York with its many delegates was a more pressing contest. Moreover, New York was where the press were paying attention. As to the results, Clinton took a majority of support in Kansas, and won by three percentage points in Wisconsin. However, Brown’s decent showing in Wisconsin received very little attention in the news. On the night of April 7 and the day of April 8,
is assumed that these guest appearances on alternative media reached a différent audience and were more in the control of the candidate. Moreover, they were reported in the mainstream news media as well.
98 the story was New York, the meaning of the results, and the disenchanted voters, both
Democrats and Republicans.
3.4 Pre-convention and General Election Maneuvering Stage
After the New York primary, there was little doubt that the Democratic nomination campaign was over. Four days later, on the day of the Virginia caucuses.
Brown pledged his support to Clinton. This was two days after ABC broadcast their lead story, “Serious charges against Jerry Brown regarding illegal drugs”.
But what about the Republican nomination? Buchanan was still running even though he admitted that George Bush would get the nomination. But he recalibrated his campaign as “fight for the soul of the Republican party” and as a fight against the
Washington establishment. Bush’s response? He ignored Buchanan. And so did the press.
From early April to the end of the campaign season, Buchanan became a non-factor. In
Oregon, for example, during the two weeks prior the Oregon primary held in May, there was no mention of Buchanan on the front or in the interior pages of The Oregonian.
Brown, however, regardless of his pledge of support to Clinton, was in Oregon trying to acquire votes for the sole purpose of winning more delegates to have some impact on policy at the convention.
The campaign had shifted to the general election. The soon-to-be party nominees.
Bush and Clinton, were thinking ahead to their conventions and to the general election campaign. And they were thinking about Ross Perot. While they earned wins in
Pennsylvania, Washington, D C., Indiana, North Carolina, Nebraska, West Virginia,
99 Oregon and Washington, Arkansas, Idaho and Kentucky, and then headed into that last big day of contests on June 2 (Alabama, California, Montana, New Jersey, New Mexico, and Ohio), the specter of Perot haunted both Clinton and Bush. In North Dakota’s
Democratic beauty contest (the final contest of the season), Perot received 29 percent of the vote while Clinton’s write-in percentage was 14.5 percent. The media changed its focus and adjusted its behavior accordingly. During the series of primaries and caucuses after New York, the number of reporters traveling with the candidates, particularly the challengers Brown and Buchanan, dropped off significantly. At this point, only the results of the more populous events were reported. And when the media, both national and local, are not excited, it is doubtful that the voters will be either. Turnouts in the subsequent contests were predicted to be and were relatively low. The media knew what the outcomes would be and instead began to focus on what the exit polls were telling them.
And they focused on the phenomenon of Ross Perot and what it would mean to the campaigns of Clinton and Bush. At this point, Clinton took the hardest hit, showing up in third place behind George Bush.
But this was to be the peak of Ross Perot. With his prominence there also came scrutiny. Ross Perot’s life and career were discussed and investigated. While prior to April a large portion of Perot’s coverage dealt with his petition signature gathering and the swell of grass roots support he was receiving, as he became viewed as a real factor in the upcoming campaign, the watchdogs responded with stories about his experience and background, evaluations of his policy stances, and analyses of why he was gaining so much support fi'om what appeared to be a populist movement among the American
100 people. At the end of the campaign season, each party had chosen its candidate, and an
independent challenger had emerged.
3.5 Conclusion: The Media and the Campaign
Observers of presidential nomination campaigns are hard put to describe the role
that the media plays in any exacting terms. While we know so much, we know so little. In
1992, it was not unusual to hear voters expressing their opinion that news media distort
candidate images, misrepresent candidate policies and are too harsh in their critiques of the
candidates’ personal lives. That role of watchdog by the media, which has long been
thought to safeguard democracy, is now viewed by many as too harsh, biased and
potentially suspect. Yet many are of the opinion that without the media voters would find
it difficult to learn anything about the candidates, and candidates would have difficulty
communicating to the populace. Regardless of one’s verdict of the media as protector or
disrupter of democracy, the fact remains that the news media does play a role. Their primary role is that of information conduit. Without the media we would not know who was running, where they were running, how they were doing and what they were saying.
As the prior description of the campaign suggests, this role can be Influential. What if the news media had never covered the character questions about Clinton? What if Paul
Tsongas had been better at playing to the media? What if he had managed his schedule with the media in mind more often? What if Jerry Brown’s three percentage point loss in
Wisconsin had been called a better-than-expected finish instead of a loss? What if Tom
101 Harkin or Bob Kerrey received as much media coverage from national outlets as Clinton had at the outset? Would it have mattered? If it does matter, then there are implications for the need of a politically responsible news media. In the following chapters, I will explore a number of these and other questions by analyzing the media coverage of the
1992 nomination campaigns using national and state level media coverage data.
102 CHAPTER 4
NATIONAL AND LOCAL NEWS MEDIA COVERAGE COMPARED
4.1 The Question of Media Homogeneity
It has been nearly twenty years since Thomas Patterson’s ground breaking work
on media coverage of presidential nomination campaigns. His study of the 1976
presidential nomination examined the nature of media coverage and its potential to
increase the awareness and knowledge of voters. One of his findings was that media
coverage of the campaign was very similar across outlets, both national and local,
regardless of geographic location. However, as noted in earlier, Patterson’s measurement
of media “does not provide a very precise indication of what was reported about the
presidential candidates” by these outlets but rather concentrates on the news themes and
news makers that dominate the news media. (Patterson 1980, P. 12). Thus, his findings of
homogeneity reflect general tendencies in news coverage rather than similarity grounded in
an actual reflection of the content, scope, focus and direction of news media coverage of
candidates.'
'Patterson's data collection involved the random selection of6,567 political news stories appearing in nine outlets (the three national networks, two news magazines, four daily papers) between January 1,1976 and November 2, 1976. A code was given to each of these stories to indicate the date it appeared in the outlet, whether it was accompanied by a news photo or film, its length, and its position (front or back page). Each story was also identified by its principal subject and actor, and whether it reflected favorably or unfavorably on this actor. The data were further analyzed for references to the candidates' personal and leadership qualities,
103 This assumption has permeated the political science discipline. But while this may have been true in 1976, how true is it today when two decades have brought broad changes to the way campaigns are conducted and have seen great alterations in the way the media system covers these campaigns?
Some of these changes have affected the way political campaigns are covered in general, while others have implications specifically for local coverage. Local press and television stations offer larger doses of national news than in the past.^ And their resources have allowed them to actively participate in covering the candidates and campaign events.
The number of local reporters shooting segments outside of the New Hampshire polling places in 1992 has increased exponentially since the late 1970s.
Moreover, there have been changes in the tactics used by the candidates themselves. For example in 1992, George Bush led the candidates in utilizing the
“bypass” strategy, going straight to local outlets which were thought to be less critical and more likely to pass on the candidate’s message without much, if any, interpretation or evaluation (Rosensteil 1994).^
styles, backgrounds, issue stands, values, campaign progress, and similar things. These were codes as references. For example, Patterson presents an example of how a story on Jimmy Carter was broken down: The references extracted were: Carter is a farmer. Carter did a Playboy interview. Carter is leading in the polls. Carter has a restrained style, and Carter is clear in his issue stands. However, breaking down a 250 word story into 5 references discards information and is potentially open to greater subjectivity.
"In many instances the local broadcast stations air breaking national news before the networks. Networks carmot deny their afBliates news footage because they worry about losing them to Cable News Network and other broadcast news organizations. Thus, local affiliates can reach the viewer with national news before the national networks.
^This is nothing new. Richard Nixon believed the national press corps was biased against him. He attempted to evade the national press and go directly to local news organizations if news media exposure was necessary. Contacting the local press goes back even farther to Charles Michelson, a Democratic National Committee publicist during the Hoover and Franklin D. Roosevelt eras. Sometimes the motivation has been to skirt the
104 While presidential candidates have paid more attention to the local news media, scholars have not: We do not know very much about how the local news media cover national political campaigns. Do th ^ simply reflect the content of the national media, and if so, which national media - assuming that they are not all alike?
Two studies suggest that variation exists. Patterson and Davis’s 1987 study focused on the New York Times and the Svracuse Post-Standard and found that during the nomination campaign of 1984 during January and February, the different Democratic candidates were perceived differently by the two news media outlets, which in turn earned them differential degrees of coverage. In addition, Buell’s study, also published in 1987, found variation in candidate coverage by five news media outlets examined during the
1984 presidential nomination campaign. Buell found differences primarily between small and large papers. The two small New Hampshire papers he examined carried fewer election stories than the larger metropolitan papers. There were also variations in how much attention the outlets gave to minor candidates and how much issue coverage was generated.
In a recent study, Bartels (1996) found different relationships between an elite newspaper, metropolitan newspapers and network television news and political actors in terms of who led the policy agenda and who followed. Bartels examined the New York
Times, the ABC World News Tonight and three local newspapers the Hartford
Courant, the Orlando Sentinel Tribune, and the Seattle Times for a three-year period
national press and other times it has been the simple pursuit of spreading a message to all possible media outlets. Local papers had news holes to fill, but lackal reporters to gather it Consequently, they were willing to publish what others sent liim.
105 during the Clinton administration. His analysis suggested that the aggregate impact of the
New York Times on the executive branch and congressional agendas far exceeded that of the local newspapers or the ABC News. Moreover, he demonstrated that the local newspapers covered relatively more of particular issues (Medicare and NAFTA, for example) and less of other issues (Bosnia and Whitewater) than the national news outlets.
This chapter examines the question of media homogeneity. Using the 1992
Republican and Democratic nomination campaigns as the focus, I examine the characteristics and degree of coverage similarity o f national and non-national (state and metro) newspapers as well as one national network evening news program. I explore similarity in three dimensions; amount of coverage, direction of coverage and coverage content.
4.2 Media Coverage of the Campaign: Attention” Differentiation Among Outlets
One question asked about the media’s coverage of presidential politics is — how much coverage was generated about the campaign overall? Are the news outlets similar when it comes to how much of their coverage they devote to the presidential nomination campaign and its players? Under the assumption of media coverage homogeneity the expectation would be that outlets cover the campaign at the same pace - printing and airing stories when something important is happening, focusing on non-campaign topics when there is a lull in activity, or ignoring the campaign when events of greater import shove it aside.
106 In order to ascertain whether such variation actually exists, I compare the number
of stories/news segments devoted to the campaigns as well as the number of days that the
nomination campaigns were featured as the “top” story.
4.2.1 National Media Attention to the Campaign
In 1992, the national press seemed to differ not only in how much coverage they dedicated to the campaign, but how much prominence they gave the campaign. The
Washington Post generally led in terms of overall front page space to the campaign. In
January, February and March, the Washington Post's allocation of campaign stories to the front page averaged 31 percent. They also devoted more headline space to the campaign.
As Table 4.1 illustrates, the Wall Street Journal generally lagged behind in terms of their aggregate coverage per month. The New York Times’ attention to the campaign was only greater than the Post’s in April.
ABC's World News Tonight, while devoting somewhat less coverage to the campaign relative to the national press, did feature the campaign prominently when it did.
Looking at Table 4.2 we see that in March of 1992, ABC led their news with a campaign story 71 percent of the time. This equals 22 days in March opening with a campaign related story. In February ABC led their news broadcast with a campaign story 15 of the
29 days.
107 Table 4.1: National Newspaper’s “Attention” to the Campaign: Percentage and Number of Front Page and Lead Campaign Stories
Month New York Times Washington Post Wall Street Journal % Front % Lead % Front % Lead % Front Dec. 1991 10% (19) 7% (12) 10% (16) 7% (10) 7% (8) Jan. 1992 13% (24) 9% (16) 27% (41) 17% (26) 14% (17) Feb. 20% (35) 10% (17) 32% (46) 21% (30) 13% (15) March 26% (48) 13% (24) 33% (51) 27% (41) 15% (19) April 20% (35) 16% (28) 17% (26) 8% (12) 11% (15) May/June 3 13% (27) 6% (13) 17% (30) 24% (23) 7% (9)
Note: The Wall Street Journal’s lavout stvie is eenerallv six vertical columns consisting of stand alone stories and columns such as the Washington Wire, What’s News and others. Generally all stories are considered “lead stories”. If a short story appears in one of the informational columns in the top half of the paper (in Washington Wire, for example), it could be considered a lead story. If below the fold, it could considered a front page story. Since only two political stories appear below the fold during the time period examined. I have eliminated the distinction for the Wall Street Journal.
Front and lead are differentiated in the following manner Front refers to stories appearing on the front page as a whole. Lead refers to stories that appear on the “top half” or “above the fold” of the newspaper (the most prominent area). Most “lead” stories appear in the space immediately below the newspaper’s masthead. Cell entries in parentheses are the actual number of stories about the campaign that appeared on the front page that month. Percentages are representative of the number of campaign stories relative to the estimated total number of stories appearing for the month.
108 Table 4.2: ABC's World News Tonight’s ‘^Attention” to the Campaign: Percentage and Number of Campaign Story Segments and Lead Segments
Month ABC World News Tonight % Story Segments % Lead Segments
December 1991 5% (15) 23% (7)
January 1992 7% (24) 43% (13)
February 12% (39) 52% (15)
March 18% (61) 71% (22)
April 11% (36) 33% (10)
May/June 3 7% (28) 24% (8)
Note: ABC World News Tonight ran an average of 11 news segments per broadcast This average was used to compute the percentage of campaign story segments relative to the total average number of segments. Lead segments are those that appear at the outset of the news broadcast
While the actual number of stories varied among the national news outlets, the overall pattern of how much attention was allocated appears similar looking somewhat like a normal curve. As seen in Figures 4.1 and 4.2 each outlet’s coverage peaked in March and trailed early and late in the process. The Washington Post diverged somewhat from the decline during May and June. The reason for their relative increase in coverage is due to their extensive coverage of the Los Angeles Riots and the candidates’s responses to it, the Murphy Brown criticism by Vice-president Quayle and Bush’s reaction, continued coverage of Ross Perot, several investigative reports of the candidates and primarily, the “local” coverage of the District of Columbia’s primary.'*
“^Examples of the stories are the foUowmg; Perot Offered to Buj’ Island for Drug Stings (5/30/92); Clinton’s Hard Lesson in Pragmatism; Arkansas’ Powerful Timber Interests Take Priority Over Environment (5/26/92).
109 30
20 i
NYT
WP
WSJ
ABC Feb March A pni
MONTH
Figure 4.1: National News Media’s Campaign Attention Levels: Percentage of National Media Campaign Coverage per Month
Analysis of variance, a procedure that divides the total variation of dependent variable, in this case campaign coverage, into components/effects and produces tests for the statistical significance of each component, in this case each national outlet, suggests that there is statistically significant variation with regard to total campaign stories per month (Significance of F=.01). Yet when the average percentages of lead stories per month are analyzed the differences are less visible (Significance of F=. 17) when comparing the New York Times. Washington Post and ABC’s World News Tonight).
110 4.2.2 Non-National Media Attention to the Campaign
The picture the non-national press paints of the campaign is less elegant. Table 4.3
presents the number of “lead” campaign articles that appeared in the listed state and metro
news outlets by month. Table 4.4 reports the same information for front page coverage.
Analysis of variance suggests that there was more significant variation across local news
outlets in the broader category of front page stories (Significance of F = .07) and less
statistically significant variation in the category of lead stories (Significance of F = .11) just as in the case of the national media.
Table 4.3: Local Newspaper’s “Attention” to the Campaign: Percentage and Number of Lead Campaign Stories
Newspaper Dec. 1991 Jan. 1992 February March April May/June 3
DMR 1% (1) 6% (10) 18% (26)* 13% (20) 3% (4) 5% (8) MUL 4% (6) 10% (16)* 10% (15)* 5% (8) 1% (2) 3% (4) editorial/1 editorial/3
BDN 5% (7) 13% (19) 14% (21)* 9% (14) 2% (3) 10% (13)
AJC 5% (6) 8% (10) 16% (19)* 16% (20)* 6% (7) 10% (13)
SLT 8% (12) 17% (27)* 9% (13) 15% (23) 5% (7) 6% (10)
MH 3% (5) 6% (10) 7% (10) 12% (19)* 4% (6) 8% (13)
CT 3% (4) 9% (14) 11% (16) 15% (23)* 4% (6) 8% (14)
MS 3% (4) 6% (10) 3% (4) 10% (15)* 3% (4) 2% (3)
NNDP 3% (5) 8% (13)* 3% (5) 8% (12)* 2% (3) 3% (5)
Note: Editorial denotes a &ont page editorial. These are not included in the percentage tabulations. Key. AJC=Atlanta Journal Constitution (GA); BDN=Bangor Daily News (ME); CT=Chicago Tribune (IL); DMR=Des Moines Register (lA); MUL=Manchcster Union Leader (NH); MH=Miaini Herald (FLX MS=Miiwaukee Sentinel (WI); NNDP=Newport News Daily Press (VA); SLT=Salt Lake Tribune (UT). An asterisk (•) designates when the news outlet gave the campaign the most prominent attention.
III Table 4.4: Local Newspaper’s ‘‘Attention” to the Campaign: Percentage and Number of Frontpage Campaign Stories
Newspaper Dec. 1991 Jan. 1992 February March April May/June 3
DMR 4% (6) 16% (25) 21% (31)* 18% (28) 5% (8) 9% (15)
MUL 13% (20) 19% (30) 23% (34)* 12% (19) .5% (8) 9% (12) editorial/2 editorial/2 editorial/4 editorial/3 editoriaI/2 editorial/2
BDN 6% (9) 15% (25) 19% (26)* 15% (24) 6% (9) 18% (25)
AJC 8% (10) 16% (20) 19% (22) 20% (25)* 11% (13) 18% (24)
SLT 10% (15) 2394(35)* 16% (23) 21% (32)* 11% (16) 8% (14)
MH 10% (15) 10% (16) 12% (18) 23% (26)* 7% (10) 9% (16)
CT 8% (12) 15% (23) 23% (34) 27% (42)* 10% (15) 12% (20)
MS 4% (6) 10% (15) 8% (12) 14% (21)* 5% (8) 6% (10)
NNDP 6% (10) 26% (38)* 16% (23) 23% (35)* 21% (31) 20% (28)
Note; Editorial denotes a front page editorial. These are not included in the percentage tabulations. Key; AJC=Atlanta Journal Constitution (GA); BDN-Bangor Daily News (ME); CT=Chicago Tribune (IL); DMR=Des Moines Register (lA); MUL=Manchester Union Leader (NH); MH=Miami Herald (FL); MS=Milwaukec Sentinel (WI); NNDP=Newport News Daily Press (VA); SLT=Salt Lake Tribune (UT). An asterisk (•) designates when the news outlet gave the campaign the most front page attention.
Figure 4.2 presents the graphic representation of the level of campaign coverage of the non-national newspapers. There is some similarity to the national pattern, but on the whole, the distributions are more diverse. Rather than normal distributions, bimodal and skewed distributions are more visible. This suggests that the local outlets’ coverage
11 2 2 0 -
AJC
BDN
CT
DMR
MH
MS
MUL
NNDP
SLT D ec1991 J a n 1992 Feb Apifl May/June
Figure 4.2; Local News Media’s Campaign Attention Levels: Percentage of National Media Campaign Coverage per Month
diverges somewhat from the national outlets’ coverage patterns. This is theoretically
explained by the dual nature of the presidential nomination process which is both a
national and a local phenomenon. Local outlets will not only focus on the process when it
is close to home, but they will focus attention on the process during times of special merit
or interest. National outlets are far more constant and universal in their coverage, but they,
too, devote more space and time when campaign activity is high or unusual in nature.
113 Aside from these general observations, there are some interesting standouts among the local press. The highest level of attention focused on the campaign came from
Virginia’s Newport News Daily Press with an average of 27 stories per month or 18% of their front page stories devoted to the campaign. The Daily Press was followed by the
Chicago Tribune which ran an average of 24 stories per month on their front page. The newspaper that covered the campaign the least was the Milwaukee Sentinel, averaging 12 stories per month, or 8% of their front page coverage. The mid-level coverage papers were the Atlanta Journal Constitution, the Bangor Daily News, the Des Moines Register. the Manchester Union Leader, and the Salt Lake Tribune. The Manchester Union Leader would rise to the first tier in terms of coverage if front page editorials were included.
Examining the number of stories can exaggerate the actual amount of coverage credited to an outlet. While the Daily Press did print a large number of campaign stories, the average size of each story was about 111 words. The Chicago Tribune’s story size average was 162 words. A few large stories and many short stories can make an outlet appear very dedicated to political coverage if one relies on the number of stories as the means of capturing coverage volume.
As to the salience level given to the campaign, if we examine the number of campaign stories that appeared on the top half (above the fold) of the newspaper (the most visible and coveted space), there is less variation overall.* But some newspapers do stand
*The greater similarity in the number of “lead” stories is due to the fact that only certain types of stories are deem ^ “lead” material. The journalistic norms that guide editors in placing “lead” stories are rather universal, and thus there is less difference. The editors’ general pronouncement of the newsworthiness of a “surprising or shocking event” is such a norm. Events of national security are always considered newsworthy. Examples would be George Bush falling ill on his trip to Japan or when a long-shot candidate wins a primary that no one expected him or her to win. Moreover, the media emphasize the returns, particularly for important contests, and
114 out. The Salt Lake Tribune had unexpectedly higher numbers of lead stories relating to the campaign than many of the other presses. This is due to the deference they placed on activity by the president.
These differences we do see from the examination of campaign stories are smoothed somewhat by summing or averaging across an entire time period. If we compared daily or weekly story totals, or broke down stories by candidate or subject, we would see potentially greater differences between and among the outlets. (See Appendix
B for national and non-national media outlet weekly candidate coverage trends.)
4.2.3 National and Non-national Campaign Attention Compared
Campaign coverage appears to vary at the national and local level. One obvious difference already suggested is that the national media outlets cover the campaign similarly in terms of when they focus the greatest and least amount of attention on it. The local outlets, by comparison, have different patterns that are generated by local campaign activity. This could be the result of reporters focusing on their own state contest or by increased candidate activity in the locale. But differences in how much coverage the outlets give the campaign are somewhat dampened by the vast attention that certain time periods in the campaign receive regardless of the state’s temporal placement within the order of primaries and caucuses. The initial contests (in 1992 New Hampshire was considered the first “real” test of candidate strength) and Super Tuesday by far received
these are generally reported as “lead” stories.
115 the most coverage. National and local media outlets are similar in that they both focus on important primaries and caucuses.
Differences are found in the relative number of stories placed on the front page on an average month. The New York Times and the Washington Post dedicated slightly greater amounts of space to the campaign tfian the local outlets. The Wall Street Journal and the ABC News fall behind most of the local outlets in terms of the percentage of coverage they gave the campaign. If we include only the first few segments of the ABC
News rather than the entire newscast, this percentage would increase. This leaves only the
Wall Street Journal, a national, but specialized newspaper, to fall behind the local outlets in terms of coverage amount.
We can statistically assess the relationship between the outlets’ coverage levels by utilizing a measure of association such as the correlation coefiScient. The value of the correlation coeflBcient is that it allows for the determination of the direction as well as the magnitude of association. While all the media outlets are positively related to other outlets when they are correlated, the magnitudes of the relationships vary. The results presented in Table 4.5 suggest that of the national news outlets, only the ABC News and the New York Times are correlated. This supports the assumption that the national network news and the New York Times cover national campaign events most similarly.
The Atlanta Journal Constitution, the Des Moines Register, the Milwaukee
Sentinel and the Chicago Tribune are all correlated strongly with the Washington Post.
These papers correlate with the Washington Post due to shared patterns of increasing or decreasing attention to the campaign. Their absolute amount of attention differs. None of
116 the local papers comes close in matching the percentage of coverage given by the
Washington Post.
There is also a cluster of regional papers: the Des Moines Register, the Chicago
Tribune and the Milwaukee Sentinel all share boimdaries and some audience overlap.
Most other clusters of correlated local papers are explained by the timing of the contests.
Newspapers located in states with contests in March are more likely to have their peak coverage during that month. The Manchester Union Leader and the Newport News Daily
Press stand out as being rather unique. The Union Leader is only correlated with the Des
Moines Register. This is expected since th^r are both the first contests of their type, primary and caucus. Other than the Register, the Union Leader, like the Daily Press is not correlated with any other national or local outlet. The Dailv Press's coverage distribution is bimodal. Peaking in January with much coverage of Douglas Wilder’s campaign ups and downs and later giving substantial coverage to the campaign during the height of its activity. The Union Leader is the most skewed. It covered its own event substantially compared to others. Moreover, when Buchanan was no longer a competitive challenger, the Union Leader’s level of coverage declined since much of it was devoted to the
Republican campaign.
When we compare the national and local outlets’ coverage of the campaign a few relationships are visible. First, the New York Times appears to have the least association with the local press in terms of per month story coverage (due in part to the distribution of stories - it had by far the most “normal distribution” with a steady incline of stories and then a steady decline). The Washington Post’s coverage has the largest number of local
117 Table 4.5: Campaign Coverage Similarities Among National and Local News Outlets
Outlet ABC NYT WP WSJ AJC BDN CT DMR MH MS MUL NNDP SLT
ABC ------.986 O.S. as. as. as. .842 as. .799 as. a a a a a a
NYT ------as. as. as. as. as. as. as. as. a a a a a a
WrP ------as. .946 .536 535 .878 as. 530 a a a a a a
WSJ .. ------____ as. as. as. as. as. as. a a a a .868
AJC ------520 .826 .843 as. .856 a a a a a a
BDN ------as. .794 as. a a a a a a a a
CT ------—— .899 .807 .703 a a as. a a
DMR ------a a .780 .733 a a .740
.MH a a a a a a a a
MS ------as. a a 838
MUL ------a a a a
NNDP as.
Key: AJC^Atlanta Journal Constitution (GAX BDN-^Bangor Daily News (ME); CT=Chioago Tribune (IL); DMR=Des Moines Register (lA); MUL=Manchester Union Leader (NH); MH=Mianri Herald (FL); MS=Milwaukec ScntiiKl (WI); NNDP=Newport News Daily Press (V.A); SLT=Salt Lake Tribune (UT) Note: Correlation coefficients appear in the cells. The notation “.oa.” means that the statistical significance was less than .10.
newspaper coverage correlated with it although the magnitude of this relationship varies across outlets (due in part to the Post’s early skewness in terms of the distribution of its stories, paralleling somewhat all of the local contests’ focus on the early contests and coverage of the incumbent president’s activities, particularly his trip to Japan and his State of the Union address). The Miami Herald and the Chicago Tribune are the only state newspapers associated with the ABC News, while only the Salt Lake Tribune is associated with the Wall Street Journal.
118 These comparisons are very simple ones. In this section I have assessed the amount of campaign coverage (measured in number of stories) the news media outlet presented on its front pages, and in the case of the ABC World News Tonight, during its
22 minute newscast. This is the bluntest of measures, thus, here we should find the greatest similarity. Yet we find there are substantial differences both among each group, national and local, and between them. A potential implication of this variation is that viewers’ and readers’ sense of the salience/immediacy of the campaign can vary depending on where they obtain their information.
Looking at the number of campaign stories provides us with limited insight into the diversity of news coverage. Moving to a more inclusive yet discrete measure of media coverage may help us determine whether variation continues to exist when we use alternative measures of media coverage, and, if so, may potentially shed some light on the nature of the variation.
Using the story as the unit of analysis is a blunt tool for dissecting news. While it may be used for comparison, it tends to overestimate an outlet’s level of coverage.
Column inches is a better measure because it accounts for story size. But, it is a difficult measure to disaggregate into party or candidate coverage. A more detailed, and thus accurate, picture is painted when we look at the outlets’ coverage percentages when measured more precisely. In this case. Table 4.6 presents a comparison of the measures;
Word percentage is defined as the total number of words, including editorials.
119 Table 4.6: Comparison of Media Coverage Measures: Story Count Versus Word Count
Media Outlet Story % Word % Difference
New York Times 14.4 (2) 10.1 (4) +4.3
Washington Post 18.8 (1) 11.9 (3) +6.9
Wall Street Journal 1.2 (13) 3.1 (12) -1.9
ABC News 10.9 (8) 12.6 (2) -1.7
Des Moines Register 10.1 (H) 6.9 (10) +3.2
Manchester Union Leader 14.1 (3) 9.4 (6) +4.7
Bangor Daily News 10.5 (10) 6.9 (10) +3.6
Atlanta Journal Constitution 12.2 (6) 8.1 (8) +4.1
Salt Lake Tribune 12.0 (7) 7.1 (9) +4.9
Miami Herald 10.8 (9) 8.7 (7) +2.1
Chicago Tribune 13.1 (4) 13.8 (1) -.70
Milwaukee Sentinel 9.0 (12) 6.5 (11) +2.5
Newport News Daily Press 12.6 (5) 9.5 (5) +3.1
Note: Storvoercentace ts defined as the total number of stcmes. izKludinft editorials. aoDearinaoa the outlet’s front Da&e relative to the total number offnnt Dace stories. Word Dercenlage is defined as the total number of words, including editorials. aDOearing on the front oace relative to the total word coverage (an estimation of the total number of words generated on the Soot page for the period examined) for the penod examined. For the ABC WorUiSewsTonight the percentages are based on the number of news s^nacnts which are diflermtiatrri by change from one reporter to another or to the anchor introdicing another topic. The cell entries represent the aggregate coverage given to the campaign from December 1 .1991 through June 3,1992.
appearing on the front page relative to the total word coverage for the period examined.
Total word coverage is defined as the estimated number of words (based on a front page average) generated on that newspapers front page for the time period examined. Story percentage is defined as the total number of stories, including editorials, appearing on the outlet’s front page relative to the total number of front page stories for the time period examined. The table also shows the difference in estimation of coverage when the two
1 2 0 measurements are compared. As we can see the number of stories measure overestimates the coverage amount and the rankings of each outlet in general. If we attempted to use stories as the candidate coverage measurement unit, this phenomenon would be even more amplified. For example, during the period of December 1, 1991 to February 18, the
New York Times ran stories featuring Clinton in the headlines five times to one relative t
Tsongas. When we actually measure the word content the difference was much less;
Clinton had a 1.5 to 1 lead on Tsongas. 1 should note, however, that a number of outlets had their campaign coverage levels underestimated by the word measure. The Wall Street
Journal, the ABC World News Tonight, and the Chicago Tribune had greater coverage levels with the word-based than the story-based measurement. This is due to these outlets running longer stories or segments rather than more short campaign pieces relative to their other front page coverage.
4.3 Coverage of the Candidates: Amount
4.3.1 National Media Outlets
To access candidate coverage, I utilized a measure of how much coverage (in meaningful groups of words) the candidate received weekly and used these values to create a summary indicator of average candidate coverage for the entire time period investigated. As shown in Table 4.7, the entries tell us on average how much coverage the candidate received relative to all the other news the media outlet printed or broadcast that week. A value of 2.0 would indicate that the candidate received, on average, two percent
121 of the outlet’s weekly coverage. The numbers in parentheses are the rankings of the candidate in terms of average weekly coverage.
In 1992, Clinton and Bush were the foci of a great deal of coverage from the beginning to the end of the process. There is little divergence among news media outlets as to their top ranked coverage recipients, as is expected. This parallels findings in other studies that frontrunners and incumbents enjoy larger amounts of coverage from the news media. Nor is there variance with regard to several also-rans, Douglas Wilder (then sitting governor of Virginia), Tom Harkin, Bob Kerrey (both sitting Senators) or Pat Buchanan in terras of their overall rankings. While their actual average coverage percentages vary somewhat, their rankings are identical across the national media outlets examined in this study. On the other hand, comparatively large differences exist for coverage levels of Jerry
Brown, Paul Tsongas and Ross Perot.
Relative to the other contenders, Ross Perot garnered more coverage from the
Wall Street Journal (probably due to this connections to the business world and Wall
Street ). The New York Times devoted more space on average per week to Brown and
Perot than Paul Tsongas. The Washington Post paid more attention to Tsongas than
Brown and Perot, respectively. ABC's coverage of Brown outranked their coverage of
Tsongas and Perot, but this was driven to a certain extent by the combative exchange between the candidate and ABC regarding a story on Brown’s alleged tolerance of marijuana and other drug use in the Governor’s mansion while he occupied it as
California’s chief executive.
1 2 2 Thus, using this summary measure of candidate coverage, I find that it is the firontrunners who are covered by the national media to a similar degree with regard to amount of coverage; it is the also-rans/early retirees that are covered similarly in that they share the award for low levels of coverage; And the real variation exists for the potential challengers who manage to hang on for a while.
Table 4.7: National News Média s Candidate Attention: Average Candidate Weekly National Coverage Amounts and Rankings for December 1991 to June 3, 1992
New York Times Washington Wall Street ABC World Post Journal News Tonight
Brown .97(3) .79(4) 33(5) 1.5 0 )
Clinton Z0(2) 2.5(2) 1.1 (2) 2.9(2)
Harkin .12(8) .25(8) .09(8) .29(8)
Kerrey .16(7) ■29(7) .10(7) 32(7)
Tsongas .89(5) ■92 P ) .27(4) .98(4)
Wilder .06(9) .05(9) .00(9) .02(9)
Buchanan 33(6) .61 (6) .17(6) .56(6)
Bush 4.4(1) 6.2(1) 23(1) 5.1 (1)
Perot .96(4) .70(5) 31(3) .75 (5)
Total 9.89 1231 4.57 IZ42
Note: The ceD entries tell us on average how much coverage the candidate received relative to all the other news the media outlet printed or broadcast that week. A value of 2.0 would indicate that the candidate received, on average, two percent of the outlet’s weekly coverage. The numbers in parentheses arc the rankings of the candidate in terms of average weekly coverage.
123 Table 4.8: Local News Media’s Candidate Attention: Average Candidate Weekly Local Coverage Amounts and Rankings for December 1991 to June 3,1992
DMR MUL BDN AJC SLT MH CT MS NNDP
Brown .31(7) .43(6) .52(5) .58(5) .35 (3) .26(6) .62(6) .42(5) .33 (4)
Clinton 1.3(3) 1.8(2) .96 (20 2.0(2) .62 (2) 1.5(2) 3.0(2) 1.5(2) 1.6(2)
Harkin 1.8(2) .16(8) .11(8) .11(8) .12(8) .19(7) .28(8) .16(8) .14(8)
Kerrey .48(4) .44(5) .36(6) .37(7) .16(7) .05 (8) .36(7) .17(7) .25(7)
Tsongas .46(5) .52 (4) .83 (3) .69(4) .29(4) .69(3) 1.1(3) .60(4) .29(6)
Wüder .01 (9) .04 (9) .00 (9) .02(9) .00(9) .01 (9) .00 (9) .00(9) .33 (4)
Buchan. .25 (8) 1.7(3) .35(7) .51 (6) .17(6) .57 (4) .92(4) .61 (3) .32 (5)
Bush 1.9(1) 4.6(1) 4.8(1) 4.1(1) 5.4 (1) 7.8(1) 5.9(1) 2.8(1) 4.8(1)
Perot .39(6) .32(7) .60 (4) 1.1(3) .28(5) .41(5) .63(5) .33 (6) 1.0 (3)
Total 6.90 5.88 8.53 9.48 7.39 11.48 12.81 6.57 9.06
Key; AJC=Atlanta Journal Constitntion (GA); BDN-Bangor Daily News (ME); CT=Chicago Tribnne (IL); DMR=Des Moines Register (lA); MLJL=Manche3ter Union Leader (NH); MH=Miami Herald (FL); MS=Milwaukee Sentinel (WI); NNDP=Newport News Daily Press (VA); SLT=Sall Lake Tribune (UT). Note: The cell entries tell us on average how much coverage the candidate received relative to all the other news the media outlet priided or broadcast that week. A value of 2.0 would indicate that the candidate received, on average, two percent of the outlet's weekly coverage. The numbers in parentheses are the rankings o f the candidate in terms of average weekly coverage.
4.3.2 Non-National Media Outlets
Table 4.8 displays the average weekly coverage for the candidates and their rankings for the local outlets. As with the national outlets. Bush ranked number one in terms of coverage levels receiving generally two to four times as much coverage as any other candidate (except in the case of the Des Moines Register which devoted 1.9 percent on average to Bush, 1.8 percent to Harkin, and 1.3 percent to Clinton). Harldn’s favorite son position in Iowa altered the patterns of coverage for the Des Moines Register relative
124 to the other local outlets. Wilder, who received very little coverage from most media
outlets before dropping out of contention, did manage to rank number four in his home
state’s Daily Press.
Clinton ranked second for almost all the local newspapers. Beyond the first and
second rankings, however, the local presses vary more; For example, Buchanan appeared
to gamer more coverage in locales in which his contest with Bush was highly contentious
such as New Hampshire and Illinois, or where his policy stances seemed to resonate or
alienate. Perot was a popular story in Georgia and Virginia where his organization at the
“grass roots” received much attention. The Atlanta Constitution featured coverage of
Perot during the early stages of the campaign in marked difference to the other outlets
who did not turn their attention to Perot until late March or early April. Jerry Brown
enjoyed his greatest coverage level relative to his peers from Utah’s Salt Lake Tribune.
Virginia’s Daily Press coverage also put Brown above a number of other candidates. Part
of this is explained by contest timing. The Virginia caucuses came rather late in the
process, around mid-April. Brown was Clinton’s only competitor, and he was also getting a good deal of attention for taking on the role of “spoiler.” As to Utah, Brown was known there due to his regional connection as well as his past success at the polls there.
Candidate coverage by the local outlets is more varied. This may be due to their distinct characteristics of geographic location, political culture, and position relative to the timing of political events.
125 4.3.3 National and Non-National Outlets Compared to The New York Times
Another means of exploring variation in media coverage is to designate a base and determine the degree of difference that exists between the base and other cases. In Tables
4.9 and 4.10, the New York Times, one of the often used indicators of national news media coverage, is used as the foundation for comparison. In this way, a general pattern of under- or over-coverage is made more visible.
As the data suggest, relative to the New York Times, the Wall Street Journal focused on the candidates much less relative to its other content.® This is particularly true of the coverage of incumbent President Bush. The Wall Street Journal was similar, however, to the New York Times in their coverage of Harkin, Kerrey and Wilder. There are differences between coverage levels of Perot, Brown and Tsongas. As to the other national outlets, compared to the New York Times, both the ABC News and the
Washington Post covered the candidates more, with the notable exceptions of Wilder and
Perot, and Brown for the Washington Post. Both also focused on Buchanan substantially more than the New York Times. ABC utilized coverage themes of voter dissatisfaction with the frontrunning candidates of both parties, and thus covered their opponents support levels and messages somewhat more. Perot may actually have been covered less due to his own disgruntlement with
® I should note that the average front page coverage for the Wall Street Journal is an estimated 4,108 words. This is closer in size to an average ABC newscast (3,991 words) than the next highest word count for a newspaper which is the New York Times with an average front page word count of 2,376. So there may be instances when the WSJ gives more coverage in words to a candidate on a particular day, but relative to its other content, the percentage may be smaller. However, in our effort to make comparability possible between these outlets, the relative measure is used. Relative attention is thought theoretically superior as well since the consumer is presented with the entire fr-ont page rather than just a portion of it
126 Table 4.9: National Coverage Differences Relative to the New York Times
Wall Street Journal Washington P#»st ABC Worltl News
Brown -.74 -.18 +.53
Clinton -.90 +.50 +.90
Haddn -.03 +.13 +.17
Kerrey -.06 +.13 + 16
Tsongas -.62 +.03 +.09
Wilder -.06 -.01 -.02
Buchanan -.16 +.28 +23
Bush -2.10 +1.80 + 70
Perot -.65 -.26 -21
Note: Cell entries are the percentage dificrences in eandidate average weekly front page coverage compared to the New Yoric Times. Negative values mean that the media outlet covered the candidate less on average during the week, and positive values mean that they covered the candidate more relative to the New York Times.
ABC and its news staff (Rosensteil 1994).
There does appear to be a difference in the amount of coverage the candidates receive when we use the New York Times as a baseline. The national news outlets appear to be rather consistent in whom they cover although there are slight variations.
The local press varies more with regard to how much space they allocate to a candidate. The Miami Herald and Chicago Tribune allocated a substantial amount of space to the campaign. Each resides in states that were competitive during the 1992
Democratic campaign. And each has a reputation for extensive campaign reporting, particularly in the area of political investigative reporting.
127 Table 4.10: Non-National Coverage DifTerences Relative to the New York Times
DMR MUL BDN AJC SLT MH CT MS DP
Brown -.66 -.54 -.45 -.39 -.62 -.11 -.35 -.55 -.64
Clinton -.70 -.02 -1.04 nodiff. -1.35 -.50 +1.00 -.50 -.40
Haiidn +1.70 +.04 -.01 -.01 nodilOf. +.07 +.16 +.04 +.02
Kerrey +.32 +.28 +.20 +.21 nodifF. -.11 +.20 +.10 +.09
Tsongas -.43 -.37 -.06 -.20 -.60 -.20 +.79 -.29 -.60
Wilder -.05 -.02 -.06 -.04 -.06 -.05 -.06 -.05 +.27
Buchan. -.08 +1.40 +.02 +.18 -.16 +.24 +.59 .28 -.01
Bush -2.50 +.20 +.40 -.02 +1.00 +3.40 +1.50 -1.60 +.40
Perot -.57 -.64 -.36 +.14 -.68 -.55 -.33 -.63 .04
Note: Cell entries are the percentage differences in candidate average weekly front page coverage compared to the New York Times. Negative values mean that the media outlet covered the candidate less on average dining the week, and positive values mean that they covered the candidate more relative to the New York Times Key: AJC=Atlanta Journal Constitution (GA); BDN-Bangor Daily News (ME); CT=Chicago Tribune (IL); DMR-Des Moines Register (lAX MUL=Mancheater Union Leader (NH); MH-Miami Herald (FL); MS=Milwankee Sentinel (WI); NNDP=Newport News Daily Press (VA); SLT-Salt Lake Tribune (UT)
Some of the difiFerences we see in candidate coverage weekly averages are due to the context and timing of specific state contests. This creates a différent setting
(temporally, geographically, and competitively) in which the local media outlet must report the campaigns. There are times when the national and local outlets (their own uniqueness set aside) do treat the candidates differently, and there are instances when there is no difference (for example, when the campaign does not make the fi’ont page or when everyone reports the New Hampshire results).
These differences and similarities are due in part to several factors: First, the national news media cover the campaign generally in a far more routine and universal
128 manner. Second, the non-national news media, while paying attention to the candidates during particularly important time periods (early contests, debates, delegate laden events) tend to be less universal and more local in their orientation generally. Will Molineaux, an editor at the Newport News Daily Press suggests, “often we don’t start paying attention to the campaign ourselves until it gets closer to our turn to go. And sometimes there are some heated state elections that we really need to cover. And that’ll take precedence until a couple or so weeks before the primary or caucus.” (Interview, November 1995). Third, candidate behavior varies. Candidates generate some difierentiation between national and non-national outlets by their own varied activity in the local areas and their attachment by means of certain characteristics to particular locales (home state status, regional attachment, ideological attachment, policy attachment and so on.)
4.4 Coverage of the Candidates: Direction
One important dimension of media coverage is the “direction” of the coverage.
During the 1992 campaign many people were surprised that after the barrage of damaging stories, Clinton was able to win the nomination. But just how much negative coverage did
Clinton receive? What is the nature of negative and positive coverage in general? Who receives the most and least negative and positive coverage? When and why do these candidates receive negative or positive coverage? Do the national and local outlets vary in their distributions of negative and positive coverage to the candidates?
129 4.4.1 National and Non-National Direction of Coverage
Conventional wisdom suggests that the national media tend to critique or evaluate
the campaign and its participants more than the local press. For this reason, one might
expect greater levels of directional coverage to be present in the national media’s stories.
Tables 4.11 and 4.12 give the totals of negative and positive coverage relative to
the candidates’ total coverage for both the national and non-national media outlets. These
values are indicative of the total positive or negative coverage a candidate received
relative to his own coverage. For example, the New York Times’s coverage of Jerry
Brown for the time period examined was distributed as 17.6 percent negative, 7.7 percent
positive and 74.7 percent neutral. We know from Table 4.7 that Jerry Brown received
close to 1 percent of all front page coverage for the time period examined, ranking third among all candidates in coverage allocated. The neutral percentages are not reported in the table.
One obvious general observation is that in 1992 the negative certainly outweighed the positive. Of the candidates, only Tsongas and Perot seemed to enjoy a relative outweighing of positive to negative, and state coverage was far more positive for Perot than his national coverage.
130 Table 4.11: Directional Coverage of Democratic Candidates: Negative and Positive Coverage Percentages
Brown Clinton Harkin Kerrey Tsongas Wilder
(-) (+) (-) (+) (-) (+) (-) (+ ) (-) (+) (-) (+)
NYT 17.6 7.7 20.4 11.1 23,1 3.7 163 2.9 10.6 7.6 791 0.0
WP 40.0 8.8 m & m 9.6 24.2 112 10.0 15.4 14.3 17.8 90.0 0.0
WSJ IZ2 9.3 17.4 12.7 14.0 16.0 24.0 26.0 5.9 14.4 31,0 0.0
ABC 27.1 3.2 18.5 5.0 1 0 2 3.8 8.6 4.2 14,0 7.6 29.0 0.0
Average 242 7.3 21.6 9.6 17.9 8.7 14.8 12.1 112 11.9 57.2 0.0
DMR 15.0 10.0 17.0 19.0 20.0 5.4 19.0 11.0 9.0 15.0 81.0 0.0
MUL 13.0 2.8 23.0 15.4 31.0 0.0 222 3.0 8.0 25.0 45.0 0.0
BDN 8.0 19.0 10.0 18.5 28.0 3.0 36.0 7.0 6.5 13.0 00 0.0
AJC IS 4 5.6 13.9 16.S 7.0 16.2 9.3 12.8 11.1 0.0 0.0
SLT 50 17.0 15.0 25.0 16.0 6.0 33.0 9.0 16.0 20.0 0.0 0.0
MH 46,0 6.9 13.0 17.1 18.3 2.8 11.1 0.0 23.1 9.8 98.0 0.0
CT 7.0 21.0 16.0 19.0 65.0 0.0 22,0 16.0 11.0 15.0 0.0 0.0
MS 10.0 3.5 24.0 16.3 42.1 0.0 12.0 3.5 16.0 10.3 54.0 0.0
NNDP 17.4 3.1 11.4 14.2 10.4 14.0 4.0 11.0 22.2 453 0.0
Average 16,0 10.0 17.3 28.0 3.8 20.6 7.0 14.0 15.7 36.0 0.0
Note: Cell entries arc the percentages of positive and negative coverage the candidate received relative to his total coverage in words for the time period examined. In other words, if Clinton received a total of 1,000 words of coverage and 400 ofthese were negative and 200 werepositive, the result would be 40 % negative, 20% negative and 40% neutral Neutral percentages are not shown. Key: AJC=Atlanta Journal Constitution (GA); BDN=Bangor Daily News (ME); CT=Chicago Tribune (IL); DMR=Des Moines Register (LAX MUL=Manehester Union Leader (NH); MH=Miami Herald (FL); MS=Milwaukee Sentinel (WI); NNDP=Ncwport News Daily Press (VA); SLT=Salt Lake Tribune (UT)
131 Table 4.12: Directional Coverage of Republican and Independent Candidates: Negative and Positive Coverage Percentages
Buchanan Bush Perot
(-) (+ ) (-) (+ ) (-) (+)
NYT I6i> 0.0 17.2 9.1 120 9.4
WP 8.0 11.0 24.0 2.1 10.0 9.8
WSJ 12.0 8.0 16.0 2.8 4.4 4.0
ABC 14.2 5.0 15.0 1.8 7.1 4.0
Average 125 6.0 18.1 4.0 8.4 7.0
DMR 17.0 4.2 16.0 4.2 0.0 15.1
MUL 9.0 18.2 30.0 6.0 9.1 7.0
BDN 27.0 14.0 9.0 4.0 9.0 5.0
AJC Î4.2 2.0 18.0 3.4 16.0 5.2
SLT 16.0 7.0 10.0 1.5 0.0 15.4
MH 4.0 8.8 10.2 2.2 6.0 11.8
CT 3 0 2.0 17.0 3.0 0.0 9.2
MS 10.0 6.2 8.0 5.1 0.0 20.0
NNDP 14.0 11.0 11.2 2.0 11.0 11.0
Average Î2.6 8.2 15.4 3.5 5.7 11.1
Note: Cell entries are the percentage of positive and negative coverage the candidate received relative to his total coverage in words for the time period examined. In other wonb, if Clinton received a total of 1,000 words of coverage and 400 o f these were negative and 200 were positive, the result would be 40% negative, 20% negative and 40% neutraL Neutral percentages are not shown. Key: AJC=Atlanta Journal Constitution (GA); BDN=Bangor Daily News (ME); CT=Chicago Tribune (IL); DMR=Des Moines Register (lAX MUL=Manchesfer Union Lender (NH); &IH=Miami Herald (FL); MS=Milwaulree Sentinel (WI): NNDP=Newport News Daily Press (V.A.); SLT=Salt Lake Tribune (UT)
The overall treatment of candidates can be assessed by looking at a comparison of their negative and positive treatment. These figures are presented in Table 4.13. Clearly there are differences between the national and non-national media. The candidates receiving the most negative treatment by the national media (relative to competitors) were
132 Brown, Bush and Clinton in that order. The state media, however, on average, treated
Harkin, Kerrey and Bush the most negatively. The national and local news outlets are more similar when we compare which candidates were treated better. The national media were generally more positive toward Tsongas, Perot and Kerrey, although Tsongas was the only candidate whose average negative to positive difference was positive. The local news media were kinder to Tsongas, Perot and Clinton. Tsongas and Perot both averaged higher positive than negative coverage, and Clinton was not far from having a balance of positive to negatives (-1.8 percent).
There are some further national vs. state differences: First, the national media coverage in this sample is slightly more negative in the aggregate that the state totals.
Second, there is variation between the national and state outlets as to which candidates received the greater level of negative coverage and somewhat less so, the positive coverage. While both types of outlets, on average gave Bush higher doses of negative coverage, the national news media covered Brown and Clinton more negatively while on average the local news media covered Kerrey and Harkin more negatively.
The state presses examined covered particular candidates far more negatively than others. This is most often a function again of timing of a state contest and other state- specific characteristics. Regional variation in negative and positive coverage seems apparent. Theoretically, if a candidate is familiar to a region, it is likely that the local news media will print and air stories about him. Often, however, these stories are a mix of original and wire stories. In Harkin's and Kerrey’s cases, a number of Midwestern and western state newspapers followed their campaigns closely. Unfortunately, this resulted in
133 some negative coverage since the candidates’ campaigns were not faring well. While local stories did not present Harkin’s candidacy in rose-colored terms, they were more likely to balance negative with positive. Yet wire stories which focused generally more on the horserace and primary results, were more likely to report poor showings and the negative implications of these poor finishes. Moreover, local news media can reflect the interaction of when they start to pay attention to the campaign and when a candidate is doing well or faltering. For example, while Tsongas generally received more positive than negative coverage, the local outlets that gave him more negative than positive coverage came in the middle and later portions of the campaign: the Atlanta Journal Constitution, the Miami
Herald, the Milwaukee Sentinel and Virginia’s Daily Press. Brown generally received more negative than positive coverage, however, in the outlets representing Maine, Utah and Illinois his positive coverage outdistanced his negative coverage.
What is not evident fi’om these aggregated over time positive and negative coverage percentages, however, is that most of the negative coverage for the peripheral candidates such as Wilder, Harkin, and Kerrey was generated after or just prior to their dropping out of the race. Therefore, this negative coverage could not be responsible for their withdrawing. Rather their general lack of media coverage is a potentially more acceptable explanation. For the candidates who are in the thick of the campaign with some chance of contesting the nomination, the negative coverage is far more serious and the positive coverage far more important.
Candidates who are actually competitive in a state and in the top tier of challengers are more likely to receive high levels of positive coverage. Of note is that the media, both
134 national and state, are more copious in their negative assessments than in their positive ones. This is partially explained by the fact that in multicandidate contests, more candidates will generally do worse than do better. Average negative coverage levels were actually lower in the Republican race, dampened by higher levels of neutral coverage generated by a great deal of issue-related news about Bush.
Table 4.13: The Directional Coverage Balance: The Difference between Positive and Negative Coverage Percentages
Brown Clinton HarUn Kerrqr Tsongas Buchanan Bush Perot
NYT -9.9 -9 3 -19.4 -13.6 -3.0 -16 -8.1 -Z6
W -31.2 -20.4 -13 +5.4 +3.5 +3.0 -21.9 -0.2
WSJ -2.9 -4.7 +Z0 +Z0 +8.5 -4.0 -13.2 -.04
ABC -23.9 -1Z5 -6.4 -4.4 -6.4 -9Z -13.2 -3.1
Nad Avr -16.97 -11.72 -9.2 -Z65 +&6S -6.55 -14.1 -1.6
DMR -5.0 +Z0 -14.6 -8.0 +6.0 -13.2 -11.8 +15.0
MUL -10.2 -7.6 -31.0 -19.2 +17.0 +9.2 -24.0 -Z1
BDN +11.0 +8.5 -25.0 -29.0 +6.5 -13.0 -5.0 -1.0
AJC -12.8 -3.9 -9.8 -6.9 -1.7 -1Z2 -14.6 -9.8
SLT +1Z0 +10.0 TOO -24.0 +4.0 -9.0 -8.5 +15.4
MH -39.1 +4.1 -15.5 -ll.I -13J +4.8 -8.0 +5.8
CT +14.0 +3.0 -65.0 -6.0 +4.0 -1.0 -14.0 +9.2
MS -6.5 -7.7 -4Z1 -8.2 -5.7 -3.8 -Z9 +20.0
NNDP -14J -24.6 -3.8 -lO.O +11.2 -3.0 -9.2 No diff.
Loc Avg -5.65 -1.8 -24.1 -13.6 +7.0 -4.6 -10.9 +&5
Key: AJC-Atlanta Joomal Constitution (GA); BDN^Bangor Daily News (ME); CT=Chicago Tribune (IL); DMR-Dcs Moines Register (lA); MUL=Manchester Union Leader (NH); MH=Miami Herald (FL); MS=Milwaukee Sentinel (WI); NNDP=Newport News Daily Press (V Ay. SLT-Salt Lake Tribune (UT)
135 There were diflFerences in the nature of the parties’ directional coverage. The
Republican campaign coverage was far more issue-oriented in general with much of
Bush’s economic and general domestic policy issue coverage being negative. There was not as much negative campaign and horse race coverage after the months of December,
January and February due to the fact that Bush’s viability was never really in question. The bulk of Bush’s positive coverage came later in the campaign when he recovered fi-om
Buchanan’s challenge and began to win primaries by substantial margins. The Democratic candidates’ coverage, both negative and positive, fell in the area of campaign and horserace coverage.
4.5 Cases of News Media Variation
One problem with assessing multiple media outlets is that information is lost when condensing the data into some presentable form. While aggregate comparisons are helpful when we are looking for general distinctions, there is still a need to focus on single points of time and data for comparison to clarify finely grained differences in media coverage.
We shall look at two examples.
4.5.1 Paul Tsongas: Coverage of a Candidate’s Defeat
Paul Tsongas was covered by the various news outlets in a varied manner. Topics as diverse as his health, his governmental experience, his ethnicity, his character, his nerdishness, his come-ffom-nowhere success in New Hampshire, his pro-business approach, his electability, and his stance on nuclear power filled the news. But these
136 topics did not appear in all news outlets, and when they did, the timing, focus, and
attention varied.
In Florida, Tsongas’ campaign confronted tragedy as one of the secret service
persons assigned to woric for the candidate was killed in an accident while on duty. This
event was not reported very widely on the front pages of either the national or the non-
national media outlets, but it did make the papers in Florida where the event occurred.
Of Super Tuesday’s many contents, Tsongas had chosen Florida as his test against
Clinton. Unfortunately, the death of the secret serviceperson was reputed to cast a pall
over his campaign. Tsongas’ poor showing in Florida was thought to have been
connected with the event and the subsequent reporting of the tragedy. Moreover,
Tsongas canceled several campaign visits in order to console the widow and attend the
funeral. The overall results of Super Tuesday and the later problems in Illinois and
Michigan resulted in Tsongas suspending his campaign.
While his Florida tragedy did not make the front pages around the nation, his
suspension did. One way in which we can assess the differences among news outlets is to
closely examine how they cover a particular event such as a “suspension” that has real
consequences for the rest of the campaign.
Table 4.14 documents the media’s coverage of the Tsongas’ suspension. He
suspended his campaign on March 19, 1992. The ABC World News Tonight broke the
story that night devoting over 7 percent of that night’s coverage to Tsongas and leading
off the news with the event. The story focused on Tsongas’ lack of money presenting it as the main reason for his dropping out. It also discussed Tsongas’ declining momentum after
137 Super Tuesday and the losses in Michigan and Illinois as well as the role that Tsongas’ support might play in the upcoming contests. It is no surprise that of the non-national papers, the Chicago Tribung, the Miami Herald and the Manchester Union Leader devoted the most attention to the suspension. Illinois’ contest had just occurred, Florida’s tumultuous contest had been about a week prior, and Tsongas had gotten his start in the
New Hampshire primary and had been endorsed by the paper. All of these outlets were in
Table 4.14: National and Non-National News Coverage of the Tsongas Suspension (March 19 and 20)
Source or Story Percentage that Day’s Content Front page Coverage Outlet
NYT Staff 9.6 Message, competition, resources
WP Staff 15.0 Resources, strategy, character, support, campaign in general
WSJ Staff 3.0 Resources, competition
Staff 7.2 Resources, campaign in gencraL outcome, support, momentum
DMR Wire Service 2.4 General reference to withdrawal
MUL Associated Press 9.0 Competition, strategy, resources
BDN Associated Press 5.0 Message, competition, resources, campaign in general
;\JC Staff 1.0 Resources
SLT New York Times News Service 0.5 Suspension of campaign
MH Staff 10.0 Message, competition, resources
CT Staff 10.0 Resources, viability, campaign in general
MS Wire Service 7.0 Background, competition, resources
NNDP Los Angeles Times News Service 3.0 Competition, campaign in general
Key: AJC=Atlanta Journal Constitution (GAX BDN=Bangor Daily News (ME); CTOticago Tribune (IL); DMR=Des Moines Register (lA). MUL=Manohester Union Leader (NH); MH=Miami Herald (FL); MS=VClwaukee Sentinel (WT); NNDP^Newport News Daily Press (VA); SLT=Salt Lake Tribut» (UT)
138 some way “close” to Tsongas’ campaign due to the circumstances of his competition in their state. Moreover, these three papers are known for their political reporting, or in the case of the Union Leader, their politics.
There are some general differences to be found between the national and non national press. The national press, excluding the Wall Street Journal, devoted, on average, more time and space to the suspension, thus giving its audience more coverage on the details and potential implications of the Tsongas’ suspension. The non-national outlets in certain cases did cover the event, but discussed it in far less detail. Only three of the nine non-national presses used their own staff to write the story. Generally these instances were from newspapers in states whose contests had recently occurred.
4.5.2 Brown: Victory Coverage Across Outlets
Another way that news outlets can vary their coverage of an event is how they frame and evaluate the event. After Tsongas’ suspension. Brown was crowned the A-B-C candidate - Anyone But Clinton. Connecticut was Brown’s first head-to-head contest with Clinton on March 24, 1992 - one week after Illinois and Michigan and only a few days after Tsongas suspended. In a surprising outcome (since Clinton was considered the likely nominee) Brown won a close vote, with 38 percent to Clinton’s 35 percent of
Connecticut’s vote. Tsongas received 20 percent of the vote as his name remained on the ballot.
Table 4.15 illustrates the coverage the Brown victory in Connecticut. ABC reported the upset on the 24th, while the newspapers followed on the 25th. Again the
139 amount of front page coverage each outlet apportioned to Brown varied substantially with the Manchester Union Leader devoting the most coverage (13 percent of that day’s front page coverage) to Brown. The local outlets actually paid much more attention to Brown’s unanticipated and thus, “surprising” victory than they did to Tsongas’ anticipated suspension. The news media also framed their coverage differently. Some news outlets chose to focus on Brown, some focused on both Brown and Clinton equally, while many focused a greater portion of coverage on Clinton and the implications for his campaign.
Not only did the focus of the story vary, but the content of Brown’s coverage varied. Five news outlets quoted Brown substantially regarding the meaning of the win.
This is categorized as candidate “spin.” Most of the stories discussed the contest in terms of Brown versus Clinton and the new competitive implications. For example, what would this mean to the rest of the campaign? How would it effect Clinton’s transition to the general election and estimates of his electability? Only two news outlets talked about
Brown’s viability and the effect of the win on his resources and future expectations.
Another difference is found in the way the titles are used to frame the event. The New
York Times and the Milwaukee Sentinel framed the event in terms of voter discontent.
Other news outlets focused on the surprise loss of Clinton.
140 Table 4.15: National and Non-National Coverage of Brown After His Connecticut primary Victory (March 24/25)
Brown’s Frmcipal Focus Title Content of Brown’s Percentage (Candidate) Coverage
NYT 9.0 Clinton Democrats rebuff Clinton in the resources, support, Connecticut Victory campaign
WP 2.0 Clinton and Brown Clinton is forced to shift strategy. competition, outcome Brown defeats Clinton in Connecticut upset
WSJ 1.0 Clinton What’s News (No Title) competition, outcome
ABC 5.0 Clinton and Brown No title strategy, viability, support, campaign
DMR 5.0 Clinton Brown shocks Clinton (AP) spin, competition, momentum, outcome
MUL 13.0 Brown Clinton Defends Himself Against competition, spin Brown Jabs (wire)
BDN 8.0 Clinton and Brown Brown upsets Clinton (AP) spin, competition, momentum, resources, viability, expectations, outcome
AJC 9.0 Brown It’s not over for Democrats; Brown message, competition upsets Clinton to Win Connecticut
SLT 7.0 Brown Brown stuns Clinton in Connecticut background, (Baltimore Sun) competition, outcome
MH 2.0 Clinton Brown Stuns Clinton in Connecticut competition, outcome vote
CT 6.0 Clinton Clinton defends himself against competition, strategy Brown jabs (wire)
MS 5.0 Clinton Brown pulls upset in Coimecticut; spin, competition, Exit poll shows dissatisfaction with outcome choices (wire)
NNDP 8.0 Clinton Brown victory confounds Clinton spin, competition, (wire) outcome
Key: AJC=Atlama Journal Constitution (GA); BDN=Bangor Daily News (ME); CT'=Chicago Tribune (IL); DMR=Des Moines Register (IA>. MUL-Mancbester Union Leader (NH); MH=Miami Herald (FL); MS-Milwsukee Sentinel (WT); NNDP=Newpott News Daily Press (VA), SLT=Sah Lake Tribune (UT).
141 4.6 Coverage of Candidates: Content
The amount of coverage given to candidates can vary. The amount of positive and
negative coverage can vary as well. But what of the substance of the coverage? Is there
anything systematic in the way that the news media decides what to include in their
coverage?
I have aggregated the multitude of coverage categories coded into five types of
coverage: 1. Policy coverage (covering race, labor, health care, abortion, the economy and jobs, the budget, taxes, education, the environment, crime, drugs, the death penalty,
school prayer, AIDS policy, trade policy, international affairs, defense, urban problems,
space, social security and welfare, general domestic policy and family values); 2.
Candidate substance coverage (covering experience and background, character and ethics,
ideology, message, personal health, spouse and family); 3. Campaign coverage (covering
strategy, organization, advertising, resources, support, outcomes of contests, general
campaign activity, debates, and candidate spin ); 4. Horserace coverage (including
coverage of momentum, viability and electability, expectations before and after events
(and in general), performance, and competitive positioning or movement.); and 5. Anti
opponent rhetoric (covering candidate quotes directed toward opposition candidates in the
nomination phase or toward anticipated opposition in the general election phase.
The results of comparing the amounts of coverage associated with these categories
suggest that there is less horserace coverage than we would have expected, more policy
coverage in the aggregate, and less candidate substance coverage.
142 LOG MEAN
NATL MEAN ^ k s S K ® 5 : . : 8 » v v ///M NNDP iiiiiiii MS ; x & i ; II» T y y / y y / « ™ CT ...... MH SLT ' ' ' K i n g z
AJC 1111 ll I I |{ BDN MUL DMR ABC
WSJ ^ '------isssssisass 1® #ï88888S 8888l//y/y® 11 WP 5»Îλ»»»»W / / H i i
20 40 60 80 100
Policy □ Candidate Campaign Horse race Anti-opponent
Figure 4.3: Combined Republican and Democratic Campaign Coverage by Outlet
143 LOG MEAN NATL MEAN ...... V i ioonnnnnnnnn / / / ixR NNDP ...... T- ...... MS ...... r ...... 1...... CT BBgaaaaagBflflaflaaaaaaBaay y y HawHM MH r- - ■■ 1 1 SLT A J C -j BDN MUL DMR ABC WSJ WP NYT
0 20 40 60 80 100
Policy [ j Candidate Campaign H orse race Anti-Opponent
Figure 4.4: Republican campaign - Percentage Campaign Coverage Devoted to Area by News Media Outlet
144 In the aggregate, the national media focus less on the horserace than the non-
national news media. (See Figure 4.3.) For the Republicans, policy coverage appears to
dominate coverage across outlets. This is driven by the coverage of the incumbent
president. Much of Bush’s campaign coverage revolved around economic and domestic
policy in general. In this instance, the news media utilized an issues framework to give
structure to the Republican campaign, rather than a horserace framework. This is an
important result given the conventional wisdom as well as previous empirical evidence that
suggests the horserace dominates campaign coverage. The framework the news media use
may be conditional rather than absolute.
As to content, the non-national press examined devoted less of their space to
information relating to candidate substance. The difference, however, is slight and generated by the high level of candidate substance coverage generated by the Washington
Post.^ And while in the anti-opponent rhetoric category variation exists, the differences do
not appear to be systematically generated by local versus national outlets.
When the content of the coverage is broken down by party, we see the differences
between the Republican contest and the Democratic campaign. Clearly if Americans knew anything about a candidate’s policy stance, it was likely to be known about president
Bush.
^The Washington Post may devote more candidate-oriented substance due to the “localness” of many of the candidates. Kerrey, Harkin, Bush ail served their respective constituents in Washington, D.C. Buchanan’s home and much of his activity were located in the District Moreover, since whomever won the nomination would be the potential occupant of Washington, D.C.’s most prominent home, the Post as a “local” paper would understandably cover the candidates more.
145 LOG MEAN
NNDP
MH YZZZZ/M
z s^ z z ^ z z
0 20 40 60 80 100
y Policy Q Candidate Campaign B Horse race | Anti-Opponent
Figure 4.5: Democratic Campaign - Percentage Front Page Coverage Devoted to Area by News Media Outlet
146 Figure 4.4 illustrates the high levels of policy coverage the Bush/Buchanan contest
encouraged. Perhaps this overabundant issue coverage could help explain his lowered support levels during this time period. While presidents do enjoy a certain deference by the media. Bush was also responsible for this unprecedented degree of policy coverage.
His need to refocus his image in the public as a leader who cared about domestic concerns led him into a policy-making craze beginning with the State of the Union. But rather than a general direction. Bush’s camp came up with detailed policies on Health Care, Tax
Reform, Education, Aid for Children, Medicare, and more. He altered his trip to a number of Pacific countries, including Japan, to focus on trade issues fi-amed by domestic concerns about jobs.
As to the Democratic campaign illustrated in Figure 4.5, the national press was generally far more likely to cover policy than the non-national (N=the national mean;
L=Iocal mean). Overall, however, policy coverage was sacrificed for campaign and horse race coverage. More disturbing is that when we move fi"om the fi’ont page to the inside pages, there is not the expected abundance of policy coverage. Rather we see a continuation of the fi’ont page stories, and if there is policy information, it is presented within the context of a general campaign story.
In 1992 individuals had more information available regarding candidate substance coverage, particularly on the Democratic side. Much of this was driven by the attention given to Clinton’s “character” issues. On average in 1992 the non-national news outlets paid much more attention to candidate substance, yet they were a bit uneven —covering
Clinton more and the other candidates less, while the national outlets focused on character
147 issues more evenly presenting ethics/character stories on Bush, Brown, Perot and Kerrey.
Disturbingly little substance coverage dealt with message or background/government service or ideology. This type of coverage was often candidate-driven as well. Harkin, for example, had more coverage pointing out his ideology. Brown had more message coverage.
The bulk of the Democratic coverage was campaign-oriented. For the
Republican’s incumbent versus challenger contest, the bulk was policy with some campaign. Given the greater number of competitors, it is not surprising that the
Democratic campaign was given substantially more horse race coverage. In addition, with more candidates competing against each there was far more anti-opponent rhetoric making its way on the front pages.
4.6.1 As the Nomination Campaign Stages Change so Does the Coverage
Campaigns are dynamic in nature. By its very association, the coverage of campaigns is as well. Appendix tables C-1 thru C-13 illustrate how coverage levels as well as content can change over time. Breaking the campaign down into four periods
(Early Days, Initial Contests, Mist Clearing and Pre-Convention/General Election
Maneuvering) we can see how the media’s focus on different aspects of the campaign can change. I will use one of these tables (Table 4.19) to illustrate how coverage changes over time. The ABC World News Tonight began its coverage of the Democratic campaign with most of its political coverage focusing on campaign activity (42.1%). However,
148 policy coverage was second with 23.3% of the coverage. This was the height of policy coverage relative to the other coverage categories.
After the Early Days, ABC focused its Democratic campaign coverage on campaign related activity, the horse race and candidate substance with the most of its coverage occurring in mid-February and early March. As to A B C 's coverage of the
Republican campaign, most of it was related to policy and issues, but campaign activity coverage did dominate during its coverage of the “Initial Contests” and took second place overall. In general, the news media outlets examined shared the dominant story frameworks of Democratic campaign activity and Republican policy with campaign activity sharing the focus during early and late portions of the campaign. Appendix B presents data from the other national and local news media outlets examined. These tables suggest that there is a rhythm to the campaign that is followed by the news media, both locally and nationally to a certain degree. Policy and candidate substance is more likely to be discussed early, but the proportions are not high for the Democratic campaign. When the campaigns were heated the news media shifted focus to campaign and horse race coverage. This can vary among the local news outlets since they may be more interested in regional contests or their own contest at times. But generally, higher levels of campaign activity do bring more campaign coverage. Early contests do so as well. Had Perot not been a feature in the 1992 campaign, the level of coverage during the latter portion of the campaign probably would have been much lower.
One other finding from the examination of these tables is that the local news media were more reliant on their own reporters when the campaign was closer to home. Smaller
149 newspapers relied on wire/news service reports when the campaign activity was distant.
Table 4.16: Substance of the Coverage forthtABC World News Tonight
Period I: December 1,1991 to February 18,1992 - Earlj- Days Average Number of words per day; 432 Party Democrats Republkaus Total Coverage in words: (34,622) 10377 (29.9«/o) 24,245 (70.1%) Area of Coverage Policy/Issue 2416 (233%) 15,513 (63.9%) Substance Re: Candidate 1853 (17.8%) 1426 (5.8%) Campaign 4371 (4Z1%) 5896 (243%) Horserace 907 (8.7%) 750 (3.1%) Anti-Opponent Rhetoric 830 (7.9%) 342 (1.4%) Misc. 0 (0.0%) 318 (13%)
Period 2: Februarv 19 to March 10,1992 - Initial Contests Average Number of words per day. 778 Party Democrats Republicans Total Coverage in Words: (16352) 9465 (57.8%) 5887 (42.1%) Area of Coverage Policy/Issue 866 (9.1%) 1214 (17.6%) Substance Re: Candidate 1009 (10.6%) 357 (53%) Campaign 5772 (60.9%) 3480 (50.5%) Horserace 1165 (12.3%) 1296 (18.8%) Anti-Opponent Rhetoric 653 (6.8%) 540 (7.8%)
Period 3: March 11 to April 7,1992 - Mist Clearing Average Number of words per day: 643 Party Democrats Republicans Independent Total Coverage in Words:(18,019) 14,330 (79.5%) 3689 (20.4%) 0 (0.0%) Area of Coverage Policy/Issue 2177 (153%) 1947 (52.7%) 0 (0.0%) Substance Re: Candidate 2317 (16.1%) 81 (2.1%) 0 (0.0%) Campaign 7061 (49.2%) 1408 (38.1%) 0 (0.0«/o) Horserace 890 (63%) 158 (4.3%) 0 (0.0%) Anti-Opponent Rhetoric 1885 (13.1%) 95 (2.5%) 0 (0.0%)
Period 4: April 8 to June 3,1992- Pre-Convention and General Election Maneuvering Average Number of words per day. 394 Party Democrats Republicans Independent Total Coverage in Words: (22,480) 9405 (41.8%) 8474 (37.6%) 4601 (20.5%) Area of Coverage Policy/Issue 1689 (17.9%) 6410 (75.6%) 494 (10.7%) Substance Re: Candidate 2710 (28.8%) 126 (1.4%) 1317 (28.6%) Campaign 3818 (40.5%) 1685 (19.8%) 2338 (50.8%) Horserace 412 (43%) 253 (2.9%) 340 (73%) Anti-Opponent Rhetoric 776 (83%) 0 (0.0%) 112 (2.4%)
150 Differing content focus is far more visible when evaluating Bush’s coverage even
after Buchanan was no longer a threat. Bush’s local coverage often focused on a policy
issue. For example, in Florida it was Haiti and Cuba, in Virginia it was defense monies,
submarines and shipyards, and in New Hampshire as well as in other states, it was the
economy. For the Democrats this is true as well but to a lesser degree. Context defines
debate. When the Democratic campaign is in Iowa the talk is about farmers; in Michigan
labor becomes the issue. And while the national press reports on these local debates, they
do not present them in the same manner or form. We will explore policy content more in
the next section.
4.6.2 Candidates and Their Issue Coverage
The content coverage given to the individual candidates in 1992 is illustrated in
Tables 4.20 through 4.32. The table entries display the percentage of issue coverage (and the actual number of words) the candidate received relative to a ll candidates, including the other party’s candidates. The categories the coverage fell under are listed. The top three categories (in terms of coverage volume) are also displayed as well as the time period
during which the candidate received the most policy coverage. I will not discuss each table in detail, but rather discuss some general conclusions drawn from the data about each of the candidates and in general.
151 4.17: New York Times Issue Coverage of Candidates
Candidate Percent issue Categories Top three categories Period coverage compared Covered in order o f coverage during with all candidates volume which candidate received most coverage
Brown 42% health policy, environment, urban taxes (flat tax) Period 3 (528) policy, domestic policy, social domestic policy welfare, taxes abortion
Clinton 2.8% domestic policy, health, economy, economy Period 1 (365) social welfare, taxes, urban policy, social welfare crime (death penalty), foreign urban policy/Los policy Angeles riots
Harkin 0.4% health policy, economy heahh policy Period 1 (51) economy
Kerrey 2.8% health policy, economy health policy Period 1 (372) economy
Tsongas 2.1 % health policy, economy, taxes, economy Period 1 (272) environment, social welfare, taxes education health policy
Buchanan 2.5 % social welfare, economy, defense social welfare Period 2 (324) defense economy
Bush 85 % race, abortion, domestic policy, foreign policy Period 1 (11,026) health policy, economy, budget, economy social welfare, taxes, environment, trade prayer in school, trade, defense, foreign policy, education, family values, environment
152 4.18: Washington Post Issue Coverage of Candidates
Candidate Percent issue Categories Top three categories Period coverage compared Covered in order of coverage during with all candidates volume which candidate received most coverage
Brown .88% taxes, environment, campaign finance Period 3 (101) campaign finance reform reform energy taxes environment
Clinton 3.5 % taxes, trade, economy, gay rights Period 4 (398) domestic policy, environment environment, labor, business domestic policy (in regulation, gay rights, urban general) policy
Harkin .07 % taxes taxes Period 1 (8)
Kerrey .07 % taxes taxes Period 1 (8)
Tsongas 1.8% taxes, education, trade, taxes Period 1 (207) energy, economy economy education
Buchanan .38 % budget foreign policy budget foreign policy Period 1 (44)
Bush 93.2 % foreign policy, defense, economy Period 1 (10,630) trade, economy, health trade policy, abortion, minority budget affairs, budget, taxes, environment business policy, domestic policy, energy, campaign finance reform, labor, social welfare, family values
153 4.19; Wall Street Journal Issue Coverage of Candidates
Candidate % Issue coverage Categories Top three categories Period compared with ail Covered in order of coverage during candidates volume which candidate received most coverage
Brown .68 % taxes, labor, trade labor, taxes, trade Period 3 (72)
Clinton 11.8% domestic policy, economy, business policy Period 4 (1.235) taxes, abortion, taxes, trade, foreign policy foreign policy, business, taxes education, environment, family values
Harkin .75 % domestic policy, taxes, economy Period 1 (79) economy, budget, foreign taxes policy budget
Kerrey 1.0% domestic policy, economy, taxes, economy Period 1 (105) taxes, abortion domestic and abortion policy (tie)
Tsongas 2.13% taxes, environment, abortion, taxes, business Period 2 (223) economy, business, trade economy
Buchanan 1.6% economy, taxes, foreign policy economy, foreign Period 1 (168) policy, taxes
Bush 82% AIDS, domestic policy, health, economy Period 1 (8.585) economy, budget, social foreign policy welfare, taxes, education, environment environment, legal system reform, trade, space, defense, foreign policy, family values, abortion, crime, business, science, urban
154 4.20: ABC World News Tonight Issue Coverage of Candidates
Candidate Percent issue Categories Top three categories Period during coverage covered in order of coverage which candidate compared with all volume received most candidates coverage
Brown 5.7 % domestic policy, health taxes, economy, Period 3 (1,689) care, economyyjobs, domestic policy taxes, environment, overall crime, foreign policy, social welfare polic>'
Clinton 9.97% abortion, domestic economy, education, Period 4 (2,934) policy, health care, environment economy, social welfare policy, taxes, environment, education, crime, defense, minority issues, foreign policy, family values, urban issues
Harkin 1.41% domestic policy, health economy, domestic Period 1 (414) care, economy, taxes, policy, education education, crime
Kerrey 1.54% domestic policy, health economy, health Period 1 (453) care, economy, care, domestic environment, crime policy
Tsongas 3.95% abortion, health care, economy, taxes, Period 1 (1,163) economy, (axes, social environment welfare policy, environment
Buchanan 2.5% economy, budget, taxes, minority issues, Period 2 (735) minority issues, social economy, taxes welfare policy
Bush 74.87% AIDS, domestic policy, economy, foreign Period 1 (22,020) health care, economy, policy, trade policy budget, taxes, education, environment, crime, trade, space, defense, foreign policy, energy, labor, social welfare policy, urban issues, family values
155 4.21: Des Moines Register Issue Coverage of Candidates
Candidate Percent issue Categories Top three categories Period coverage Covered in order of coverage during compared with all volume which candidates candidate received most coverage
Brown 1.7% economy, taxes economy Period 1 (105) taxes
Clinton 2.1% (axes, death penalty, death penalty Period 1 (131) agriculture, gun control taxes economy
Harkin 8.5% domestic policy, death agriculture Period 1 (516) penalty, health care, healthcare defense, taxes, economy, taxes elderly, energy, agriculture
Kerrey 3.0% domestic policy, trade, agriculture Period 2 (181) health care, taxes, economy agriculture health care
Tsongas 2.5% economy, taxes, energy, agriculture Period 2 (151) agriculture taxes economy
Buchanan 1.6% healthcare foreign policy Period 1 (95) foreign policy healthcare
Bush 80.5% defense, education, budget Period 1 (4,868) environment, domestic economy policy, health care, urban policy economy, budget, social welfare, taxes, trade, environment, urban policy, agriculture, family values
156 4.22; Manchester Union Leader bsue Coverage of Candidates
Candidate Percent issue Categories Top three categories Period coverage compared Covered in order of coverage during with all candidates volume which candidate received most coverage
Brown J29% abortion, taxes abortion Period 3 (10) taxes
Clinton 1.5% social welfare, abortion, social welfare, abortion Period 1 (51) taxes taxes
Harkin 12% education education Period 1 (40)
Kerrey 1.1% health, taxes health Period 1 (33) taxes
Tsongas 0 no issue coverage no substance coverage no issue coverage
Buchanan 7.8% domestic policy, health, taxes Period 2 (264) taxes. labor, economy economy labor
Bush 88% domestic policy, health, taxes Period 1 (3.004) economy, budget, taxes, economy social welfare, foreign healthcare policy, environment, urban policy
157 4.23: Bangor Daily News Issue Coverage of Candidates
Candidate Percent issue Categories Top three categories Period coverage compared Covered in order of coverage during with all candidates volume which candidate received most coverage
Brown J0% taxes, economy economy Periods (22) taxes
Clinton 13% abortion, economy, taxes, gun campaign finance Period 3 (93) control, campaign finance reform, economy reform gun control
Harkin 0 no issue coverage no issue coverage no issue coverage
Kerrey 0 no issue coverage no issue coverage no issue coverage
Tsongas 2.1% taxes, social welfare, economy taxes, economy Period 2 (152) social welfare
Buchanan .55 % budget, taxes budget Period 1 (152) taxes
Bush 95.8% agriculture, domestic policy, urban policy Period 1 (7.015) health care, social welfare, taxes taxes, trade, defense, foreign trade policy, economy, defense, urban policy, family values, abortion, environment, AIDS, campaign finance reform
158 4.24: Sait Lake Tribune Issue Coverage of Candidates
Candidate Percent issue Categories Top three categories Period coverage compared Covered in order o f coverage during with all candidates volume which candidate received most coverage
Brown 0 no issue coverage no issue coverage no issue coverage
Clinton 30% labor, taxes, urban labor Period 4 (49) urban taxes
Harkin .13% energy energy Period 2 (20)
Kerrey .04% energy energy' Period 2 (7)
Tsongas .41% energy, labor, economy energy Period 1 (66) economy labor
Buchanan .12% abortion abortion Period 2 (19)
Bush 99% homeless, trade, business, foreign policy Period 1 (15,810) domestic policy, health care, trade economy, budget, social taxes welfare, taxes, envirorunent. family values, defense, foreign policy, urban, education
159 4.25: Atlanta Journal Constitution Issue Coverage of Candidates
Candidate Percent issue Categories Top three categories Period coverage compared Covered in order of coverage during with all candiotes volume which candidate received most coverage
Brown 0 no issue coverage no issue coverage no issue coverage
Clinton .16% taxes taxes Period 1 (9)
Harkin 0 no issue coverage no issue coverage no issue coverage
Kerrey .75 % health, taxes health Period 1 (43) taxes
Tsongas .97 % economy, domestic policy economy Period I (55) domestic policy
Buchanan 2.7% foreign policy foreign policy Period 1 (154)
Bush 95.4% domestic policy, health care, foreign policy Period 1 (5.430) economy, budget, social taxes welfare, taxes, trade, AIDS, health care defense, foreign policy, environment, urban policy, education, family values
160 4.26: Miami Herald Issue Coverage of Candidates
Percent issue Categories Top three Period dining Candidate coverage Covered categories in order which compared with of coverage volume candidate all candidates received most coverage
Brown .13% environment environment Period 4 (7) Clinton 2.2% immigration, social welfare, foreign policy Period 4 (119) foreign policy, family values, gay immigration rights gay rights
Harkin 0 no issue coverage no issue coverage no issue coverage
Kcrrcy 0 no issue coverage no issue coverage no issue coverage
Tsongas 34% immigration immigration Period 2 (19) Buchanan .70% foreign policy foreign policy Period 2 (39)
Bush 96.6% foreign policy, immigration, foreign policy Period 1 (5335) health care, economy, budget, economy taxes, environment, drug policy, immigration trade, minority policy
161 4.27: Chicago Tribune Issue Coverage of Candidates
Candidate Percent issue Categories Top three categories Period during coverage Covered in order of coverage which candidate compared with volume received most all candidates coverage
Brown 0 no issue coverage no issue coverage no issue coverage
Clinton 2.8% labor, economy, taxes, labor, taxes Period 4 (222) urban policy urban policy
Harkin 1.1% labor, economy labor Period 1 (87) economy
Kerrey 0 no issue coverage no issue coverage no issue coverage
Tsongas 22% economy, taxes economy Period 2 (175) taxes Buchanan 1.8% immigration, trade, arts and immigration, arts and Period I (146) humanities, economy humanities, trade
Bush 92% health, economy, budget, economy Period 1 (7458) taxes, trade, defense, foreign policy foreign policy, arts and taxes humanities, urban policy, environment
162 4.28: Milwaukee Sentinel Issue Coverage of Candidates
Candidate Issue coverage Categories Top three categories Period compared with all Covered in order of coverage during candidates volume which candidate received most coverage
Brown 1.7% taxes, drug policy, defense taxes Period 3 (85) drugs defense
Clinton 7.0% economy, taxes, labor Period 4 (344) environment, labor, drug economy policy, defense taxes
Harkin .82% economy, taxes economy Period 1 (40) taxes
Kerrey .34 taxes taxes Period 1 (17)
Tsongas 1.1% economy, taxes economy Period 1 (55) taxes
Buchanan 0 no issue coverage no issue coverage no issue coverage
Bush 88.9% health, economy, budget, foreign policy Period 1 (4,355) taxes, trade, defense, foreign economy policy, social welfare, domestic policy domestic policy
163 4 J8: Newport News Daily Press Issue Coverage of Candidates
Candidate Percent issue Categories Top three categories Period coverage Covered in order of coverage during compared with ail volume which candidates candidate received most coverage
Brown 23% abortion, domestic policy, abortion Period 3 (20) foreign policy domestic policy foreign policy
Clinton .76% economy, taxes, foreign economy Period 1 (62) policy, domestic policy foreign policy taxes
Harkin 0 no issue coverage no issue coverage no issue coverage
Kcrrcy 2.58 % economy, foreign policy, foreign policy Period 1 (209) defense economy defense
Tsongas .42 % economy economy Period 3 (34)
Buchanan 23% the arts & humanities the arts and humanities Period 2 (19) Bush 95.7 % race, abortion, domestic trade Period 1 (7,746) policy, economy, budget, economy taxes, environment, trade, foreign policy defense, foreign policy, the arts and humanities, term limits, urban policy, labor policy, social welfare, AIDS, education
164 Brown received relatively small levels of policy coverage. Most of this coverage
was generated between Super Tuesday and New York. The prominent themes of Brown’s
coverage were taxes and the economy. The New York Times and the ABC News covered
the widest gamut of issues on Brown and gave him the highest volume of issue coverage -
4.2 and 5.7 percent of the issue coverage pie, respectively. State newspapers covered
Brown policy-wise to a lesser and more limited degree.
Kerrey and Harkin were treated rather similarly by the national media with regard
to issue coverage as they were considered to be fighting for a similar vote in the Midwest and western regions. Kerrey, on balance, however, received more coverage from the
national media, although the issues covered for both candidates were often the same.
Locally, there was greater diversity in terms of how much issue coverage they garnered
and the type of issue discussed. In Utah the issue was energy. In Iowa it was agriculture, taxes, health care and the economy. In Virginia’s Newport News, they discussed Kerrey on defense issues (important to that area), but ignored Harkin. In Illinois, Harkin’s stand on labor was discussed while Kerrey was ignored. In general, these candidates received more issue coverage from the local media outlets. Most of their coverage was focused early in the campaign since both of them dropped out before Super Tuesday.
Clinton received a substantial amount of issue/policy coverage from the Wall
Street Journal and the ABC World News Tonight. The Washington Post and New York
Times, did not focus on Clinton’s policies substantially more than his close competitors.
The bulk of his policy coverage came, however, either early, during period one, or late in the campaign, after April 7. His pattern of issue coverage is like that discussed by Graber
165 (1994). She suggests that much issue coverage comes early in the campaign prior to the action. When the action begins, the news media cover the events and results of the campaign. It makes sense that we see this occur for Clinton. But in addition to early coverage, Clinton received high levels of late coverage. Generally, media outlets paid their greatest amount of issue coverage to Clinton at the outset of the campaign and at the end of it. The later issue coverage is indicative of pre-general election coverage. Comparisons were then made of Clinton, Bush and Perot. Moreover, events such as the Los Angeles
Riots and the Rio Earth Summit gave the Clinton the opportunity to comment on Bush’s policies and thereby state his own at the invitation of the news media.
The local media did not cover the range of issues to be found in the prominent headlines of the national media. Their focus was more limited compared to the national media. Moreover, Clinton did not dominate in terms of how much issue coverage he received. He would often place second or third behind other candidates. However, the breadth of issues touched upon with reference to Clinton was substantially more than his competitors. On average, there would at least six issues discussed in relation to Bill
Clinton compared to two for his competitors.* Moreover, it seems Clinton was able to focus his message to the local media. In Miami, Clinton was covered on the topics of
Haitian and Cuban policy as well as immigration in general and gay rights. In Chicago,
Clinton on labor, urban policy, the economy and taxes made the front page of the Tribune.
*Of course, Clinton and Brown were the two candidates who stayed in the race the longest This would give t hem more time to touch on various issues and respond to various national and international ev'ents that occurred later in the campaign. However, the news media appeared to care more about what the frontrunner had to say on the issues rather than his distant competitors. This was truer of the national news media, however, than it was for the local press examined in this study.
166 Tsongas, generally fell behind Clinton in terms of issue coverage volume.
Moreover, he did not enjoy the breadth of issue coverage that Clinton received. The national media concentrated on Tsongas’ stands on taxes and the economy. He received most of his coverage early, while he was doing well in the polls and better than expected.
As to his local coverage, it too was limited in breadth. Tsongas’ main areas of issue coverage were the economy and taxes. His local issue coverage was generated early in the campaign as it was for the national media as well.
The main generator of issue coverage was George Bush. For both national and local news media. Bush’s issue coverage portions were always 80 percent or higher
(again, relative to all candidates, not simply Buchanan). Moreover, his coverage had the widest spectrum of topics. However, much of his issue coverage came early in the campaign and was generated by actions taken by Bush and his campaign organization.
Thus, in 1992 the bully pulpit was quite dominant in the generation of issue coverage.
While the absolute levels of coverage and some of the topics varied, this is one area where we see a strong coverage resemblance at the national and local level.
As to Bush’s competitor, Buchanan had relatively small amounts of issue coverage. Nationally, Buchanan received under two percent of the issue coverage, on average. His primary coverage topics were the economy and foreign policy, the budget and social welfare policy. Most of Buchanan’s issue coverage came very early in the campaign. Locally, he did worse in issue coverage volume. The two local outlets that discussed Buchanan’s polices substantially were the Manchester Union Leader (which endorsed him) and the Atlanta Journal Constitution, which did one story on his foreign
167 policy stances. The issues covered most with reference to Buchanan were taxes, the budget and the economy (usually as an attack on Bush policy rather than advocation of a new policy), foreign policy, and the arts and humanities. The aggregate coverage on these topics ranged from a high o f264 words to a low of 19 words.
4.7 Summary and Implications
This chapter has explored the similarities and differences among the national and non-national (state and metro) media outlets’ presidential nomination campaign coverage and found both. Their correspondence in some areas and variation in others sheds some light on how the news media cover campaigns and how campaigns influence this coverage.
Generally, the national news media included in this study devoted more coverage to the campaign than the non-national outlets. However, the overall coverage levels were not as far apart as some might expect. Yet one area they did differ was in the area of pre primary coverage. The national news begin their coverage of the campaign earlier than most local news outlets.
The national media are more likely to have similar patterns of when and how much they cover the campaign, whereas the local news outlets examined are more varied in when and how much they allocate to the campaign overall. Both types of outlets give more coverage to early, competitive and delegate-laden events such as Super Tuesday and
New Hampshire’s primary.
168 As to the news media’s treatment of the candidates, all of the outlet’s examined
here focused more of their attention on the frontrunners. However, beyond the first tier of
candidates we see more variation. The national media focused on the challengers and
gave little coverage to the third tier candidates. Therefore the individual national outlets
examined varied in their attention to the challengers whereas their coverage of Harkin and
Kerrey, for example, was similar. The local media, however, gave more varied attention to
the second and third tier candidates. George Bush, the incumbent president, dominated
coverage across all news outlets examined.
An examination of the negative and positive coverage given to the candidates
points out a number national and local patterns. The fi'ontrunners and their second tier
challengers receive more valenced coverage. There is little evidence of a bias toward any
particular candidate other than the one doing well at the time. Harldn, the most liberal of the candidates, certainly did not benefit from a “liberal” bias in the media. His coverage
level was low and his negative coverage relative to his positive coverage was high. Even
his own home state newspaper, the Des Moines Register gave him more negative than
positive coverage. Generally the better the candidate is doing, the more coverage that
candidate will receive. And while more coverage often brings more negative coverage as
other studies have found, candidates also receive more positive coverage. A frontrunning candidate may be scrutinized more closely as one of the top tier candidates, however, his negative coverage can be offset by positive coverage of his success and support levels.
Outside of candidate evaluations dealing with experience and character, most valenced coverage of the Democrats in 1992 dealt with the horse race. For the Republicans,
169 however, Bush received negative coverage on policy coverage and early campaign organization, while later receiving more positive coverage in campaign and horse race areas. The analysis also suggested that in 1992 the news media allocated more valenced coverage in general to the more competitive Democratic campaign. The national and local news outlets had some differences in their valenced coverage. Overall, the national media allocated more negative coverage to candidates than the local outlets examined.
The content of the news media’s coverage in 1992 evinced an interesting pattern.
When we examine the content of news coverage we find that policy coverage does tend to come early. Yet, it may also be that when the candidates left in the race are the imminent nominees, we will see policy coverage at the close of the campaign as well as the media begins its transition to general election coverage as occurred in 1992. How much of a role the likely entry of Ross Perot played is unknown in this resurgence of policy coverage. It is likely, however, that if fi-ontloading continues, leaving a few late and thus insignificant primaries and caucuses in May and June, the nominees will be determined early and begin to differentiate themselves on the issues with their general election competition.
As to this policy coverage of the candidates, in 1992 the incumbent president and the Democratic frontrunner received the bulk of national issue coverage. Mainly due to the fact that these were the candidates who not only did well, but stayed in the race longer than their opponents. Since the national media cover the campaign consistently, it is likely that these candidates would receive more policy coverage. Still, candidates who win and continue to win are more likely to have their policies examined.
170 Locally, the incumbent president dominated, but there was more variation in
coverage of the Democratic campaign with Clinton, Tsongas, Buchanan, and Kerrey and
Harldn claiming second place to Bush in some local outlet. Brown, Harkin, Kerrey and
Buchanan, overall received the least policy coverage. This suggests that it may be diflScult
for information consumers to learn about the issue stances of any candidates but the top
tier from the national media. While the local news outlets appear to be more diverse in
their allocation of policy coverage, this appearance is partly due to the timing of their
coverage. Buchanan received more policy coverage from the Atlanta Constitution
because at the time they were covering the campaign the most (near their own state’s
primary) Buchanan was active in Georgia and doing well in the state’s polls.
In this chapter I have demonstrated that there are differences between the national
and the non-national press with regard to how much attention they gave the campaign, the
direction of this coverage, and the content of the coverage in 1992. While there are
instances during the campaign and for certain candidates when similarity is more dominant
than dissimilarity, nevertheless, making a general statement that the news media’s
coverage is all alike would certainly be a mistake. However, the degree of difference is
lesser for the national media.
While I have described both the differences and similarities of coverage, two
important questions have yet to be addressed; First, what drives these differences? Is it all
about the audience one serves, one’s resources and the nature of one’s place in the order
of events? Second, what potential impact could national media have on outcomes, and more importantly can more localized media have an independent effect?
171 CHAPTERS
WHAT MOVES NEWS MEDIA COVERAGE OF THE
PRESIDENTIAL NOMINATION CAMPAIGN?
5.1 The Changing Face of Local News Coverage of Presidential Campaigns
Media and politics in today’s society are often inseparable. The mass media has become the primary conduit fi’om which voters gain most of the information they need for making a judgement on election day. While most scholarly and political practitioner attention focuses on the potential effects that media coverage has on political outcomes, often little attention is given to the forces that motivate this coverage in the first place.
Thomas Patterson’s (1976; 1981; 1993) work and that of other scholars (Arterton
1984; Brady and Johnston 1987) suggest that the “game” dominates the conceptual framework within which much election coverage is developed. The presidential nomination campaign, for example, is presented as a game to win the nomination. It is contested by various candidates (the players) who use many strategies to gain attention for themselves, eliminate their opposition, and position themselves for the general election.
The media have been self-critical of this game-oriented presentation of campaign news. As evidence of this, after the 1988 presidential election, the national press went through a
172 process of self-evaluation. Their subsequent coverage of the 1992 nomination and general
election campaign showed their own determination to concentrate less on the traditional
horse race framework and move to voter-centered coverage of the campaign (Twentieth
Century Fund, 1993). In fact, according to the Pavlik and Szanto (1994), “many national
news organizations consciously sought out ways to talk to voters, gauge their opinions
and incorporate these views into their coverage. Many assigned reporters to talk directly
to voters, and get beyond the campaign rhetoric.” (P. 8)
Moreover, the national media decision makers also used different strategies in
structuring their coverage to avoid some mistakes of the past. For example, some news
agencies rotated reporters on and off the campaigns. Hal Bruno, political director for ABC
News, believed that such strategies helped move the media away from pack journalism, a contributor to homogeneous coverage, and bring more diversity and vitality to the campaign coverage (Pavlik and Szanto 1994). One of the leading network news organizations, ABC, had the lowest levels of horse race coverage of the media outlets analyzed in this study. (See Chapter 5.) On average, the local outlets did present more horse race coverage than their national counterparts.
These self-scrutinizing routines by the national press, which were evident in 1992, are not typical of the local press, however. In 1992, the Gennifer Flowers allegations were picked up by the tabloids, then the state and regional press and finally the national news media, which then followed suit with misgivings. The national media were more willing, however, to air and print the questions regarding Clinton’s draft deferment; the story was broken by the Wall Street Journal, perhaps giving those allegations more
173 credibility. While Clinton’s competitors were somewhat less enthusiastic about attacking
the frontrunner’s morals, they were more inclined to attack his patriotism. This, along with
Clinton’s uneven, and some say poor, handling of the situation, drove the media’s
continued pursuit of the issue.
During the early 1990s the non-national news outlets moved further into national
news territory. With expansive technology from satellite relays to Internet data bases, local
reporters are far more situated to report on national events, particularly national
campaigns. In 1992 we were just beginning to see the implications of this expansion of
political information availability.
Changes in the media system have given local media outlets more room to flex their national reporting muscle. In the interest of democratic principles, this may be a
positive thing. Local reporters, particularly those from the printed press, are more likely to address issues of local importance forcing candidates to speak to policy concerns they might choose to avoid. The vastly open information system also allows reporters and editors to keep abreast of candidate statements about and commitments to policies. The days when a politician could pledge policy support in one locale and pledge the opposite in another are gone. Thus, as for candidate accountability, the changes are normatively beneficial. Moreover, given that a local news outlet’s audience is sometimes vastly different from that of the nation as a whole, we should expect that local media will respond to their needs as information consumers, much as the incentive model suggests.
Chapter five’s demonstration of the peaks and valleys of local coverage of candidates during the campaign supports this assumption. We know that there are differences
174 between national and locaJ coverage of the campaign and, more important, its participants, but in truth, we know relatively little about what political factors might drive these differences.
5.2 The Politics of Local and National Media Coverage of Candidates: Theoretical Questions
I have established that the national and local media often vary in their coverage of the campaign and candidates during the campaign. We know that how much attention they give the campaign varies over time. Their treatment of the candidates can vary as to how much attention they give, but also regarding the direction and focus of the coverage. Yet what moves this variation? How can we better understand the potential impact of national and local media if we do not understand what political forces may generate these differences at the outset?
I have shown that over the course of the campaign there are notable differences in coverage amount, valence and content of the national and local press. While differences are expected at the local level, the degree of difference found among the national media was unexpected. However, when we compare national coverage the week of a contest or group of contests, we find that there is greater, if not perfect, similarity. Thus, we can create a more inclusive and efficient measure of national media coverage by averaging the four national outlets’ coverage levels.
Some might be inclined to create an indicator of prior event coverage that attempts to capture a running average or summary judgement of coverage up to that point.
175 However, this operationalization may be problematic given the dynamic nature of the campaign. The content analysis of national and local coverage over time (from December
I, 1991 to June 3, 1994) suggests that such a measure would not capture accurately the
“current” media assessment of a candidate or the competitive nature of the most current events. These are potentially important factors if we are interested in focusing on the information that may be most relevant to voters, candidates and reporters in making assessments before a primary or caucus.
Primary and caucus goers are approached as “boundedly rational.” It is assumed that they make decisions based upon the limited information that they are willing and/or able to find (Simon, 1985). The information costs accompanying political information are also important. The lower the cost, the more likely this information will be consumed.
Thus, likely voters are assumed to have access to the greatest amount of “free information” during the time of their own primary or caucus when the salience level of the nomination campaign is increased in their particular locality. This may also trigger their attention to other sources of information, such as the national media. A.L. May, of the
Atlanta Journal Constitution, states, “. .. you can see it from the polls. When the campaign gets to Atlanta, people start to pay attention. They may already know who the frontrunners are, but they learn about the other candidates when the circus comes to town.” Moreover, according to David Yepsen of the Des Moines Register, people use the
“latest” information they receive to make their judgements, if that is available to them.
“One thing you notice about news coverage is that it changes from day to day... What we say today we will change the next day... generally because whatever we said about
176 the campaign was wrong,” he said. In addition, voters realize that the process alters and
that candidates may drop out before their primary or caucus. It makes rational sense for
them to limit information to that which comes easily to them nearest the time of their
decision. It is at this point that potential primary and caucus participants are thought to
begin actively calculating their decision to vote and whom to vote for, thus making the
local and national media coverage at this point deserving of particular attention.
The variation in national and local coverage before an event may be candidate- related. Thus, one might expect that candidate-behavior and other candidate-related characteristics contribute to the variation found in the national and local media outlets. To explore whether national and local news coverage of candidates is explained by the same or different factors, I conducted various bivariate statistical analyses and estimate several multivariate models to test the following hypotheses:
General Hypotheses:
1. Media outlets, both local and national, will pay more attention to earlier contests.
2. Media outlets, both local and national, will pay more attention to important
contests - defined as those with many delegates at stake or highly competitive
contests.
Local Media Coverage
Level
1. Local media coverage is positively related to how much attention the candidate
has paid to the state financially and personally.
177 2. Local media coverage is positively related to prior national media coverage.
3. Local media coverage is positively related to a candidate’s state or regional
afiSliation.
4. Local media coverage is directly affected by candidate viability and momentum,
but less so than national media coverage.
5. Local media coverage is directly related to a candidate’s competitive status in a
state; local competitive status meaning leading in the state polls.
6. If a scandal breaks during the week of a state’s primary or caucus, the
expectation is that the candidate related to the scandal will receive a higher share
of local coverage.
Direction
1. Local media coverage is more deferential to incumbents. Therefore, we would
expect incumbency to be associated with positive local media coverage more so
than with positive national coverage.
2. The share of positive and negative local coverage of the candidates is positively
related to the national media’s positive and negative coverage levels, respectively.
3. States with highly competitive contests are expected to generate higher levels of
negative coverage, particularly for the state’s frontrunner and runner-up.
4. When a candidate pays a high level of personal attention to a state, that candidate
will receive a higher level of positive local media coverage.
178 5. Candidates are expected to gamer higher levels of positive coverage in their home
regions and home states.
6. Momentum and viability are expected to be positively related to positive local
coverage.
7. As the number of competing candidates decreases, positive levels of local coverage
are expected to increase and negative coverage levels decline.
8. Local media positive and negative coverage is directly related to a candidate’s
competitive status in a state; thus, candidates who are leading in the polls will
receive higher shares of both positive and negative local coverage.
National Media Coverage
Level
1. Candidate viability is positively related to national media coverage levels.
2. Momentum is positively related to national media coverage levels.
3. National media coverage generally ignores lower tier candidates, and focuses on
the frontrunners.
Direction
1. The incumbent president is likely to receive more directional national news media
coverage than the challenger.
2. Negative and positive national media coverage is influenced by viability and
momentum levels.
179 3. Negative national media coverage is influenced by scandal.
4. As the number of active candidates declines, national media coverage becomes
more negative.
5. As candidates decline in the national polls, their national media coverage becomes
more negative.
Substance of Local and National Coverage
1. National media pre-event coverage is less likely to focus on issues than local pre
event media coverage.
These hypotheses regarding coverage of the campaign overall and of the
candidates before a primary or caucus may help us better understand what campaign-
related factors influence how much and what kind of coverage the campaign and the
candidates receive before a primary or caucus event.
5.3 Data
To answer these questions, I have used media coverage data generated from the national and local media outlets examined in chapter five to test the general hypotheses that require an over-time analysis. However, to test the hypotheses relating to prior-event local and national coverage, I have, of necessity, added cases to the data set. Thus, in this chapter I will be working with the two data sets discussed in chapter three: one consisting of weekly coverage breakdowns of four national and nine local news outlets composed of
13 cases (each outlet) over 25 time points (weeks); the other, the augmented sample,
180 comprising the week of (seven day units) and the week prior (previous seven day units) content analyses of 21 state daily newspapers, and a composite variable for national media coverage for the same time period.* Each data set is used to answer separate questions regarding over time coverage of the campaign and pre-event coverage of the campaign.
The local and national pre-event coverage variables are these; the relative percentage of front page local coverage the candidate received during the week (seven days) prior and inclusive of the day of the event; local coverage amount the week ( seven days) prior to the event week:, a negative and positive coverage amount for each of these weeks, and a national coverage version of these same variables. To create these variables
I coded the news coverage for each candidate for the period and operationalized it as the percentage of that week’s coverage given to the candidate compared with all other participating candidates in their party. For example, Clinton’s coverage before a state event might be 22 percent for local coverage and 56 percent for national coverage. This means that he received 22 percent of all the front page coverage devoted to the campaign’s Democratic candidates that week in the local media outlet and an average of
56 percent of the front page and newscast coverage of the four national news outlets.
5.4 Predicting General Trends in Media Coverage
5.4.1 Early Contests
When political scientists write of the flaws of the presidential nomination campaign, one fault often placed at the feet of the media is that they are biased in their
'Daily totals were aggregated into weekly totals.
181 reporting — not of the candidates, but of the contests themselves (Loevy 1993). For example, New Hampshire and Iowa, when they are competitive, outdistance other state primaries and caucuses in the attention they receive from the news media both before and after the event. Paul Gurian’s (1992) documentation of the national media’s high levels of attention to early states, and primaries (rather than caucuses), and delegate rich states gives testimony to the media’s focus of attention.
While the national media’s preferences have been documented, we know little about the tendencies of state media outlets. In their coverage of the campaigns, do they focus, as the national media does, on early contests and those deemed important due to their “must win” status or their delegate offerings? Moreover, do we see a difference in these general trends when the nature of the campaign is different? For example, the 1992 campaign on the Democratic side was a multi-candidate competition. However, on the
Republican side, the competition was limited to two candidates, an incumbent and his ideological challenger.
Challenges to incumbent presidents are frequent, and often occur when an incumbent is thought to be “weakened” in some way; however, usually the challengers are fringe candidates who are treated as such by the news media. Yet, the very presence of a challenger may contribute to negative coverage of the incumbent. A comparison of challenged versus unchallenged incumbent media coverage would be necessary to determine if this is true. Any generalization could not be determined by a study of the
1992 campaign only.
182 When the correlations of national and local weekly campaign coverage levels to
time are compared, some interesting patterns emerge. I should note that this analysis deals
with over time comparisons of coverage rather than pre-event coverage which will be
explored shortly. (See Table 5.1.) These correlations are based on comparisons of each of
the outlets’ weekly coverage levels (as to how much coverage was given to the campaign
compared with other news) with a variable measuring the time. The time variable is a
scale ranging from one to four and corresponds to the time periods discussed in chapter
five: early days (coded 1), initial contests (coded 2), winnowing and mist clearing (coded
3), and pre-convention and general election maneuvering (coded 4).
Excluding the pre-primary (early days) period as they contain no primaries and caucuses, we see that early contests are covered more heavily by both the national and the local press. There are, however, differences in how the local press treats the different campaigns. The Democratic campaign was rather conventional in that early contests received more press coverage. Note that the correlations for the Des Moines Register
(Iowa) and the Manchester Union Leader (New Hampshire) are both more than .80. This signifies the strong tendency of media outlets in these first and important states to cover their own events heavily and thus, even relatively substantial coverage of later events pales in comparison. The only local outlet not to have its coverage significantly correlated with time was the Newport News Daily Press (Virginia). Part of this is explained by the
183 Table 5.1: The Relationship of News Media Outiet Coverage and Campaign Period
Media Outlet DEM REP lÆcal AJC -.58 (.01) n.s. BDN -.57 (.01) n.s. CT -.65 (.01) -.44 (.05) DMR -.82 (.01) n.s. MH -.45 (.06) as. MS -.49 (.04) -.57 (.01) MUL -.83 (.01) -.60 (.01) SLT -.65 (.01) -.47 (.05) NNDP n.s. as.
National ABC -.65 (.01) as. NYT n.s. -.55 (.02) WP -.53 (.02) -.67 (.03) WSJ n.s. as.
Note: Time period is coded one (weeks during the pre-primary/cariy days stage); two (weeks of initial contests including Iowa, New Hampshire, Maine, South Dakota, Junior Tuesday and &>uth Carolina); three (weeks of Super Tuesday, Illinois and Michigan, Connecticut, Vermont, Kansas, Minnesota, Wisconsin and New York); and four (weeks including Pennsylvania, D C., Indiana and North Carolina, Nebraska, W. Virginia, Oregon, Washington, Arkansan, Idaho, Kentucky, and June’s Super Tuesday and North Dakota). The N is 16 weeks for each outlet when the early days stage is dropped, 25 weeks when it is included. A negative sign indicates a decrease in coverage over time.
Table entries are bivariate correlations with statistical significance in parentheses. The entry n.s. means the correlations were not statistically significant at the .10 or less level.
Key: AJC=Atlanta Journal Constitution (GA); BDN=Bangor Daily News (ME); CT=Chicago Tribune (IL); DMR=Dcs Moines Register (lA); MUL=Manchester Union Leader (NH); MH=Miami Herald (FL); MS=Milwaukee Sentinel (WI); NNDP=Newport News Daily Press (VA); SLT=Salt Lake Tribune (UT).
consistent and relatively proportional coverage devoted to the campaign over time
(generally all from wire or news service reports unless they dealt with Governor Douglas
Wilder’s short venture into the nomination). Rather than running large stories on occasion that dominated the front page, the Newport News Daily Press ran smaller stories on a very consistent basis.
184 The national media varied somewhat in their coverage over time of the Democratic campaign. The ABC news organization covered the campaign to a greater extent during the earlier portion of the campaign, as did the Washington Post. The New York Times and the Wall Street Journal’s coverage were not significantly correlated with time. The national media followed a somewhat similar pattern in that they devoted more time to periods two and three on average, the earlier, competitive campaign events, than the latter ones. Like the local media, the national media provided lesser coverage of the campaign during the pre-primary period or in the post-New York time period, during which Clinton hit the halfway mark for delegates needed to win the nomination. It was at this point that the campaign was no longer even remotely competitive.
The coverage of the Republican campaign was somewhat different. We see much less of a correspondence with time and coverage in the local press. Five of the nine outlets included in the over time local media analysis did not have coverage levels related to time.
Part of this is due to the coverage of President Bush whose coverage levels remained relatively consistent throughout. Moreover, Buchanan had never dominated the headlines for a for a sustained period of time, and after Super Tuesday, Buchanan all but disappeared fi'om the media’s coverage of the campaign.
S.4.2 Critical Contest Weeks
Contests that become “critical” ones are also the focus of both the national and local media. First contests, like New Hampshire, and competitive contests such as South
Dakota for the Democrats and Georgia for the Republicans, are the focus of a great deal
185 of media attention. Because these contests have implications for the competitive structure
of the subsequent contests (who will be still be competing), they are given attention and
analyzed both before and after their voting takes place. It is generally a combination of the
efforts of the media and the candidates that determines which contests are critical. Often
candidates will set expectations themselves — forecasting what they or their opponents
need to do in a state. The media, in their role as analyzers of the process, make assessments of which contests are important. For example, Florida was a key state for
Super Tuesday rather than the many other contests because it was considered a good test market, free from regional biases toward one candidate or the other, thus making it
“competitive.” Moreover, the results of the Florida primary had significant implications for
Tsongas’ campaign. Critical contest weeks were coded one, and all others coded zero. I use the assessments of political analysts, reporters and candidates to determine whether or not a particular week’s event should be categorized as critical for a candidate or candidates."
The results of an analysis of the correlations between the local and national weekly coverage levels and whether a critical contest occurred during that week lend support to the hypothesis that the news media, both local and national, do pay attention when an important contest takes place. Yet, it is supported more for the Democratic campaign with its multi-candidate field than for the incumbent-challenger Republican campaign (See
Table 5.2).
"Critical contest weeks are listed in Table 5.2's note.
186 Table 5.2 The Relationship of News Media Outlet Coverage and Critical Contest Week
Local DEM REF MUL .31 (.06) .43 (.08) DMR .28 (.06) as. BDN .85 (.01) n.s. SLT .67 (.01) .62 (.01) AJC .75 (.01) as. MH .54 (.01) as. CT .58 (.01) .61 (.01) MS .55 (.01) .76 (.01) NNDP .26 (.10) .41 (.10)
National DEM REP ABC .57 (.01) as. NYT .55 (.01) as. WP .45 (.02) as. WSJ .48 (.01) as.
Note: Critical Contest Weeks inelude; for the Republicans - the week of the New Hampshire primary, the week of the Junior Tuesday (due to Georgia’s primary), the week of Super Tuesday, and the week of the Dlinois and Michigan primaries. For the Democrats the weeks are the same with the addition of the week of the New York primary. Pre-primary cases are not included in the analysis. The N is 16 weeks for each outlet when the early days stage is dropped. 25 weeks when it is included. A negative sign indicates a decrease in coverage over time.
Table entries are bivariate correlations with statistical significance in parentheses. The entry n.s. means the correlations were not statistically significant at the .10 or less level.
Key: AJC=Atlanta Journal Constitution (GA); BDN=Bangor Daily News (ME); CT=Chicago Tribune (IL); DMR=Des Moines Register (lA); MUL=Manchester Union Leader (NH); MH=Miami Herald (FL); MS=Milwaukee Sentinel (WI); NNDP=Newport News Daily Press (VA); SLT=Salt Lake Tribune (UT). ______
The strength of the relationship for the Democratic campaign varies somewhat among the local outlets, again, the Des Moines Register and the Manchester Union Leader standing out with relatively weak and fewer significant relationships. The Newport News
Daily Press, too, stands out with the lowest correlation and a significance level of only. 10.
Again, this is due to the structure of their coverage and their consistent treatment of the campaign.
187 The national press were very evenly correlated regarding their attention to the
Democratic campaign during critical contest weeks. What is interesting however, is that
national coverage of the Republican campaign is not related to critical event weeks. The
Bush campaign dominated the press during the nomination campaign generally and received relatively high levels of coverage despite whether or not an important contest occurred that week. Buchanan’s coverage increased slightly, but by Super Tuesday, he was not longer significantly featured in the news nationally.
Some local media, on the other hand, do appear to have varied on whether their coverage of the Republican campaign was influenced by the timing of important primary and caucus events. The Chicago Tribune shows a moderate correlation with coverage level and the occurrence of a critical contest. Most of this, however, is driven by coverage of Bush rather than by Buchanan. While the Chicago Tribune covered the president heavily during the campaign, his level of coverage intensified when something important happened such as his less-than-expected vote in New Hampshire and in Georgia.
The Salt Lake Tribune’s coverage again generally focused on the Bush campaign.
Buchanan received substantial coverage for a six-week period fi'om February 9 (a week or so before New Hampshire) until the week after Super Tuesday. The same can be said of the coverage of the Milwaukee Sentinel and the Newport News Daily Press. Most wire stories dealing with the Bush campaign found their way onto the fi'ont pages of these three newspapers making them appear to be deferential to any presidential activity.^
^During the campaign, the Milwaukee Sentinel devoted 6,333 front page words to Bush and 985 to Buchanan. The Chicago Tribune, a non-deferential outlet, while affording Bush a great deal of coverage, also covered Buchanan substantially. Buchanan ranked fourth overall in terms of the Chicago Tribune’s candidate coverage.
188 Overall, the hypotheses that the news media, both local and national, are more likely to give more coverage to the campaign when early and critically important events are involved are confirmed for the coverage of the Democratic campaign. Given the circumstances surrounding the incumbent president, a badgering opponent and a critical populace, the coverage levels appeared quite high during the early days and initial contest stages; fell during the mist-clearing stages when Buchanan’s campaign was all but ended, and increased during the last stage when the Bush campaign moved into its pre-convention stage, and his media coverage began to be increasingly affected by the Perot factor.
5.5 National and Local Pre-Event Coverage
Chapter five’s examination of media coverage of the campaign and candidates over a 26 week period found that certain candidates tend to dominate the stories printed and broadcast by the news media. When we turn to an examination of media coverage immediately before an event, we see that this is true here as well.
Table 5.3 displays the means and standard deviations for the relative levels of national coverage for each candidate compared with his party competitors. Clinton and
Bush received the highest levels of national pre-event coverage as expected. Tsongas was a distant second behind Clinton, and Buchanan fell 74 percentage points behind Bush, on average. Kerrey stands out somewhat. Even though he left the campaign early, he received, on average, more pre-event coverage than Harkin and Brown. While Harkin and
Brown were in the campaign. Harkin generally received less coverage than Kerrey.
Brown’s lower average level is due to little early and late coverage. Most of his coverage
189 came near the middle of the campaign in March and early April. Although Kerrey’s win in
South Dakota boosted his share of national and local positive coverage significantly for a
brief period, there was not a subsequent movement in his national poll standing. South
Dakota failed to bring Kerrey more support or money.
In contrast to the national media, the local media exhibited far more equality in
their distribution of total coverage as shown in Table 5.4. As for the Democrats, Tsongas’
and Clinton’s mean coverage levels were separated by less than two percentage points.
Of course, each locale had a varied campaign context, and we would expect greater
diversity consequently.* Yet this varied context did not seem to afiect the coverage of the
Republican campaign. Republican candidate local coverage means were very similar to the national coverage means. Bush dominated local press coverage although the
dispersion of coverage levels varied for him more at the local level than the national. The local media also gave Buchanan on average almost twice as much pre-event coverage than the national media. Of course the Republican campaign was less complex, involving only two candidates one of those an incumbent and one an insurgent Republican challenger
with few hopes of winning the nomination.
* Aggregating the local news outlets pre-event coverage does not control for the differing placement of outlet in states with different positions in the temporal sequence of events, varying competitive levels and attachments to particular candidates. Moreover, these figures might be different (although I suspect a relatively marginal difference) were more outlets included in this sample. Therefore, these are not meant to be generalized to all local outlets, but rather utilized as a gross indicator of local coverage to be compared with national coverage taking place at the same time.
190 Table 53: Pre-event National Coverage Means by Candidate
Democratic Candidates Total Coverage Mean Std. Dev. Brown 13.78 9.62 Clinton 45.40 18.32 Haridn 9.54 2.99 K err^ 18.89 10.11 Tsongas 28.23 12.96
Republican Candidates Total Coverage Mean Std. Dev. Buchanan 17.32 11.40 Bush 80.74 11.27
Note: Statistics based on national coverage before a state event Only those pre-event time points corresponding to the time points examined for the local news media are included.______
Table 5.4: Pre-event Local Coverage Means by Candidate
Democratic Republican Candidates Total Coverage Candidates Total Coverage Mean Std. Dev. Mean Std. Dev. Brown 12.76 10.02 Buchanan 23.53 19.76 Clinton 33.51 24.68 Bush 73.72 18.72 Harkin 15.33 21.82 Kerrey 18.96 14.09 Tsongas 3S32 18.00
Note: Statistics based on 21 states included in the augmented local sample. The N’s are determined by the time period in which the candidate was active. Clinton (N=21); Brown (N=2I); Harkin (N=l 1); Kerrey (N=9); Tsongas (N=I7)._____
5.6 Possible Factors Influencing Local Media Pre-Event Coverage of Candidates
5.6.1 Candidate Attention
It has been hypothesized that local media coverage of the candidates may be affected by the attention that a candidate gives a particular state in both personal and
191 surrogate appearances. This variable is operationalized as a dummy variable coded one if a candidate and his surrogates devoted substantial time to a state relative to his opponents.
If two candidates are investing similarly high levels of time, both receive a value of one.
While the actual number of candidate and surrogate visits could not be tabulated with exactness, the fact that a candidate spent the most time in state compared with competitors could be established through a variety of sources (books, magazine and newspaper articles). The relationship between these factors and how much attention the candidate will receive from the local press is displayed in Table 5.5. This table also presents the data on the relationship of candidate spending (relative and absolute) and local news coverage of the candidates.
Candidate attention and spending far better predict the coverage a candidate receives right before a contest. The coverage of the campaign during the previous week is often devoted to prior campaign events that may not be state-related, or the local outlet may devote no attention to the campaign at that time.^ This varies substantially across local outlets and was very true of some outlets in states with later contests that were no longer competitive, generated low turnouts, and were predictable by that point. Many local newspapers delayed full attention to their own primary or caucus until the week of the event. They may be following the campaign in sequential order, as suggested by the campaign model of news coverage, almost as if they were deferential to the other contests until it was their own turn to take the spotlight. Moreover, with the frontloading and bunching of primaries, it is diflScult for candidates to get to states more than a week before
*This appears to be true of inside page coverage as well.
192 the primary or caucus is scheduled. This, in turn, could be responsible for the local news media delaying attention to their own state’s event.
Candidate attention captures whether or not the candidate spent a great deal of time and effort contesting a state primary or caucus. Note the relatively moderate and statistically significant correlation for the Democratic campaign reported in Table 5.5. This is to be expected. Personal attention should matter, particularly in a complex, multi candidate campaign, but is sometimes outweighed by other factors such as national viability or competitive status in the state. The local press, for example, may cover a candidate’s visits, but if he is not the frontrunner or rurmer-up, he still may not gamer local media attention. However, this does not mean he will not gamer votes. Jerry Brown did not receive significant levels of prominent local media attention fi’om the Bangor Daily
News the week prior to the campaign, but he did end up winning the Maine caucuses.
Sometimes deciding not to visit a state gets the candidate on the fi’ont page
— if the candidate is a frontrunner. Clinton made a phone call to the Raleigh, North
Carolina, News and Observer explaining that he was needed in Los Angeles after the
Rodney King riots. His phone call made the front page (“Clinton views LA as pivotal:
Candidate skips state visits today”), while Brown’s visit to Chapel Hill made the back of the second section. (News and Observer. May 4,5) Note however, that candidate attention has no correlation with local coverage shares in the Republican campaign.
193 Table S.S: The Relationship of Candidate Attention and Spending with Local News Media Coverage
Democrats Local Coverage 1 Week Prior 2 Weeks Prior Candidate Attention .30 (.01) n.s. N=79 Relative Spending .47 (.01) n.s. N=79 Absolute Spending n.s. as. N=79
Republicans Local Coverage 1 Week Prior 2 Weeks Prior Candidate Attention n.s. as. N=35 Relative Spending .85 (.01) .64 (.01) N=35 Absolute Spending n.s. as. N=35
Note: Table entries are bivariate correlations with statistical significance in parentheses. The entry n.s. means the correlations were not statistically significant at the. 10 or less level.______
5.6.2 Candidate Relative and Absolute Spending
Theoretically, relative and absolute spending are two different factors. Relative spending is a measure of competitive spending - is the candidate spending more than his competitors? Absolute spending signifies the magnitude of the investment. A candidate can be spending far more than his opponents, but it may be that they are all spending relatively small amounts of money. Spending data were gathered fi'om the Federal Election
Commission reports on financial activity for the 1992 presidential nomination campaign.
Relative spending is operationalized as the candidate’s share relative to all party competitors, while absolute spending was simply the actual number of dollars spent in a state. As Table 5.5 displays, relative spending is a significant predictor of immediate prior- event coverage. The absolute amount of spending is unrelated to coverage levels. This suggests that outspending one’s opponents may have merit, but it is potentially a spurious relationship. Candidates who are doing well can spend more money relative to their
194 opponents. Candidates who are doing well also receive more coverage. The actual
number of dollars spend does not appear to be related to how much local coverage the
candidate receives. Thus, local news reporters appear to use financial information only as
it is a means to designate who is able to outcampaign challengers rather than paying
attention to the actual amount spent in their state.
5.6.3 Performance Factors
While it is expected that performance factors such as viability and momentum will
influence local coverage, the relationship is expected to be weaker than for the national
media. Local news outlets are expected to pay more attention to local factors that may
dampen the importance of a candidate’s prior performance. Viability is operationalized as
the number of delegates accumulated prior to a contest relative to half the number
required for nomination. Table 5.6 suggests that viability is an important indicator of the
level of coverage a candidate will receive fi'om a local media outlet for candidates
participating in both partisan campaigns. Momentum, on the other hand, had no
relationship with local media coverage share for either the Democrats or the Republicans.
Having won a prior contest or group of contests did not translate into a significant
increase or decrease in one’s share of the local news coverage. Momentum was
operationalized as a dummy variable coded one (weighted by the number of contests won)
if a candidate won a prior contest or group of contests, and zero if not.
Examining the relationship for each candidate, we find that Clinton’s local coverage share is strongly correlated with viability, as was Bush’s. Clinton’s relative share
195 of the local media’s coverage, in particular, grew steadily as he amassed delegates and as his competitors dropped out. This is true especially after New York when Brown was still in the campaign, but gave no challenge. Clinton won every subsequent contest by a strong margin, increasing his share of the delegates, and garnering, relative to Brown, most of the local media’s coverage.
As to the potential relationship of momentum to local coverage level, Tsongas appears to have been the only candidate to have short term “winning” associated with local coverage share. This is explained by his surprising New Hampshire win. His local coverage rose significantly after this contest, however, as he could not sustain momentum, his coverage levels decreased.
As to the Republican contest, the effect of momentum carmot be measured in terms of winning contests, for Buchanan won none. Since the momentum value can increase when multiple contests are won, we can still test for an effect of Big Mo versus
Small Mo for Bush. For example. Bush’s winning all the Super Tuesday contests generated a score of 11 for the momentum value. His win of the South Carolina primary generated a value of one. Candidates who lose, receive a zero. But even if we measured momentum in terms of expectations, Buchanan’s better than expected performances were few and early.
As to viability, as Bush amassed delegates his coverage in the local press did increase by a statistically significant amount, while Buchanan’s decreased as he stayed in the race garnering a few delegates here and there, and thus, increasing his viability score in
196 small but real numbers. In fact, his viability score increased incrementally while his coverage levels dropped to zero.
I should note that being the frontrunner and being the most viable are often used synonymously by the media and other interested observers of the campaign. Yet they are different concepts. Being the frontrunner means that the candidate is ahead, but it gives no indication of how far ahead. Tsongas was called the frontrunner after New Hampshire, but his viability level was not significantly greater than Clinton’s. After March 3 (Junior
Table 5.6: The Relationship of Performance Factors and Local News Media Coverage
Democrats Local Coverage 1 Week Prior Viability .49 (.01) N=79 Momentum n.s. N=79
Republicans Locai Coverage 1 Week Prior Viability .63 (.01) N=35 Momentum n.s. N=35
Viability by Candidate Momentum by Candidate Brown n.s. n.s. N=21 Clinton .71 (.01) n.s. N=21 Harkin n.s. N=ll Kerrey n.s. n.s. N=9 Tsongas n.s. .29 (.10) N=17
Bush .63 (.01) n.s. N=19 Buchanan -.53 (.03) N=16
Note: Table entries are bivariate correlations with statistical significance in parentheses. The entry n.s. means the correlations were not statistically significant at the. 10 or less level.______
Tuesday in 1992) the race was described by media across the country as “lacking a frontrunner” and “having no frontrunnner emerge” but rather a host of regional winners.
197 After Super Tuesday, Clinton became the fi'ontrunner, and his delegate totals propelled
him quite a distance fi'om his nearest challenger giving more weight to his first place
position.
5.6.4 Home State and Home Region Affiliation
Another factor thought to influence the local coverage of a candidate would be whether the candidate is fi'om a particular state or region. The assumption is that the local
press will be more interested in covering the exploits of a political figure familiar to them or perhaps having some history with their area. Home state means the candidate was bom or lived in a state for a significant amount of time and is considered “fi'om” that state.
Home region refers to states that border a candidate's home state, excluding an opponent's home state.For example, Bob Kerrey’s home state is Nebraska, and the states comprising his home region are South Dakota, Wyoming, Colorado, Kansas, and
Missouri. Iowa is not counted since it is the home state of another candidate. These states would be coded one as a dummy variable for a home region state. All other states are coded zero.
As Table 5.7 suggests, in general, home region is not a strong indicator of a candidate’s level of coverage by the local press. Home state, however, has a relatively strong correlation to coverage two weeks before an event, but a lesser correlation with coverage the week of the event. The local sample used here includes only three home state cases, Tsongas in Massachusetts, Harkin in Iowa, and Bush in Maine. Given the Boston
Globe’s reputation for tough political reporting (of all the outlets in the state sample, the
198 Boston Globe focused mainly on the game aspect of the campaign and was relatively critical of Tsongas throughout its coverage). Bush’s coverage in Maine, while somewhat higher than in other states, was not as unique as the Register’s or Globe’s coverage of
Harkin and Tsongas, respectively.
It is often assumed that one reason candidates do so well in their home states is that the media there give them a free ride. This is not so. Tsongas received no free ride.
Harkin, while receiving a great deal of coverage from the Des Moines Register, was often covered negatively (generally in the wire stories that were run but also by local reporters writing in the state or from their Washington bureau). Bush’s coverage in Maine seemed very objective rather than biased toward him as a native son. Yet home state may still be a necessary control in predicting local coverage levels given the local media’s penchant for devoting more coverage on average to the native candidate.
Home region, while unrelated to coverage the week of the event except in the case of Harkin, is a weak predictor for Brown, Clinton, and Kerrey of coverage received in the second week before an event. Brown and Kerrey were visible in the local press in the West and Midwest, respectively, more so than in other regions. Part of this however, is due to the fact that they concentrated their eftbrts there. Clinton, as well, had somewhat higher levels of coverage in the areas surrounding Arkansas, but his coverage levels were high enough everywhere else to negate the impact of region.
199 Table 5.7 The Relationship of Home State and Home Region and Local News Media Coverage
Democrats Local Coverage 2 Weeks Prior Local Coverage 1 Week Prior Home Region as. as. N=79 Home State as. .28 (.01) N=79
Republicans Local Coverage 2 Weeks Prior Local Coverage 1 Week Prior Home Region as. as. N=35 Home State as. as. N=35
Home Region by Candidate Brown .24 (.10) as. N=21 Clinton .27 (.09) as. N=21 Kerrey .53 (.07) as. N=9 Harkin n.s. .87 (.01) N=ll Tsongas n.s. as. N=17 Bush as. as. N=19 Buchanan as. as. N=16
Note: Table entries are bivariate correlations with statistical significance in parentheses. The entry n.s. means the correlations were not statistically significant at the .10 or less level.______
5.6.5 Competitive Field
We might expect that the competitive nature of the contest might affect how much coverage the locai press gives the candidates. For example, the expectation is that early when many candidates are competing, the media will be forced to be less equitable in their coverage, simplifying the complex competition by focusing their attention, and thus the voters’ attention, on the top tier of candidates. When the competition moves from many candidates to fewer, the media is expected to portion out its coverage more equitably, particularly if the campaign is still competitive. Table 5.8 suggests that the local media follow this pattern somewhat, but not strictly, given the other variables that come into play.
200 At the outset of the primaries and caucuses in February when the Democratic campaign included five candidates, the local press appears to have been somewhat equitable in the distribution of coverage; Tsongas 29.88 percent, Clinton 22.43, Harkin
19.9 (although this is somewhat inflated by the Des Moines Register’s very high level of coverage of Harkin); Kerrey 18.41 percent. The national press concentrated far more of their attention on Clinton and Tsongas, followed by Kerrey some distance behind and then the rest of the candidates. In a sense the national press appeared to follow the suggested hypothesis more closely than the local press. After the field diminishes to four, we do see a rather drastic change. Harkin’s coverage drops significantly, followed by low levels for
Brown as well, and most of the coverage is left for the top tier candidates, Clinton and
Tsongas, to divide. While the national press divided it rather equitably, the local press averages were favorable of Tsongas, both while Harkin was in the campaign and after he dropped out. While many state newspapers did not endorse primary candidates, Tsongas was generally the pick of many of those that did in this sample of 21 state newspapers.
When Tsongas suspended his campaign, leaving only Brown to challenge Clinton, the coverage of the two remaining candidates appeared somewhat even at the national level. Brown, as the only challenger, saw his coverage levels surge to heights they had never enjoyed previously. This was short-lived. After the New York primary, where
Brown was unable to defeat Clinton, his coverage levels dropped to their previous low levels of coverage. Brown’s national poll standings dropped as well.
201 Table 5.8: The Relationship Between Candidate Field Size and Local News Media Coverage
Democrats: Average Local Coverage Amount Relative to Competitors
No. Candidates Five Four Three Two Two prior NY post NY Brown 12.03 7.95 10.10 52.0 16.0 Clinton 22.43 20.0 33.58 29.0 78.0 Harkin 19.9 2.50 ------Kerrey 18.41 ------Tsongas 29.88 42.8 38.79 ------
Correlation with Local Coverage Share: -.31 (.01)
Note: The local coverage sample includes the following states at each stage of competition: Five competitors - Iowa, New Hampshire. Maine, South Dakota, Georgia, Idaho, Colorado, Maryland, Utah; Four competitors - Arizona and Nevada; Three competitors - Florida, Louisiana, Massachusetts, Mississippi, Tennessee, Illinois; Two competitors prior to New York - Wisconsin; Two competitors after New York - Virginia, North Carolina and Oregon.______
5.6.6 National Attention
One other important influence on the local level of candidate coverage is the attention that the national press is giving the candidate. No man is an island unto himself, nor is any news media outlet. The local news media still gains much of their information about the campaign from the same sources as the national media (so similarities are to be expected), and they follow the national media’s analyses and reports of the campaign as well. As Mark Woodward of Bangor, Maine’s Daily News said, “As a reporter, you really envelope yourself in the information available. And a great deal of this comes from the leading national newspapers and the wire services. Television is a part of it, too. You watch the speeches on C-Span and the weekend political talk shows, and you pay attention to what the major networks are saying. But it all blends in with your own context
202 the needs of your audience, and what you may be interested in and know personally.”
(Interview March 1996)
The correlation between the national coverage of a candidate during the week of an event and the local coverage of the same is significantly related as shown in Table 5.9.
Yet, this relationship varies across candidates as well and should not be thought to suggest a local media that simply replicates the national media’s reports. Rather the strong correlation is influenced by very similar coverage of the fiontrunners.
Table 5.9: The Relationship Between National News Media Coverage Levels and Local News Media Coverage Levels
Democrats Local Coverage 1 Week Prior National Coverage .63 (.01) N=79
Republicans National Coverage .84 (.01) N=35
Note: Table entries are bivariate correlations with statistical significance in parentheses. The entry n.s. means the correlations were not statistically significant at the .10 or less level.______
5.6.7 Local Competitive Status
Finally, a candidate is expected to receive more local coverage relative to his peers when he is the local frontrunner, or in the case of a very close contest, the two local frontrunners. The correlation (.68, .01 two-tailed) between a candidate’s local competitive status and local coverage suggests that this is true. This relationship only applies for the
Democratic campaign since the Republicans had only one frontrunner nationally and in each locale George Bush. While there were areas where he had more support,
203 Buchanan was never close enough to mount a serious challenge to Bush in more than a symbolic manner.
5.7 Toward a Model of Local Candidate Coverage
The information garnered from the bivariate statistical analyses suggests that a great portion of the amount of local coverage a candidate receives is a function of a limited set of variables. The level of local coverage a candidate receives is purported to be a function of the following; candidate attention given to the state, the candidate’s level of relative spending, the candidate’s viability score before the event, whether or not the contest is in his home state, the number of candidates competing for attention, the candidate’s level of national coverage during the same time period, and finally, the candidate’s competitive status in the state. This general model is applied to both party campaigns.
Table 5.10 presents the estimates from an ordinary least squares regression of the aforementioned variables on the dependent variable of percentage of local coverage relative to competitors for the Democratic campaign. The cases used in the regression are the levels of pre-event coverage for each of the state newspapers in the augmented sample.® The results of the regression analysis suggest that different competitive
®For example, if Tsongas received 34 percent of the coverage in state newspaper X relative to his fellow pursuers of the Democratic nomination for the seven day period before but including the day of that state’s primary, the dependent variable for that case would be 34 percent. Thus, the each candidate represents a case. The data are pooled cross-sectional and the range of candidates from the beginning of the campaign to the end are 5 to 2.
204 circumstances (multi-candidate competitive versus two-candidate, incumbent dominant) are covered differently by the local media.
Table 5.10: Model of Local News Media Coverage: Democrats^
Variable B SEB Beta Sign, of T Candidate Attention 3.64 3.34 .084 .2804 Relative Spending .08 .07 .099 .2359 Viabihty .12 .18 .083 .4842 Home State 25,88 9.71 195 .0095*** Candidate Field Size 1.63 1.79 .085 .3660 National Coverage .45 .11 .402 .0002*** Loc. Compet Status 16.13 3.85 .377 .0001*** (Constant) -4.34 8.52 .6121
SE of Estimate: 12.87 Adjusted R Square: .62 N=79
Note: The correlation between national coverage and viability is .698, .01 two-tailed. Table entries are ordinary least squares regression estimates. ***=<.01 .**=<.05, *=<.10
^There are special problems associated with the estimation of regression equations describing the behavior of individuals or institutions over time. Problems such as autocorrelated error or nonconstant error variance. These are not aways present, however, given the potential for their presence I attempted to assess the degree of distortion by using, as Dielman (1989) suggests, an ANCOVA or LSDV model. 1 ran each of the regressions using dummy variables to represent time and individual variation. The estimates for the variables remained relatively stable and the same ones were generally statistically significant One difference, however, was that the R-squareds generally increased. A more accurate estimation of the variables influence would be achieved with pooled cross-sectional times series analysis.
205 Table 5.11: Model of Local News Media Coverage: Republicans
Variable B SEB Beta Sim. of T Candidate Attention 15.34 5.44 .246 .0036*** Relative Spending .30 .12 .423 .0168** Viability .21 .09 .247 .0333** Home State -27.15 14.64 -.152 .0739* National Coverage .30 .15 .345 .0583* (Constant) 8.85 5.45 .0748
SE of Estimate: 13.53 Adjusted R Square: .7968 N=35
Note: Candidate field and local competitive status arc constant for tiie Republicans. Table entries are ordinary least squares regression estimates. ***=<.01, •*=<.05, *=<.10
The Democratic campaign was by far a more competitive campaign, although in
1992 the ffontloading and the mix of regionally diverse early contests created an interesting dynamic. The Democratic candidate’s local coverage shares were best explained by the level of national coverage received, their competitive status in the state, and whether or not they were competing in their home state.
This finding is normatively disappointing. Local media attention to candidates may be driven partially by the national media's assessment of who is doing well, their own assessment of who is doing well in their state (based on local polls and endorsements generally) and whether or not the candidate is a native son. While we see that the local media is not immune to horse race coverage, at least we can assume that this attention is driven by polls generated by a constituency that is somewhat different from the national public both in terms of issue concerns and perhaps in ideology.
206 As to the Republican campaign, almost 80 percent of the variance in local coverage shares for Bush and Buchanan is explained by the presence of six variables. The regression results suggest that personal attention to a state could increase one’s local media coverage substantially. Moreover, the more one spent, the more coverage one would receive in the local press. This is verified by the fact that Bush generally out spent
Buchanan and received more local coverage on average. But when Buchanan did spend more (although he never did outspend his opponent) it was in targeted states that he believed would be more supportive of his candidacy. Subsequently, his coverage levels increased in these states. Generally, these were also the states in which he devoted much personal attention.*
Viability is a statistically significant indicator of local pre-event coverage.
Buchanan won some New Hampshire’s delegates on February 18, increasing his viability and media coverage simultaneously. As the Republican campaign progressed, Buchanan’s coverage dwindled to nothing. He managed to attain a few delegates here and there over the course of the campaign; thus, his viability increased in absolute terms, but the distance between himself and Bush was wide. His campaign’s significance began to diminish after
Super Tuesday and continue on a downward spiral. After Michigan, Buchanan’s efforts were a “non-story.” Bush, after Super Tuesday, began a continued sweep of the primaries and caucuses and generated a great deal of post-event coverage. With Buchanan out of
When Buchanan’s coverage shares are examined individually, it becomes obvious that not only did his share increase when he targeted money, but also when he invested a great deal of personal attention. Buchanan’s coverage levels were higher in more conservative states as well. Finally, greater levels of relative spending suggested Bush’s incumbency advantage and could be conceptualized as a defacto control for incumbency. He was bankrolled exceedingly well, and generally spent substantial amoimts of money despite Buchanan’s strength in a state.
207 the way, however, the focus of the Bush coverage switched to comparisons between the
incumbent and his potential general election challengers, and problems Bush might have
come November. The strength of viability is mainly due to Bush’s viability increases
corresponding to increased levels of coverage, particularly during the latter weeks of the
campaign. In some respects, viability, while significant, is not as theoretically interesting in
the Republican context as it is in the Democratic context.
5.8 Possible Factors Influencing the Direction of Local Media Coverage of Candidates
While a substantial portion of the variance in the amount of local media coverage a candidate receives can be accounted for by candidate performance and candidate characteristic variables, what about the direction of this coverage the positives and negatives to be found within its content? How might they be accounted for?
As discussed earlier, the valence of coverage is operationalized as the relative portion of positive or negative coverage the candidate receives as discussed in Chapter
Three.® For example, the amount of coverage is determined for a week, and of this
® Alternate measures of negative and positive coverage were considered Analyses using a measure that captured only the absolute amount of negative or positive coverage and a measure that captured a candidate’s negative and positive share relative only to his own coverage totals were conducted However, both analytically and theoretically, the measure that compares a candidate’s negative and positive coverage to that of the negative and positive coverage garnered by his opponents appears to be most worthy. Theoretically, voters are taking in information about a candidate that is often set in a comparative context with other candidates. Thus, the measure must reflect this relativity. However, while measure of absolute negative and positive coverage and the candidate- relative measure perform badly in the statistical tests, and this competitor-relative measure does comparatively better, the measure would be enhanced if the absolute and relative components could be collapsed For example, an even more empirically and theoretically satisfying measure would allow us to compare a candidate’s relative directional coverage controlling for absolute levels of coverage. How do create this measure is somewhat problematic, but must be addressed in future analyses.
2 0 8 coverage the amount of positive and negative coverage in words is also tabulated. Then each candidate’s portion of the total negatives and positives is assessed. For example, if during the week, the total negative coverage was 150 words and 15 words of this came from a comment on Brown, his negative score for the week would be 10 percent. It is assumed that candidates getting more positive coverage would do worse than candidates receiving more negative coverage in the eyes of information consumers and potential voters. However, the relationship may be conditional upon other candidate factors and structural factors (such as the time period of the campaign).
These directional variables do not indicate the absolute level of negative coverage
(for this would be difBcult to compare across outlets with differing space constraints) nor does it compare negative coverage to some outlet-related base such as total front page or 22-minute broadcast, but rather the more theoretically valuable comparison is made to the candidate’s competing peers— the assumption being that it does not matter so much the absolute value of negative or positive coverage, but rather the relative value. Is one receiving the most negative coverage relative to one’s competition? the most positive? How will voters respond to a candidate who is consistently given negative coverage in the news media? Does it matter from which news outlet the negative or positive coverage originates? In this section, the bivariate relationships between a number of variables and the relative proportions of negative and positive coverage will be explored.
*°How would we compare a 100 words in the Miami Herald which on average only prints 810 words on its front page versus 100 words in the Des Moines Register which usually prints 1,814?
209 5.8.1 Incumbent Deference
One might expect the local media to be more deferential to the incumbent president. The bypass strategy is founded on the premise that the local media tend to be less critical and more willing to pass on the candidate’s message. But this may be true only in the case of personal interviews. With regard to what they print, the local press does not seem to be substantially deferential in terms of how much positive or negative coverage they give to the president prior to their own primary or caucus.
Table 5.12 displays the results of an analysis of variance of local positive and negative pre-event coverage given to the candidates in the Republican party testing for the effects of Bush’s position as incumbent. It is evident that the local media had far more positive coverage for the president than for his challenger, Buchanan. However, the same can be said of negative coverage. Bush’s average portion of the local positive coverage was 53.90 percent (Std. Dev. =40.02), however, his negative portion was 75.81 percent
(Std. Dev.=31.41). However, we may say that the local press is somewhat deferential when we compare these totals to the national media’s distribution of positive and negative coverage; Bush received only 43.49 percent of the national positive coverage on average with a standard deviation of 19.94 percent. His national coverage negative portion was
77.28 with a standard deviation of 13.13 percent. These numbers suggest that the president’s national positive coverage is rather low and with much less dispersion among the over time points examined in thus study, while the local positive coverage, which was somewhat higher, ranged substantially among outlets and over time points. Comparing his
210 coverage to Buchanan’s, however, demonstrates a wide gap in directional coverage of both positive and negative coverage.
Table 5.12: Incumbent Deference by the Local News Media
Republicans Negative Coverage Positive Coverage Mean Std Dev. Mean Std Dev. Buchanan 26.42 33.72 17.30 26.79 Bush 75.81 31.41 53.90 40.02
Note: If we compared the two candidates’ directional coverage as a percentage of their individual coverage rather than as a percentage of the total given to both candidates, the values deflate somewhat. Bush's mean negative coverage relative to his own total coverage was 22 percenL In other words, on average. Bush’s pre-event total coverage was comprised of 22 percent negative coverage. The corresponding value for Buchanan was 10 percent As to positive coverage, the differences are less visible between the two candidates. Bush, on average, had 10 percent of his pre-event coverage comprised of positives, while Buchanan had 8 percent However, these numbers do not do justice to the sheer amount of coverage that Bush received. His 22 percent negative means almost one-flflh. Since Bush could average 439 words in coverage during a pre-event week, this compares significantly to the one-tenth of Buchanan’s coverage which only averaged 188 words. These differences suggest that the directional variables should include some reflection of the amount of coverage the candidates received since this is assumed to potentially make a difference in the assessments of potential voters. Moreover, it is assumed that it is the relative activities and characteristics of the candidates that drive the news media’s directional coverage of them.______
5.8.2 National Directional Coverage
Just as the attention that the national media pays the candidates influences the attention the local press gives the candidates, so too might we expect that the relative assessment of the candidates in positive or negative terms by the national media would be somewhat reflected in the local press, particularly prior to a contest. This influence may be offset by local circumstances and candidate activities in the locale, but the expectation is that if the national media pronounces a candidate doing very well or quite poorly, this information will be reflected in the local press to a certain degree at that time point.
211 Data presented in Table 5.13 suggest that this is the case. National positive and negative coverage percentages are correlated with local positive and negative percentages.
The linear relationship between national and local positive coverage levels is the stronger at .56 (.01 two-tailed). Looking at the candidate variations we find some interesting relationships.
Table 5.13: The Relationship Between National Directional Coverage and Local Directional Coverage
By Candidate National Negative w/ National Positive w/ Local Negative Local Positive Brown n.s .44 (.05) Clinton n.s. .60 (.01) Haridn -.59 (.05) n.s Kerrey n.s n.s Tsongas n.s .55 (.02)
Bush n.s .43 (.03) Buchanan n.s. .39 (.06)
Note: Table entries are bivariate correlations with statistical significance in parentheses. The entry n.s. means the correlations were not statistically significant at the .10 or less level.______
The candidates who were fi'ontrunners or runner-ups at some point in the campaign each have moderate correlations between prior national positive coverage and local positive coverage, with Clinton having the strongest relationship. Part of this is explained by the general nature of positive coverage - it is often campaign- (relating to finances, organization, strategy) and game-related. No candidates fi’om the Democratic campaign had positive relationships between their national negative and local negative coverage percentages.
212 Harkin actually had a negative relationship between his national negative level and his local negative level of coverage. As his national negative portions rose, his local negative levels declined. Some of this explained by the fact that as it became more apparent that Harkin was not going to be able to continue, his coverage at the national level declined in general, as did his local coverage, but his national coverage negative levels increased in describing the failings of his candidacy, whereas the local media examined mentioned his potential dropping out fleetingly and talked more about the implications of a three person race, who the labor vote would go to, and so forth, i f they covered him at all. In other words, they (the local media in general) put less emphasis on analyzing why Harkin came to his end, and more on what was going on, particularly since they had an event of their own coming up, in terms of present competitive activity. Again, it is worth mentioning that the national media focus on a more global picture of the campaign and the state media focus more on their island o f activity and its inhabitants and visitors prior to an event. This is one of the practical explanations for the variation that sometimes exits in national and local negative pre-event coverage.
One of the reasons we find candidates, particularly fi’ontrunners (Tsongas, Clinton and Bush) receiving higher levels of negative coverage is that as they become more viable their general level of coverage increases, and thus proportionately their negative and positive coverage increases as well. The nature of the coverage may be different — positive coverage relating to gaining delegates and the nomination, winning by substantial amounts in relatively unchallenged contests; and the negative coverage may be more substantive relating to character or experience or relate to the upcoming general election
213 or be generated by that last challenger that the party is trying to convince to drop out of the race.
5.8.3 Candidate Attention
When a candidate pays a high level of personal attention and invests his time and effort in a state, that candidate is expected to receive a higher level of positive media coverage than when he does not. Table 5.14 suggests that this hypothesis can be confirmed in a bivariate analysis for the Democratic campaign.
Candidates such as Clinton, Harkin, Kerrey, Buchanan, and Brown appeared to benefit in terms of their allotment of positive local coverage if they were a presence.
Tsongas and Bush did not see differences in their level of positive coverage based on the personal attention they gave a state. Negative coverage was not related to candidate attention. Candidates are not “punished” if they did not visit the state.
214 Table 5.14: The Relationship Between Candidate Attention and Local News Outlets
Democrats Mean Std. Dev. Cases Moderate to Low Attention 16.7434 23.841 43 High Attention 37.0824 27.4973 36
F=I 1.9257 Sig.=.OOOI
Republicans Mean Std. Dev. Cases Moderate to Low Attention 32.7152 39.7744 17 High Attention 39.5892 38.0880 18
F=.2422 Sig.=.6259
Candidate Means High Attention Low to Moderate Attention Brown 16.56 7.90 Clinton 51.20 38.50 Haridn 47.98 14.96 Kerrey 55.20 22.41 Tsongas 26.74 26.52
Buchanan 22.43 14.97 Bush 57.28 52.99
Note: Table entries were generated using an analysis of variance.
5.8.4 Home State and Home Region
As to whether or not home state or home region media are biased in favor of candidates with roots in their area (See Table 5.15), we find that Bush received little extra favoritism fi-om one of his home bases - Maine. Although Bush did sweep the Maine caucuses, which were not competitive, there was not a great deal of pre-event coverage.
There was plenty of coverage of the nomination campaign itself^ but Maine’s Republican caucuses were not “news”. Buchanan received only 3.7 percent of the caucus vote. Over the entire course of Maine’s campaign coverage fi"om December 1991 to June 3rd, much of the campaign coverage was pulled from wire and news service reports. Few stories
215 about Bush or Buchanan were locally written. In fact, George Mitchell, one of Maine’s
Senators, appeared in more locally written front page stories, compared to Bush or
Buchanan, when he became a potential vice presidential contender."
The Democrats, however, were easily identified by region. Tsongas was the northeastern candidate; Clinton the southern candidate; Harkin and Kerrey the
Midwesterners; and Brown was the candidate of the west. Each had served or was currently serving the state as a political oflBce-holder - either governor or senator.
Candidates are expected to have higher levels of positive coverage from media outlets in their home state or home region, although this may be conditional somewhat on where they rank in the overall competitive hierarchy. For example. Brown was well liked in
Oregon, but by May, Clinton had pulled ahead so substantially in the number of delegates won that Brown did not see much positive coverage, or much coverage generally, for all the effort he put forth and all the good will that Oregonians felt towards Brown.
While there is a positive and statistically significant correlation between home state, home region and positive coverage share, the relationship is not very strong. Based on an examination of the three home state outlets in this sample, what appears to occur is that home state presses will cover their home state candidate to a higher degree, but this does not mean that their coverage will be positively biased toward that candidate. Indeed,
Boston Globe for Tsongas and the Des Moines Register for Harkin were among their toughest critics.
" if we examined the Washington Post as a local paper, we would see that it gave no favoritism to Pat Buchanan who claims Washington, D.C. as his home.
216 Table 5.15: The Relationship of Home State and Home Region and Local News Outlet Positive Coverage
Local Positive 2 Weeks Prior Local Positive I Week Prior Republican Home Region n.s. n.s. N=35 Home State n.s. n.s. N=35
Democratic Home Region n.s. .22 (.02) N=79 Home State n.s. .19 (.04) N=79
Note; Table entries are bivariate correlations with statistica] significance in parentheses. The entry n.s. means the correlations were not statistically significant at the .10 or less level. ______
5.8.5 Performance Factors
One relationship that is fairly strong, at least in a bivariate comparison, is that of positive and negative coverage with performance factors. (See Table 5.16.) For the
Democratic candidates, momentum and viability are both positively correlated to a candidate’s positive local coverage. Viability, however, has the stronger linear relationship. This is due mainly to the local media, on the cue of the national media, reporting on the wins (momentum) of candidates and the progress towards winning enough delegates to gain the nomination (viability). This focus is often what makes up the bulk of positive coverage.
We do not see, however, a relationship between viability and negative pre-event coverage for the Democrats. Rather than seeing candidates get increased negative coverage when they have done poorly, we see them ignored by the press, a hazardous situation for any candidate. Increased viability can bring, as mentioned earlier, more
217 scrutiny by the media and more criticism by opponents, but the local pre-event coverage presented here does not give support to premise for coverage immediately prior to an event. While evidence in other studies suggests that frontrunners are covered more harshly as th^r become more successful (Hagen 1992), local pre-event negative coverage does not seem to be systematically affected.
Table 5.16: The Relationship Between Candidate Performance and Local Directional Coverage
Democrats Local Positive Coverage 1 Week Local Negative Coverage I Week Viability .51 (.01) il s. Momentum .26 (.01) n.s. N=79
Republicans Viability n.s. .32 (.06) Momentum .39 (.05) .31 (.06) N=35
Note: Table entries are bivariate correlations with statistical significance in parentheses. The entry n.s. means the correlations were not statistically significant at the .10 or less level.______
The Republican candidates viability scores were not correlated with local positive coverage during the pre-event periods examined. Rather, momentum has a moderate correlation with local positive coverage. This suggests that George Bush’s momentum value going into a state’s primary was associated with local positive coverage. The bigger his win in the previous event, the more likely he would receive more positive local coverage. Viability and momentum appear to have some relationship to local negative coverage for the Republican candidates. For Bush, while momentum appeared to bring
218 positive coverage it was accompanied by negative coverage as well. As to viability, here
the relationship is more complex. Buchanan’s viability level rose from 0 to 11.9 percent of
half the delegates needed for nomination during the period examined. Bush’s viability
scores rose from 0 to 100 percent. He actually passed the halfway point in April. As
Buchanan’s viability increased his negative coverage actually decreased on average, while
Bush’s negative coverage remained rather high throughout, and marginally increased at the end of campaign when Buchanan was no longer receiving coverage. For local news outlets, the relationship between performance factors and positive and negative coverage is not a strong one.
5.8.6 Candidate Field Size
Candidate directional coverage may also be affected by the number of competitors in the race. Table 5.17 illustrates what occurs to the local media’s positive and negative distributions of coverage when candidates drop out. As the number of competitors decreases there is an increase in the level of positive coverage, although the relationship is somewhat weak. The relationship between negative coverage and competing candidates is even weaker, and is in the other direction. In 1992, as the number of candidates decreased from five to two, the positive coverage levels tended to increase. This appears to be due to the local media’s positive coverage being more dispersed when there are multiple candidates. As candidates drop out, this means some other candidates are doing better - usually the frontrunner(s) in their performance and thus in their positive coverage.
219 The reason we see negative coverage decline in general for most candidates left in the race, is that frontrunners receive more pre-event negative coverage, both by the local media and the national media, leaving the other competitors with less coverage, but less negative scrutiny as well.
Table 5.17: The Relationship Between Candidate Field Size and Local Directional Coverage
Local Positive Week oFEvent Local Negative Week of Event Candidate Field Size -.25 (.01) .19 (.05) (2 through 5)
N=79
Note: Tabic entries are bivariate correlations with statistical significance in parentheses. The entry n.s. means the correlations were not statistically significant at the. 10 or less level.______
5.8.7 Scandal
It is also expected that the outbreak of a scandal during the week of an event may increase the candidate’s negative coverage level by the local media. Table 5.18 suggests that rather than increasing that candidate’s share of the negative coverage, we see instead a decrease in the positive coverage allocated to that candidate. But this relationship is very weak, which sheds some light on the often disbelieved fact that even though Clinton was riddled with scandalous stories, they did not seem to make their way into as much of the press coverage as has been assumed.
220 Table 5.18: The Relationship Between Scandal and Local Directional Coverage
Local Positive Week of Event Local Negative Week of Event Scandal -.14 (.10) as.
N=79
Note: Table entries are bivariate correlations with statistical significance in parentheses. The entry n.s. means the correlations were not statistieally significant at the .10 or less level.______
5.8.8 Local Competitive Status
One obvious point is that local coverage is driven by state as well as national
politics. One effect of the state-by-state nature of the nomination campaign is that a
candidate who is not the national frontrunner or even the runner up can be a state
ffontrunner. This is particularly true early in the campaign when no candidate has garnered a substantial number of delegates.
We know that being the national ffontrunner can bring increased negative scrutiny by the national press, and indirectly these evaluations of the national press trickle over to the local press. More indicative of how much negative and positive coverage a candidate receives is whether or not that candidate is the local frontrunner or challenger. The Argus
Leader, one of South Dakota’s dominant dailies, focused much of its attention and criticism on the candidate leading in the state polls - Bob Kerrey. In fact, he received a 61 percent share of the negative coverage produced by the Argus Leader during the week of the South Dakota primary. Table 5.19 shows that the association between one’s state-level competitive status and one’s level of negative coverage is rather strong (.59, .01 two- tailed). The correlation with positive coverage is somewhat less, but still substantial.
221 Table 5.19: The Relationship Between State-Level Competitive Status and Local Directional Coverage
Local Positive Coverage I Week Local Negative Coverage I Week Democrats Only State Competitive Status .49 (.01 ) .59 (.01)
N=79
Note: Table entries are bivariate correlations with statistical significance in parentheses. The entry n.s. means the correlations were not statistically significant at the. 10 or less level.______
5.9 Toward a Model of Local Directional Coverage
Using the bivariate indications of factors associated with a candidate’s levels of positive and negative coverage, a multivariate model of positive coverage share and negative coverage share can be determined for both the Democratic and Republican campaigns.
The general model of local positive coverage share for both campaigns is specified to be a function of prior national media positive coverage share, personal attention paid to the state, home state affiliation, viability, momentum values, the number of competing candidates, competitive status in state (state poll leader and runner-up, if close) and scandal (as a negating influence) and incumbency deference. Table 5.20 provides the results fi'om an ordinary least squares regression of these factors on candidate local positive coverage share.
For the Democratic campaign, the model predicts close to 60 percent of the variance in positive coverage share. The model also suggests that positive verdict is
222 Table 5.20: Model of Local Positive Coverage: Democratic and Republican Campaigns
Denocratk Candidates Variable B SEB Beta Sie. of T National Pos. Coverage .22 .06 .368 .0002*** Candidate Attention 11.60 4.46 .208 .0223** Home State 34.94 14.52 .205 .0188** Momentum -2.10 2.40 -.080 .3914 Viability .82 .24 .415 .0011*** Candidate Field Size 4.67 2.71 .189 .0897* Scandal -6.17 10.82 -.050 .5701 Loc. Compel Status 12.65 5.40 .229 .0221** (Constant) -17.85 13.06 .1761
Adjusted R Square: .50 SE of Estimate: 19.01 N=79
Republican Candidates Variable B SEB Beta Sie. of T National Pos. Coverage .66 .29 .377 .0307** Candidate Attention 1.83 13.18 .022 .8904 Home state 2.14 38.38 .009 .9559 Viability -.10 .20 -.099 .6067 Momentum 2.89 3.52 138 .4184 Incumbent 29.07 14.67 .378 .0574** (Constant) -2.05 12.36 .8696
SE of Estimate; 33.27 Adjusted R Square: .27 N=35
Note: Scandal, local competitive status and competitive field are constant for ftie Republican campaign. Candidate field and local competitive status are constant for the Republicans. Dependent variable is local positive pre-event coverage. Table entries are ordinary least squares regression estimates. ***=<.01, **=<.05, *=<.10 ______
somewhat dictated by the national media. Candidates who are receiving positive coverage from the national media will see it matched, while not on a one to one ratio, but rather substantially by the local press. For example, if Clinton were to receive a 40 percent share of the national media’s positive coverage, we would see this generally translate into 26 percent of a positive share at the local media level. Moreover, if he gave the state a great
223 deal of personal attention, outcourting all of his rivals, his local coverage level would rise
12.6 percent. However, should a personal scandal emerge the week of this event, his positive coverage level would decline by a hefty 17 percent, but if he were the local frontrunner or competitive runner-up, he could expect an additional 14 percent positive share of the local coverage.
As to the Republican campaign, the ability of the model to predict is weaker. A model of positive coverage share for the Republicans includes all the variables in the general model although three drop out due to lack of variance. This model predicts only
27 percent of the variance in positive coverage share. Buchanan generally did not receive much positive coverage. His base level of positive coverage when the effects of the model’s variables are held constant is close to zero. The estimate for “incumbent” captures
Bush’s general level of positive coverage share - approximately 29 percent. As Bush’s positive national coverage increased, so did his local coverage. Much of Bush’s positive coverage dealt with issues of viability, support, and contest outcomes. And while in absolute terms his positive coverage was actually quite low, relative to Buchanan his shares were high. The results also suggest that the local media’s assessment of the
Republican campaign was driven primarily by the national media and by Bush’s position as incumbent president. At least 60 percent of the positive coverage levels for Bush were at
75 percent or better suggesting some similarity in the dependent variable. The lesser variance makes estimating the model a more difficult task and may account for the lower adjusted R-square.
224 As to candidate local negative coverage share, a model consisting of candidate
field size, local competitive status, incumbency, scandal and viability predicts only 32
percent of the variance in the dependent variable for the Republicans and 33 percent for the Democrats. (See Table 5.21) In the Republican model, only one of the explanatory
variables is statistically significant - incumbency. The model suggests that the Republican candidates started out with a higher intercept, 26.5 percent of a negative share. For the incumbent, we can add another 22 percent for a total of 48.4 percent.
The model predicts only a moderate amount of the variation in the candidates’ negative coverage. Part of the explanation for this may be found in the variance of the dependent variable. Over 50 percent of the local outlets gave Bush 80 percent or more of the negative coverage shares during the week prior to their event. Buchanan averaged 20 percent in half the outlets included in the analysis. The variable incumbent captures this repeated pattern of high negative coverage for Bush and lower negative coverage for
Buchanan. The range of variation for the other variables in the model is also strongly associated with the incumbent and challenger dichotomy. Bush enjoyed more movement in his viability. He was the only candidate to win contests, thus depriving Buchanan of any momentum.*'
'^Other regression analyses suggest that prior to Super Tuesday, the Republican campaign was more fluid. During this period. Bush’s less than expected performance in New Hampshire increased his negative coverage levels. As he won Maine, South Dakota and so on, his negative coverage levels decreased incrementally. At this point in time, the fact that he was an incumbent was not the critical factor, but that he was shaken by Buchanan. Moreover, at this time period, Buchanan was receiving more negative coverage than he woifld afler Super Tuesday, when he was no longer considered a factor in the campaign. After Super Tuesday, these minor fluctuations in viability and doing less than expected were no longer relevant since Bush won by large nutrgins leaving Buchanan far behind. Regressions of cases falling in this time period which include the same variables as the model reported on the next page, suggest that a control for the incumbent president has the strongest predictive ability for local negative coverage. Moreover, for this time period, the model explains 44 percent of the variance in local negative coverage.
225 As to the Democratic campaign, while a similar level of variance is explained, the standard error of the model is much less. Yet, only local competitive status appears to have a systematic relationship with local negative coverage. Local frontrunners (which may or may not coincide with national frontrunners) are more likely to gamer higher levels of negative coverage. Momentum is almost significant and in the correct direction. It would appear that winning prior to a state’s primary or caucus could potentially negate the amount of negative coverage a candidate receives. Surprisingly, having a scandal break during the week of a primary or caucus had no influence a candidate’s negative coverage.
Table 5.21: Model of Local Negative Coverage: Democratic and Republican Campaigns
Democratic Candidates Variable B SEB Beta Sign, of T
Candidate Field Size 2.49 3.23 .097 .4422 Scandal .04 12.32 .000 .9974 Loc. CompeL Status 36.06 5.97 .629 .0000 *** Viability .12 .28 .063 .6494 Momentum -4.44 2.79 -.166 .1153 (Constant) -1.38 15.02 .9270
SE of Estimate: 22.88 Adjusted R Square: .3734 N=79
Republican Candidates Variable B SEB Beta Sign. oFT Incumbent 47.76 13.76 .594 .0016*** Viability -.02 .19 -.019 .9103 Momentum 1.58 3.48 .072 .6520 (Constant) 26.50 8.37 .0035
SE of Estimate: 32.40 Adjusted R Square: .3227 N=35
Note: Local competitive status, competitive field and scandal are constant Table entries are ordinary least squares regression estimates. ***=<.01. **=<.05, *=<.10 ______
226 Momentum is the only variable that appears to have some potential for negating
local negative coverage levels. The implication is that candidates who are entering the
campaign week with a win from the prior week, may have less negative coverage in the
state’s local newspaper.
5.10 Factors Influencing National Coverage
Having determined, to a degree, the relationship between candidate-related
performance, resource and attention variables, as well as competitive structure and other
candidate characteristics and local media coverage, I now turn to national coverage.
National media coverage is assumed to be explained by “national” factors. For example, a
news media outlet covering the campaign for a national audience is expected to pay close
attention to the overall performance of the candidates. Viability and momentum are
assumed to be positively related to national media coverage levels. The correlations
reported in Table 5.22 suggest that for both the Republican and Democratic candidates,
the general relationship of performance factors was similar. Momentum had a moderate
and positive linear relationship with national media coverage share. Viability, on the other
hand, had a very strong linear relationship with national coverage share.
Another assumption regarding national media attention is that frontrunners and
incumbents (who are generally frontrunners throughout) gamer the most attention from
the national media. The analysis of variance results reported in Table 5.23 suggests that this is the case.
227 Table 5.22: The Relationship Between Performance and National News Outlet Coverage
Democratic Candidates National Coverage Share Momentum Value .46 (.01) N=79 Viability .83 (.01) N=79
Republican Candidates National Coverage Share Momentum Value .48 (.01) N=35 Viability .78 (.01) N=35
Note: Table entries are bivariate correlations with statistical significance in parentheses. The entry n.s. means the correlations were not statistically significant at the .10 or less level. ______
Table 5.23 National Coverage Differences Between the Incumbent, Frontrunners, and Challengers
Republicans Mean Incumbent 86.29 Challenger 12.80
F=807.2196 Sig.=.0000
Democrats Mean Frontrunner 55.97 Challengers 19.85
F= 129.6925 Sig.=.0000
Note: Table entries were generated using an analysis of variance.
Utilizing only these three variables, the models can predict 61 percent of the variance in the national media’s attention levels to the Democratic candidates and 94 percent of the variation for the Republicans.
2 2 8 The model estimates for the Democratic campaign suggest that for every increase in viability (the percentage of the number of delegates necessary to win nomination a candidate has won up to that point) the candidate received .75 percent of the national media coverage share. If a candidate garnered 20 percent of delegates needed to win, we could expect that candidate to receive 15 percent of the national coverage share.
Achieving the status of frontrunner, brings with it an additional 12.3 percent of the national media’s attention. Momentum, surprisingly, does not appear to influence national coverage systematically, although sharp rises in viability may mute the effect of winning.
As to the Republican campaign, a similar scenario is seen. As viability increases, so does the share of national media attention. And the incumbent, in this case Bush, could expect 72 percent of the national coverage given to the campaign, while Buchanan’s average is captured by the intercept’s value of 11.47 percent.
Viability is an important indicator of national media coverage levels for both candidates competing in a traditional multi-candidate campaign, as well as one featuring a challenged incumbent. Since viability is a the sum of success in a number of state primaries, we must ask ourselves if media coverage plays a role in influencing these vote outcomes. If it does, then we have a situation where gaining delegates leads to increased national media attention, which influences local media coverage in part, which then influences the vote outcome. For this reason, we may need to turn to dynamic models which capture the full effect of media coverage.
229 Table 5.24: Model of National News Coverage: Democratic and Republican Campaigns
Democratic Candidates Variable B SEB Beta Sipi. of T Viability .75 .06 .651 .0000 * * * Momentum 1.25 1.07 .057 .2424 Frontrunner 12.30 2.97 .232 .0001* * * (Constant) 16.05 1.33 .0000
SB of Estimate: 13.21 Adjusted R Square: .72 N=79
Republican Candidates Variable a SEB Beta Sim. of T Viability .21 .02 .268 .0000*** Momentum -.75 .71 -.032 .2936 Incumbent/Frontrunner 60.84 2.90 .784 .0000*** (Constant) 11.47 1.48 .0000
SE of Estimate: 9.0073 Adjusted R Square: .93538 N=35
Note: Table entries arc ordinary least squares regression estimates. **=<.05. *=<.10 ______
5.11 Possible Factors Influencing the Direction of National
Media Coverage of Candidates
Just as local coverage has its positive and negative dimensions, so does national coverage. What factors appear to influence the national media’s allotment of negative and positive coverage to the candidates? Are the factors the same ones that influence local media verdict? Do they have a similar degree of influence?
230 5.11.1 Incumbency
Theoretically, incumbents are hampered when they challenged from within their own party. This is an indication of potential factiousness in the party or a lack of responsiveness from the incumbent president to his party’s various sectors. A challenge may be seen as simply arising when an incumbent appears weakened. An ambitious opponent might utilize the nomination campaign to lay down a groundwork for future presidential aspirations or to position him or herself as an alternative party leader. When an incumbent is challenged we can generally expect the national media to apply their analytical expertise to the situation. The challenge itself is newsworthy and potentially indicative of problems in the incumbent’s campaign or within the party itself. Even if the president’s renomination is thought to be a foregone conclusion, the presence of an antagonistic opponent may heighten the critical light in which the incumbent’s performance is reviewed in retrospect.
Moreover, an incumbent must deal with the context of his term in office - what has he done for the American people? The media, like the voters, can assess the incumbent’s performance retrospectively as well as prospectively. Table 5.25 illustrates that incumbents are covered more negatively than their challengers to a significant degree. It is interesting to note that several occasions during the 1992 nomination campaign, Buchanan reflected on his negative portrayal in the media. Observing Buchanan’s coverage, however, suggests that very little of his coverage related to his ideas on women, immigration, protectionism and so on. Rather, the astounding fact is that there was relatively little negative coverage of Buchanan. Outside of a few stories and editorials,
231 most media outlets seemed to shy away from “Buchanan bashing.” Bush, by far, received
higher levels of negative coverage. And when the absolute numbers are compared across
all candidates. Bush is decidedly given more negative coverage.
Table 5.25: National News Media Outlet s Negative Coverage of Incumbent President Bush
Republicans Mean Std. Dev. Cases Incumbent 80.91 7.62 19 Challenger 11.29 7.66 16
F=721.74 SigofF=.OOOG
Note: Table entries were generated using an analysis of variance. ______
5.11.2 Performance Factors
How well a candidate does is intuitively associated with negative coverage. When
a candidate does well, we assume more positive coverage, when they do worse, we expect
more negative coverage. Table 5.26 presents the results of the performance variables with
national positive and negative coverage.
As we see, momentum and viability are correlated with positive coverage as
expected. The relationships are more substantial for the Democratic candidates, but still
moderate for the Republicans. As to negative coverage share, rather than an inverse
relationship (as viability increases negative coverage decreases) we see a direct one - as viability and momentum level increase, so too does negative coverage. This again is
supported by the observations that when a candidate becomes a member of the top tier of
232 candidates — as the frontrunner or runner up and begins to accrue delegates and momentum by winning state primaries and caucuses — the national media pay more
Table 5.26: The Relationship Between Candidate Performance and National News Media Outlet Directional Coverage
Democratic Candidates National Positive Coverage Share National Negative Coverage Share Momentum Value .46 (.01) .20 (.01) N=79 Viability .64 (.01) .46 (.01) N=79
Republican Candidates National Positive Coverage Share National Negative Coverage Share Momentum Value .34 (.02) .46 (.01) N=35 Viabüity .29 (.01) .56 (.01) N=35
Note: Table entries arc bivariate correlations with statistical significance in parentheses. The entry n.s. means the correlations were not statistically significant at the .10 or less level.______
attention to those candidates. The national media analyze their background, experience and current campaign, sometimes within a critical framework. In addition, candidates sometimes increase their viability scores and momentum scores prior to dropping out of the campaign. Kerrey, for example, won South Dakota (adding delegates and momentum and a surge of national media attention) only to drop out shortly thereafter. But this is a rare event. Generally, candidates who are doing poorly are ignored by the national media as soon as their viability is ascertained to be quite dubious. This happened to Harkin despite success in a number of state caucuses. Immediately after New Hampshire, Harkin sunk in the level of attention the press gave him nationally.
233 Candidates like Harkin receive little pre-event coverage (mostly a few words regarding their share of a recent vote) and often there is some negative dimension to the coverage (financial problems, no coherent message, lack of support, and internal campaign problems.) Again and again, however, there is evidence that doing better brings with it critical judgement by the media triggered not only by their assumed role of ‘watchdog’ which in normative terms is a critical part of our democratic process; but also because as one does better, one’s opponents become desperate and begin to turn negative. Tsongas had to resort to this when it became apparent that Clinton had taken over the role of frontrunner.
S. 11.3 Competitive Field
The structure of the competition among candidates appears to have a moderate relationship with directional coverage. The number of candidates participating in the competition is directly related to positive coverage and inversely related to negative coverage as illustrated in Table 5.27. The larger the number of candidates competing the more likely their positive coverage shares will be greater than if fewer candidates are competing. In other words, when their are many competitors, generally at the outset of the campaign, the national media are careful to partition out their positive coverage. This takes place rather naturally since the early contests, particularly in 1992 saw a number of candidates doing well in particular state contests. When the number of candidates begins to decrease due to wirmowing, we see the national media shift their positive weighting somewhat and concentrate on those who are left in a somewhat critical fashion.
234 Table 5.27: The Relationship Between Candidate Field Size and National News Media Outlet Directional Coverage
Democrats Onlv National Positive Coverage Share National Negative Coverage Share
Competitive Field .46 (.01) -.26 (.01) N=79
Note: Table entries are bivariate correlations with statistical significance in parentheses. The entry n.s. means the correlations were not statistically significant at the .10 or less level.______
5.11.4 National Polls
The polls take the pulse of American politics. In attempting to establish which
candidates appear to be resonating with the national public the national media pay
attention to these polls and utilize them in their “calling” of campaign race. Table 5.28
suggests that the national poll standing of a candidate is moderately correlated with positive coverage and strongly correlated with negative coverage (again positively rather than negatively). The strength of the association between national poll standing and
positive coverage share is very similar for the two parties’ candidates. However, national poll standing is very strongly associated with negative coverage share, particularly for the
Republican candidates. Again, this is a spurious relationship. In 1992, the incumbent president was covered far more heavily than his challenger. In general we might assume that incumbents presidents who are challenged received more negative coverage. In 1980,
Jimmy Carter’s treatment by the press was thought be clearly negative given his within party challenge, his loss of support among Democratic members of Congress and the hostage crisis that emerged. And even though he received a great deal of negative media,
235 he still won the nomination rather handily. Nelson Polsby states, “The Democratic
nomination of 1980 was won by the incumbent president, principally because he was the
incumbent.” (P.37) Polsby goes on to suggest that incumbency may overshadow other
factors in relation to gaining the nomination, but the context of the campaign may be
defined by retrospective review of the incumbent’s administration. In 1992, Bush’s
position garnered him those benefits of party organization, ample funding, endorsements
firom party regulars and a great deal of media exposure. However, given the context of the Bush administration and particularly the latter time period, there was more ample fodder for the news media to analyze and criticize. Moreover, having spent time in the
White House, Bush developed a policy portfolio that his opponents both of his own and the other party could spin in directions positive to themselves.
Table 5.28: The Relationship Between National Poll Standing and National News Media Outlet Directional Coverage
Democratic Candidates National Positive Coverage Share National Negative Coverage Share NaU. Poll Standing .42 (.01) .62 (.01) N=79
Republican Candidates National Positive Coverage Share National Negative Coverage Share Natl. Poll Standing .46 (.01) .94 (.01) N=35
Note: Table entries are bivariate correlations with statistical significance in parentheses. The entry n.s. means the correlations were not statistically significant at the .10 or less level.______
236 5.11.5 Scandal
As we have seen, scandal occurring in a given week appeared to have little eflfect on a candidate’s local negative share, but did decrease his positive coverage. What is the nature of scandal and national directional coverage? We see from the results in Table 5.29 that the national media in 1992 were more likely to cover a candidate more harshly if a personal scandal emerged that week. However, the relationship is a relatively weak one for negative coverage and non-existent for positive coverage.
Table 5.29: The Relationship Between Scandal and National News Media Outlet Directional Coverage: Democrats Only
Democratic Candidates National Positive Coverage Share National Negative Coverage Share Natl. Poll Standing n.s. .24 (.01) N=79
Note: Table entries ate bivariate conelations with statistical significance in parentheses. The entry n.s. means the correlations were not statistically significant at the. 10 or less level.
5.12 Toward a Model of National Directional Coverage
The bivariate analyses of directional coverage and a host of structural, performance and other candidate related variables suggest that the national media’s direction of coverage may be estimated with variables which capture the competitive field size, momentum, viability, incumbency, national poll standing and scandal. Table 5.30 displays the results of a multivariate model of the Democratic and Republican candidate’s positive coverage shares.
237 The Democratic candidates’ share of positive coverage the week prior to an event
is predicted within 14 percentage points on average by this model. The estimates suggest that the larger the number of candidates competing the lesser the overall shares of positive coverage. Momentum brings with it a burst of positive coverage. Winning a big event, such as the multiple contests of Super Tuesday, can bring a significant boost of positive coverage from the national media. The larger the number of contests won, the greater the positive coverage level.
Viability does not reach statistical significance. However, reaching the halfway point of viability (fifty percent of the required delegates) makes a big difference for the candidate who achieves it and his opposition. For example, Clinton’s positive coverage after New York’s primary was very high. During those first two weeks of April, Clinton hit the 50 percent mark (the halfway point).He had accumulated over 50 percent (a majority) of the delegates needed to win the nomination. At this point, candidate positive coverage levels diminished for Brown by 32.9 percent and increased for Clinton by 42.65 percent. With Brown effectively out of the race, anything positive that appeared in the national media would be owned by Clinton.
Scandal does not appear to have the same negating eflfect on positive coverage levels for the national media as it does for the local media. The eflfect of a breaking scandal appears to increase positive coverage. This is probably a spurious relationship given that
Clinton’s post-Flowers scandals emerged as he was doing better on the campaign trail.
*^The halRvay point is a dummy variable that designates that a candidate has reached the 50 percent of delegate point All candidates still in the race receive a value of one for this time period. Cases occurring before the halfway point has been reached are coded zero.
238 Nevertheless, scandal did tend to increase negative coverage almost three times as much.
Positive movement in a Democratic candidate’s national poll standings also
increased positive coverage share. As national polls indicated an increase in general support for a candidate so did national positive coverage increase. For every one point in poll upward movement, a candidate received on average .28 percentage increase in positive coverage relative to competitors.
The Republican campaign’s model generated an explained variance of 56 percent.
A candidate’s national poll standing was inversely related to the share of positive coverage the candidate received. This suggests a five point national poll increase would generate a one percent decrease in positive coverage. Of course, this relationship is not thought to be a causal one, but rather another spurious relationship. For Bush, as his shaken poll numbers steadily inclined, Buchanan was no longer foil for comparison of vote outcomes, so little coverage related the large victories Bush won in May and June. Rather, he recovered somewhat in the polls, and the national news media began to focus on the Perot versus Bush contest. Moreover, there was already negative discussion of the Republican
National Convention and possible ramifications of the anti-Bush movement, and Bush’s issue stances were taking critical attention fi’om the press and attacks fii’om both Clinton and Perot.
Viability, in general, had a direct relationship to national positive coverage share, but did not reach statistical significance. Momentum boosted a candidate’s national positive coverage share. When Bush reached the halfway point in accumulating delegates.
239 Buchanan’s positive coverage diminished while Bush’s increased for similar reasons
discussed for the Democrats.
Table 5.30: Model of National News Media Outlet Positive Coverage
Democrats Variable B m 3 . Beta Sign, of T
Momentum 3.80 1.57 .179 .0034*** Viability -.23 .22 -.211 .2802 Competitive Field -4.73 1.57 -.243 .0030*** Natl. Poll .28 .11 .205 .0083*** Scandal 9.99 5.18 .096 .0558* Halfway point reached -32.92 4.61 -.560 .0000*** Halfway x Frontrunner 75.57 10.62 .997 .0000*** (Constant) 36.77 7.03 .0000
Adjusted R Square .66 Standard Error 14.37 N=79
Republicans Variable B SEB Beta Sign, of T
Momentum 3.29 1.36 .227 .0179** Viability .19 .15 .377 .1898 Natl. Poll -.20 .09 -.263 .0445** Halfway point reached -45.78 5.33 -.916 .0000*** Halfway x incumbent 26.07 15.47 .479 .0956* (Constant) 51.21 5.09 .0000
Adjusted R Square .56 Standard Error 16.47 N=35
Note: Competitive field and scandal are constant for the Republican campaign. Table entries arc ordinary least squares regression estimates. ***=<.0U **=<05. *=<.10 ______
As to the model’s effectiveness of explaining negative coverage share, it is
moderate for the Democrats, but strong for the Republican campaign. The multivariate model applied to the Democratic campaign in which these variables were regressed on
240 candidate negative share is able to predict a candidate’s negative national coverage share within 16 percent. The model accounts for 49 percent of the variance in negative coverage levels. The appearance of a scandal in a given week appears to have increased the candidate’s negative coverage by 27 percent. In addition, national poll standing (a variable moderately correlated to viability) is directly related to negative coverage share.
Again, this could be explained by a concurrent phenomenon rather than a causal one.
Clinton’s national poll standing rises mirror not only his viability increase, but also his increase in national media attention. In other words, as he became the viable frontrunner, he generally did better in the polls and received more coverage in general. When he received more coverage in general, he was also critically analyzed. The more coverage a candidate receives the more likely some directional coverage would be included. This is true for both positive and negative coverage.
Momentum is not statistically significant, but the direction of the estimate suggests that winning may negate negative coverage briefly as the media shifts to covering the positive outcome. In 1992, when Clinton reached the halfway point in the number of delegates needed to win the nomination, in general, the level of negative coverage decreased. However, the interaction between Clinton and halfway suggests that the net effect was a 5.26 increase in negative coverage. Thus, for Clinton, becoming the nominee apparent dampened the level of negative coverage, but did not do away with it completely.
The Republican’s negative coverage share is predicted well by the model which includes viability, national poll standing, momentum and an indicator for when the leading
241 candidate reached the halfway point in viability (50+ percent). An interaction to control for incumbent cases is included as well. The model suggests that increasing viability and poll standing were accompanied with increased negative share for the leading candidate.
This makes sense since as Bush came closer and closer to the nomination, the Buchanan story disappeared. With Buchanan no longer in the race, any negative assessment would be directed only at Bush.
Table 5.31: Model of National News Media Outlet Negative Coverage
Democrats Variable B SEBBeta Sign. ofT
Momentum .55 1.42 .123 .2056 Viability .11 .24 .106 .6486 Natl. Poll .73 .12 .574 .0000* * * Scandal 26.87 5.77 .280 .0 0 0 0 * * * Haltyvay x Frontruimer 17.06 11.82 .237 .1510 Halfway -10.80 5.13 -.199 .0308** (Constant) 5.54 7.83 .4801
Adjusted R Square .49 Standard Error 16.00 N=79
Republicans Variable B SEB Beta Sign, of T
Viability .34 .09 .457 .0005*** Natl. Poll 1.04 .06 .943 .0 000 * * * Momentum 1.33 .87 .063 .1318 Halfway -4.78 3.43 -.066 .1669 Half\vay x incumbent -42.93 9.95 -.545 .0000* * * (Constant) -4.52 3.27 .1703
Adjusted R Square .91 Standard Error 10.59 N=35
Note: Table entries are ordinary least squares regression estimates. ***=<.01, **=<.05, *=<. 10
242 When Bush reached the halfway point, his negative coverage appeared to decrease substantially, particularly since stories relating to the protest vote declined marginally due to the lack of negative horserace coverage. However, his negative coverage did continue, but it shifted to issue criticism as the general election neared. For example, the
Washington Post ran a number of stories in April through June, for example/'Critics fault
President’s Education Attention Span” (4/16), “MIA issue shaped Perot’s view of Bush”
(4/17), “Bush reasserts his urban ideas: I am Proposing Them Again: We Must Try something new” (5/9); “Reagan, Bush Blamed for Urban Decline” (5/14); “Bush Curbs
Clean Air Provision” (5/17). Clinton’s coverage somewhat paralleled this. As he continued to collect delegates, discarding Brown as an opponent after New York, the coverage he received from the national media altered their coverage almost on cue with the candidates desires. The Washington Post ran this fi’ont page story on April 19, “Seeking the
Presidential Clinton: Campaign will Return to Stressing Issues in Eftbrt to Engage Bush.”
While Perot managed to dominate news coverage by the national media in April and May,
Clinton did manage to utilize the Los Angeles riots to engage Bush on urban policy. And the television news featured stories on Clinton dealing with education and crime; the Wall
Street Journal discussed Clinton’s push for family values.
5.13 Pre-event Substance Coverage by the National and Local Media
Reformers of the news media often demand less horserace and more issue coverage. Evidence suggests that candidates are producing more commercials and speeches that are issue-oriented, yet the news media, particularly the national news report
243 the competitive aspects of the campaign in greater quantity. Thus, while it appears that
candidates and thus parties are more responsible, the news media may not be. Ironically, the analysis of substance coverage in the 1992 Republican and Democratic campaigns
suggests that voters may see more substance when there is more competition. A great deal of coverage of public policy issues occurred during competitive time periods such as prior to New Hampshire, Super Tuesday, Illinois and Michigan, and New York. (See
Tables 5.32 and 5.33.)
For the Republicans, national issue coverage was consistent throughout except for a period in late March/early April. When Buchanan was still a player. Bush’s substance coverage at the national level was heavy. When there was a competitive void. Bush’s policy coverage declined in the national media. When Perot became Bush’s foil in early
April, once again, issue coverage increased somewhat, but more importantly, a national event of such shocking significance, the Los Angeles Riots, created a context in which the incumbent president had to act and be evaluated. Table 5.33 illustrates the distribution of issue coverage of the Republican campaign. Most of the coverage was about Bush.
Buchanan’s issue coverage was very sparse in the national media. The Republican’s national issue coverage outweighs issue coverage at the local level, but local coverage was not far behind. Moreover, there was substantial issue similarity between the local and national media. The economy, budget and taxes dominated initially. Later, questions of urban policy, the Los Angeles riots, the environment and family values took precedence.
244 Table 532: Pre-event Issue Coverage of Democratic Candidates: National and Local News Media Outlets Compared
Week Caatest(s) Issues
yio u 189 heahh care, general domestic, taxes
2/18 NH Budcar energy. Haiti 86 economy, taxes, nuclear energy, education.
2/23 ME 138 taxes.
2/25 SO 678 ■gncuiture. health, energy
3/3 143 abortion, energy, taxes, economy, economy health, water, environment, message, homeless. Ibrergn policy
3/5 ND
3/7 SC.AZ.WY 50 drugs, economy, business
3/8 NV 78 nuclear energy, agnculture
3/10 97 economy, disabled, message, social 162 social welfarepolicy. taxes, welfare policy, racial policy. economy, environment, health
3/17 lUMI 315 taxes, economy, message. labor 43 environment, labor, energy, taxes
3/24 CT 23 general domestic policy
3/31 VT 359 taxes (flat tax)
4/2 AK 207 taxes, social welfare urban policy, gun control, foreign policy
4/7 KS. MN. MY. WI 33 drug policy 144 aboitioii. sm 4/11 VA 37 economy, abortion, foreign policy 4/27/28 PA 5/5 IN. NC, DC. Ill domestic policy generally 5/12 NE.VVV 112 jobs, urban policy, social issues in general 5/19 OR.WA 512 environment, health, domestic policy, economy education, abortion. AIDS, message 5/26 AR.KV 125 gqr rights, business, environment 6,2 MT. NJ. 43 Average 183.4 Average 73 words words Note: Cell entries are based on the total nunü)er of words devoted to issue coverage. When one event took place on the date givetutho local coverage cell entiv reflects the total coverage in words given to issues. Ifmore than one event took place on the dat& the ceU entry is an average across me state ouilets examined For the national outlets» cell entries represent the average issue coverage over the four outlets examined. 71» topics, however» are the range of topics covered. 245 Table 5.33: Pre-event Issue Coverage of Republican Candidates: National and Local News Media Outlets Compared Week Contest(s) Issues Issues 1 ^ 2/10 LA 439 economy, budget, foretgi afiairs, domestic 326 economy, budget, domestic policy, policy Totetgn affairs, trade 2/18 NH 196 econrrmy, budget, foreign aflairs, domestrc 98 health, taxes, economy, budget policy 203 ME 440 message, taxes, economy, budget 97 taxes, budget, ecooomy 205 SD 86 message, economy 74 economy 3/3 219 A^golicy. Iraq, taxes, environment, taxes. 377 domestic policy, economy, budget, racial pobcy, message 3/5 ND 190 economy, domestic policy, budget, --- faxes, message 3/7 SC,AZ,WY 73 economy. Japan/trade 291 economy, taxes, budget, environment 3/8 NV 152 taxes 286 economy, taxes, budget, errvironment 3/10 182 economy, immigration, trade, race, Cuba, 202 ectxtomy, budget Haiti, welhirc, ideology 3/17 IL,MI 264 taxes, economy, ideology 180 economy, foreign policy 304 CT ---- 112 economy, health, education, taxes, domestic policy 3/31 VT 0 40 AK ’ ' ’ 11 domestic policy, economy 4 0 KS, MN, WI 105 economy, b u d ^ taxes, domestic general 137 domestic policy, economy, message 4/11 VA 85 welfare, campaign finance reform 139 foreign aflairs 408 PA ----- 22 taxes 5/5 IN, NC, DC 390 Los Angeles riots, morality, message 277 Los Angeles nots 5/12 NE,WV ----- 802 social issues, Los Angeles riots, health care, domesuc policy 5/19 O RW A 184 economy, Los Angeles riots, domestic 413 Los Angeles riots economy, urban policy policy, environment 506 AR,KY,ID 151 family values 196 family values, foreign affairs, Los Angeles riots, etrvtronment 6,2 308 drug policy, taxes, abortron, environment Average 218.8 Average 716 words words Xote: Cell entries are based on the total number of words devoted to issue coverage. When one event took place on the date given, tho local coverage ceil entry reflects the total coverage in words giveQ to issues. Ifmote than otte event took place on the date, the cell entry is an average across the state outlets exatntrted. For the natrtma otrtlets. ceU errtnes represent the average issue coverage over the four outlets exartrtned. The topics, htrwever. arc the range of topics covered. 246 5.14 Discussion and Implications Local media coverage and national media coverage have similar requirements for determining the level of attention th^r give to the campaign and its participants. Candidate performance is a critical factor. The news media are more equitable in their coverage when there is not highly viable frontrunner. But when a strong frontrunner emerges, the coverage basically follows a formula that says cover the frontrunner and compare the others to the frontrunner in terms of their performance. If the challengers do well, they receive coverage. If they do this more than once, they receive more attention from the media. If they falter, they become the proverbial “old news.” One difrerence that emerges, however, is that the national media follow a relatively simple formula for determining the attention they will give a candidate, whereas the local media have a more diverse set of factors to contemplate. In 1992, the national media looked at two factors to determine how much coverage they would allocate to particular candidates: First, was the candidate the frontrunner or the incumbent? Second, how viable was the candidate? The local media (when covering a competitive race) have factors such as home state or regional connections, candidate activity and local competitive factors to consider, and they take into account the national media’s focus of attention. As to local coverage of the Republican campaign, the focus was on the incumbent once again having the amount, focus and tone of the coverage set by the national media. This was generally the case unless the incumbent paid special attention to their state’s contest, which Bush only did if Buchanan was sufficiently threatening in the state. Although Bush did send 247 numerous surrogates, spend substantial amounts of money and contribute various executive “pork” monies to various states at crucial time periods. We see larger differences between the national and local media in the area of how much directional coverage they allocated to a candidate prior to an event. For the Democratic campaign, a candidate’s local level of positive coverage was determined by the national media’s positive verdict, by how much attention the candidate gave the state, and whether or not the candidate was the local favorite. A candidate’s positive coverage could be hurt by the appearance of scandal in the national media. The national media’s positive verdict was determined primarily by a candidate’s performance. Had he won his last contest(s)? Was he accruing delegates? These factors basically are the foundation for positive coverage. On a purely structural note, the size of the field of candidates appears to influence the distribution of positive coverage as well. George Bush’s performance, both in the national polls and in the primaries and caucuses drove his positive national coverage. The national positive coverage, in turn, influenced the local coverage. One of the reasons we see such a strong linear correlation between national coverage of the Republican campaign and local coverage of the Republican campaign is that there was little candidate activity to influence the local news outlets. Buchanan did not have the financial resources to campaign everywhere and Bush had no need to do so. There was little activity on a local level to create a context in which the local media deviated fi’om the general campaign pattern set by the national media. As to negative coverage of the candidates, what determines who receives more negative coverage? At the national level, Clinton received more negative coverage through 248 the revelation of a series of “character” issue scandals ranging from marital infidelity, draft evasion, illegal loans and land deals, marijuana use, exemption of a state ethics law to his governor’s oflBce, to a golf excursion at an all white private club. Brown also encountered scandal dealing with the tolerance of drug use while he was governor of California. Scandal appears to influence a candidate’s level of negative coverage for the Democrats. For candidates who do not receive much coverage to begin with, however, it is doubtful that the news media, national or local, will generate positive or negative in general. As to local negative coverage, being a national and a local frontrurmer, or both simultaneously will lead to greater levels of negative coverage. Overall candidate performance is a critical factor in determining the level of candidate coverage. For the Democratic campaign, performance was also indicative of directional coverage. The national media are strongly influenced by a candidate’s viability and momentum. The local media are moved by this as well, for they follow the lead o f the national media somewhat. But the local media are also affected by candidates who invest time and money in their state. Candidates who are local frontrunners also gamer coverage. In many ways, the local media provide an informational conduit that may not be as inclined to winnow out lesser candidates as quickly as the national media. Although at a time point such as the halfway mark discussed earlier, it is doubtful than any but the frontrunners in either party as nominees apparent will be covered by the news media. In the days when there was more time between nomination contests, candidates could use this performance criteria to their advantage. Working hard in states to gain traction and momentum. Today, the frontloading of the contests may make victory for a 249 poorly financed candidate meaningless. Winning one state in Super Tuesday’s event does not match up to winning ten. In 1992, most candidates had little time to spend in individual states. Rather they had to concentrate their efforts many places at once. Candidates with greater financial resources could use paid advertising, more surrogate campaigners, higher profile events (bus tours, glitzy rallies, satellite town halls), and more campaign materials in general to blanket an area with the candidate’s name, message, and persona rather than relying on low cost events and personal interaction. Thus, the process favors certain candidates, and these candidates are then favored with higher levels of media coverage, often more positive levels of media coverage, but also greater levels of scrutiny. It is almost as though the media know they have helped create a frontrunner in a imperfect system and wish to administer test after test to assure that they have not made a terrible mistake. The implications of this behavior are that the news media can play a substantial role in reviewing the plans and policies of the candidates, if they so choose. In doing this they can do much to coalesce and solidify the supporters of these candidates as well as their opposition. Bruce Buchanan (1978) discusses a similar phenomenon in the relationship of the press and the president. Relevant to the incumbent presidential candidate, Buchanan suggests that the press can play a role in disciplining the president. He cites for example, “revelations of corruption and graft, television coverage of the Vietnam War, and investigative reporting such as that by Woodward and Bernstein.” (P. 118) In 1992, George Bush may have been disciplined by the news media. Bill Clinton certainly was showered with a barrage of revelations of scandal yet went on to win the 250 nomination. This implies that the news media’s ability to discipline is conditioned upon the voters’ response, which may be conditioned upon the choices that they are given. 251 CHAPTER 6 THE POLITICAL CONSEQUENCES OF NATIONAL AND LOCAL MEDIA COVERAGE While scholars and practitioners believe that the news media do matter, the issue in dispute is the nature and degree of tins influence and the question of whom among the various news agencies wields this influence. In this chapter, the influence of two types of media outlets, national and local, is compared. After the exploration of how the national and local news media vary in their coverage (Chapter 5), and determining what political factors drive their differences (Chapter 6), I am now in a position to address the question of whether the news media outlets discussed actually influence primary and caucus outcomes. In the presidential nomination campaign, the candidate attempts to win enough delegates to ensure his nomination at the party’s national convention. To accomplish this, each candidate must compete for delegates in all of the United States. Voters are in an odd position in that in a political process that is both national and local in nature. When the campaign begins in Iowa, the nation is rapt at attention as this is the first “legitimate” contest of the series — straw polls discounted. lowans have been paying 252 attention to the campaign for a longer period than the public because they have been the targets of nearly two years of candidate courtship. Thus, lowans (and New Hampshirites) go through a far longer “localized” campaign period that most voters in other states. For the rest of us, we begin to pay some attention to these important contests, but it is not “localized” yet. We watch the campaign as a “national” electoral event. We perceive of our place in this event as national in nature. Later, when the campaign moves to our state, it becomes a state-level (local) event. A few weeks before our turn at the ballot box, we are aware of much more campaign activity (depending on our state’s importance and the timing of our event within the sequence of primaries and caucuses). Most of the activity we experience is through a combination of local and national news media. We hear our radio talk show hosts, our local anchors and read the local newspapers to gamer some knowledge about the campaign. Also our opinion leaders — friends, coworkers, ministers, teachers, and acquaintances share the information that they have accumulated. For some of us, the vote choice is already made. However, for others, particularly those who have not paid great attention to the campaign, it may be during the localized portion of the campaign that opinions are formed and the decision for whom to vote is made. Generally, information costs are much lower at this period for local consumers. Candidates campaign in a way that suggests that this localized phase is important. They air their ads right before the individual state contest (post-Iowa and New Hampshire contests), generally eight to 14 days before the primary or caucus. Most campaigners believe it is a waste of money to air ads earlier because their time and resources are 253 generally tied up in the contest of the moment. When candidates do turn their attention to a state, they do so with paid television and radio, press releases to newspapers, phone banks, interviews with the local news media and a bevy of campaign appearances all orchestrated to gamer attention from the local media and support from voters in the state. Ofren we see the national media reporting on national trends in support for the candidates — polls based on samples from throughout the nation. These polls at times do not always reflect the levels of support a candidate may be drawing in a particular state. The nation was surprised when Jerry Brown won the Colorado primary, but Colorado’s media had been tracking his surge for days before the event. The fact that local media may be more “in tune” to the state’s political pulse is particularly true of caucus states. State and/or regional reporters are expected to be more familiar with politics in their local area and thus a better reflection of the preferences of voters and party regulars in a particular state. They often evaluate the candidates within the framework of important local issues, within the context of the state’s particular political culture, and within a competitive framework that changes from week to week. Regardless, the local news media provides information to voters that may be diflerent or at least complementary in some respects to the national news. For example, comparing Harkin’s and Kerrey’s coverage in South Dakota before its contest is vastly different from their national coverage at that point. Comparing the New York Time’s coverage of Kerrey and Harkin to that of the Argus Leader (Sioux Falls, South Dakota) the week of the South Dakota primary shows this difference. From February 19 through February 25, the New York Times devoted a total of 32 front page words to Kerrey and 254 Harkin combined. The coverage dealt with their battle to eliminate each other. In contrast, the Argus Leader devoted 526 front page words to Kerrey and 325 to Harkin. Topics included competition, support levels, viability, agriculture, health care, economic and energy issues, message, candidate spin, debate evaluations, and general campaign activity. The Argus Leader, not only gave Kerrey more coverage, they also endorsed him. 6.1 The Intermediary Model of Presidential Nomination Campaign Outcomes Since voters receive very little first-hand information about candidates, we can only assume that most of their information comes directly or indirectly from information intermediaries. For the presidential nomination campaign these would consist primarily of the news media, both national and local. In certain instances, interest groups at the national and local level may attempt to send information to voters, but unless a voter is already a member, it is unlikely that he or she would receive materials from the group. Thus, the mass media, particularly the news media, is the primary conduit to the “mass” public. Even if opinion leaders are involved, it is likely that they too have received their information from the news media. This is not to say, however, that every individual voter is exposed, or if exposed, at the same level. Bartels (1988) documents familiarity levels of two 1984 Democratic candidates, Gary Hart and Walter Mondale. He found that individuals polled had more familiarity on candidate traits than issues. Yet what is more important, he could illustrate that early in the campaign, media use (low and high) differentiated between familiarity levels among the candidates, particularly less well-known candidates. Yet, as the campaign 255 continued, the familiarity gap narrowed considerably. Average information levels about Hart and Mondale were separated by a wide gulf early in the campaign with Mondale the more well-known, but differences narrowed and information parity resulted as the campaign continued. This suggests that as the campaign progresses, moving from locality to locality with the field narrowing along the way, the public becomes more aware of the candidates. Nor can one assume a uniform behavior response when a message does reach readers and viewers. The model simply suggests that candidates given more exposure by the national and local media are more likely to do well overall in an upcoming contest. Whether the national or the local press shows a stronger relationship to the choices made by the voters, is a question with little precedent. One might expect that as to prediction the local media would be a better bet given its greater knowledge of the specific audience and the presentation of customized information about the candidates’ activities in their state. To test the hypotheses that local media coverage and national media coverage affects primary and caucus outcomes in the presidential nomination campaign, I have developed a theoretical model that can be empirically tested. This model purports that we can predict a candidate’s vote share by taking into account the 1. candidate’s national media coverage levels before the contest (coverage volume relative to competitors), 2. national directional coverage (positive and negative volume relative to competitors), 3. local media coverage levels before the contest, 4. local media coverage valence, 5. local and national valence coverage differences, 6. state-specific spending (which represents the 256 candidate’s resource investment in the state), 7. the candidate’s ideological position relative to the state’s ideological culture, 8. the candidates’ home state or regional aflBliation, and 9. a control for incumbency. This model will be tested with pre-event coverage data gathered &om the 21 state newspapers and four national media outlets discussed in previous chapters. Before testing the multivariate model, however, the variables will be explored regarding their bivariate relationships to the dependent variable, vote outcome. 6.2 Bivariate and Multivariate Hypothesis Testing The national and local coverage volume and positive coverage shares are hypothesized to be positively related to a candidate’s vote outcome. These variables capture the candidate’s exposure levels by the national and local media, and the relative degree of positive exposure. Positive coverage should be positively related to vote share. Negative coverage, intuitively, is assumed to be negatively related to vote outcomes. The more negative coverage one receives, the lesser one’s expected vote share. In marketing circles, this type of information leads to consumer skepticism. If one hears that a product has flaws, purchasing it not as likely - unless it has no alternative competition or unless it offers something of greater value than its competitors. We know that Clinton won the nomination in 1992, yet compared with his competitors, he received the highest levels of negative coverage. Bush received much negative coverage but won the nomination. However, he did go on to lose the general election. So the evidence suggests that negative coverage may have no impact, have a 257 positive relationship with vote outcomes, or may have its negative impact negated by positive coverage’s impact. It may be that both positive and negative coverage have little relationship with vote outcomes. Perhaps all that is important to candidate success is repeated exposure and not the substance of what is exposed during the presidential nomination contests. The findings in Chapter Six suggest that candidates who do well receive the bulk of exposure, both nationally and locally.' Table 6.1: The Relationship of Local News Media Outlet Coverage and Vote Share Controlling for National News Media Outlet Coverage Partial Correlation Controiling for... Democrats Republicans Local Coverage Level & Vote Share National Coverage Level .55 (.01) n.s. Local (+) Coverage & Vote Share National (+) Coverage .59 (.01) .38 (.05) Local (-) Coverage & Vote Share National (-) Coverage .31 (.01) ILS. Note: Table entries are the partial corrélation coefficients and their significance levels (in parentheses). The partial correlations allow the measurement of the linear relationship between two variables when the linear effixt of the control variable is removed. 'Moreover, directional coverage is generally found to be of three çpes: policy evaluation (found to be heavily negative for Bush, particularly in the economic and trade areas) and campaign/horse race coverage (positive and negative evaluations of campaign activity and outcomes) and character issues (in 1992 Clinton and Brown both had negative evaluations in this area while Tsongas had positive ones). As suggested earlier, perhaps the direction of coverage is not important in some areas, but critical in others. While I do not test this hypothesis here, it would be a worthy topic for future examination. 258 Table 6.2: The Relationship of National News Media Outiet Coverage and Vote Outcome Controlling for Local News Media Outlet Coverage National Coverage Level & Vote Local Coverage Level a s. .78 (.01) Share National (+) Coverage & Vote Share Local (+) Coverage as. as. National (-) Coverage & Vote Share Local (-) Coverage .29 (.01) .89 (.01) Note: Table entries are the partial cocrelahoa coefficients and their significance levels (in parentheses). The partial cocrelatioos aUow the measurement of the linear relatitmship between two variables when the linear efiect of the control variable is removed. Table 6.3: The Relationship of Overall Direction of Coverage and Vote Share Local Positive/Negative Difference Natl. Positive/Negative .63 (.01) .54 (.01) Difference Natl. Positive/Negative Difference Local Positive/Negative as. -.75 (.01) Difference Note: Table entries are the partial correlatioa coeffiaents and their significance levels (in parentheses). The partial cmrelatioos allow the measurement of the linear relationship between two variables when the linear efiect of the control variable is removed. The partial correlation ccefficieirts and their significance levels for the media variables and vote outcome are presented in Tables 6.1, 6.2 and 6.3. The Democratic campaign vote outcomes have all the appearances of being driven by local coverage to a greater extent than national coverage. The Republican campaign’s results were better reflections of the national media. Thus, in 1992, the competitive campaign, boasting many candidates, was more likely to have its outcomes associated with local news media coverage. The campaign involving the incumbent president, whom the national media 259 covered to a greater degree, was more closely associated with the national media’s coverage, particularly since an incumbent might use a “rose garden” strategy. Such a strategy would not lend itself to the factors that generate more “local” campaign coverage. Note, however, that there is no negative direction for the correlation coefficient for either local or national negative coverage. In fact, the higher Bush’s level of negative coverage the greater the vote outcome. Again, this is a reflection of the spurious relationship of the incumbent president receiving more negative coverage and more votes without the two being causally related. This is better illustrated by the entries in Table 6.3. For both the Republican and Democratic candidates, receiving more local positive than negative coverage is positively related to vote outcomes. During the week before a primary or caucus, if a candidate received more positive local coverage than negative coverage compared with his or her opponents, then the candidate would generally have a greater vote share. We do not see, however, the same relationship for the national media’s directional coverage. There is no relationship between receiving more positive national coverage compared with one’s opponents and receiving more vote shares for the Democratic candidates. A relationship does exist for the Republicans, but it is in a negative direction, again, reflective of a spurious relationship. The level of and direction of the national and local media is assumed to influence the aggregate vote outcomes in state primary and caucus contests. However, other variables that reflect a candidate’s direct or surrogate contact with the populous are thought to have some relationship to that candidate’s share of the vote as well. 260 How much time and money spent in a state should help explain how much support a candidate has on election day. Two variables are used to tap these concepts - candidate attention to and candidate spending in the state. It is assumed that candidates who spend time and money heavily in a state will do better in that state. Not only because candidates usually target funds to states in which they think they will do well, but because it is assumed that much of this money is used to convey information about the candidate to the public. Big chunks of campaign money end up going the purchase of television and radio time, sometimes newspaper advertisements. Moreover, candidate and surrogate visitations are used to reach the local population. The correlations reported in Table 6.4 suggest that candidate attention had more of a relationship with vote outcomes in the Democratic contests, while spending was associated with vote outcomes in the Republican campaign. State level financial investments did not seem to play a role for the Democrats in 1992. Candidates often spent more money in states in which they were in trouble. Harkin spent much of his money early, as did Kerrey, and both were near complete campaign bankruptcy before the March contests. Tsongas had little money initially, and when it did come in after his New Hampshire win, he could not convert it into state by state investments fast enough. Clinton had the most money and used it. Generally, Clinton would do well both in states he spent heavily and those in which he spent less. Part of the reason for this may be that as frontmnner he received a great deal of “free” exposure through the local media. 261 Table 6.4: The Relationship Between Candidate Resource Investment and Vote Share Candidate Attention Candidate Spending (in S) Democrats tL& Republicans .47 (.o n Brown .56 (.01) n.s. Clinton .44 (.05) as. Haiidn .82 (.01) n.s. Kerrey ILS. as. Tsongas .58 (.01) as. Bush as. Buchanan 0.$. .61 (.03) Note: Table entries arc bivariate correlation coefficients and their significance levels (in parentheses. Finally, in attempting to predict and explain a candidate’s state primary or caucus vote outcomes, one would necessarily want to account for ideology. The presidential nomination campaign is sometimes dominated by particular candidate messages that articulate in brief a candidate’s ideological position within the party. They rely on this message to differentiate themselves from their partisan competitors and select issues to package their political ideology. Harkin was an old-time liberal who focused on the New Deal premise of building America’s infrastructure and generating jobs by this means. That would lead to a better economy and a better society. Clinton was the New Democrat, appealing to the middle class with tax breaks and speaking of getting the economy rolling again. Tsongas was the tough love liberal Democrat - a fiscal conservative with a soft side 262 on social issues. Kerrey had trouble developing a message and positioning himself ideologically. Brown was viewed by many as a far-Iefter, but many of his speeches spoke of reform that appealed to conservatives, moderates and liberals. His flat tax, for example, played well in a conservative state like Utah, while his environmental and energy positions did well in more liberal states like Vermont and Connecticut. To capture the potential barrier or benefit to a candidate from a state’s political culture, I use Wright, Ericson and Mclver’s index for state ideology. This variable gives an indication if a particular candidate did better or worse in more liberal or more conservative states. States with higher levels of conservatism have positive scores while liberal states have negative scores. Table 6.5: The Relationship Between State Ideology and Candidate Vote Share Brown Clinton Haridn Kerrey Tsongas Bush Buchanan Bivariate Correlation -.35 .05 -.24 .20 -.41 -.02 .01 CoefiBcient (.01) (n.s.) (n.s.) (n.s.) (.03) (n.s.) (n.s.) Brown and Tsongas were the only candidates to have statistically significant relationships between their vote outcome and the ideological culture of a state. Both Tsongas and Brown did better in states that were more liberal. While the other candidates’ correlation coefficients do not reach statistical significance, it is noteworthy that the direction of the relationship is as expected. Harkin appears to have done better in more liberal states, while Kerrey did better in more conservative states. Clinton’s 263 coefiBcient value of .05 suggests that his vote was not tied to ideology. He would draw from both liberal and conservative states. Bush and Buchanan’s coefficients suggest that Buchanan had a slight edge in more conservative states as expected. Still, overall, state ideology did not appear to be related to their vote shares. Table 6.6: The Relationship Between Home State and Home Region And Vote Shares Democrats Republicans Home State .29 (.001) .09 (as.) Home Region .40 (.001) .07 (as.) The variables that represent whether or not the contest is taking place in the candidate’s home state or home region (those states bordering the candidate’s home state, inclusive of the home state) are correlated significantly to vote outcomes for the Democrats but not for the Republicans. The correlations between home state and region and votes are presented in Table 6.6. 6.3 Media Coverage and Vote Outcome Before moving to the results of the ffdl multivariate model, the relationships between the media variables and vote outcome will be discussed more extensively. Table 6.7 presents the bivariate correlation coefiBcients for the media variables for both parties. 264 Visibly, the relationship between local coverage and vote outcome is much stronger for the Democratic campaign. On the Republican side, the national news coverage dominates. Moreover, among the media variables, there is a moderate relationship between national and local coverage (.63, .01 two-tailed) for the Democratic campaign, whereas in the Republican campaign we find a much higher correlation (.83, .01 two-tailed). Moreover, for the Republican campaign the level of national coverage and national negative coverage are very highly correlated (.92, .01 two-tailed). The largest correlation, however, is found in the Republican cases with national negative coverage correlated with vote outcome at .94, .01 two-tailed. Again, reflective of a spurious relationship. Note that national positive coverage for the Republicans was only weakly correlated with vote totals. Not only do we find the media relationship with vote outcome differentiated between the campaigns, but we find that local and national media coverage was more similar with regard to their linear association in the Republican party. This is not a surprise, again, due to what we know about the nature of the 1992 media coverage and the factors that moved it. We see this differentiation in the regression coefGcients of the media variables presented in Tables 6.9 and 6.10. The national media variables perform well in explaining the variance in the Republican candidates’ vote outcomes, but poorly for the Democrats. Local media coverage variables, do well for both parties. 265 Table 6.7: The Relationship Between National and Local News Media Outlet Coverage (Amount and Direction) and Candidate Vote Share Democrats Local Local Local Pos. National National National Coverage Neg. Coverage Neg. Pos. Local Coverage .53 (.01) .67 (.01) .63 (.01) .28 (.01) .43 (.01) Local Neg. .14 (.01) .19 (.01) .16 (.01) .03 (.01) Local Pos. .61 (.01) .25 (.01) .61 (.01) National Coverage .57 (.01) .63 (.01) National Neg. .27 (.01) Vote Outcome .6 6 (.01) .34 (.01) .65 (.01) .44 (.01) .32 (.01) .37 (.01) Republicans Local Local Local Pos. National National National Coverage Neg. Coverage Neg. Pos. Local Coverage .60 (.01) .47 (.01) .83 (.01) .72 (.01) .27 (.01) Local Neg. .20 (.01) .59 (.01) .61 (.01) .13(01) Local Pos. .47 (.01) .49 (.01) .49 (.01) National Coverage .91 (.01) .35 (.01) National Neg. .28 (.01) Vote Outcome .76 (.01) .69 (.01) .45 (.01) .91 (.01) .94 (.01) .26 (.01) 266 Table 6.8: Local News Media Outiet Coverage (Amount and Direction) Variables and Their Relationship to Candidate Vote Share Democrats B SEB Beta Sig.T Local Coverage Level .27 .12 .29 .0284** Local Positive .31 .08 .44 .0003*** Local Negative .08 .06 .12 .2438 (Constant) 7.50 2.39 .0024 Adjusted R-squared .50 Std. Error of the Estimate 13.47 N=79 Republicans BSEB Beta Sig.T Local Coverage Level .41 .15 .45 .0134** Local Positive .11 .08 .18 .1795 Local Negative .20 .10 .33 .0520** (Constant) 14.46 5.96 .0228 Adjusted R-squared .61 Std. Error of the Estimate 14.44 N=30 Note: Table entries arc ordinary least squares regression estimates. *=< .10, •*=S05. •••=<.01 267 Table 6.9: National News Coverage (Amount and Direction) Variables and Their Relationship to Candidate Vote Share Democrats B SEB Beta Sig.T Natl. Coverage Level .27 .16 .26 .0962* Natl. Positive .16 .12 .17 .1933 Natl. Negative .11 .11 .11 .3550 (Constant) 10.40 3.49 .0039 Adjusted R-squared .19 Std. Error of the Estimate 17.32 N=79 Republicans BSEB Beta Sig.T Natl. Coverage Level .25 .12 .35 .0499** Natl. Negative .40 .10 .63 .0007*** Natl. Positive -.06 .07 -.06 .4354 (Constant) 18.81 3.17 .0000 Adjusted R-squared .89 Std. Error of the Estimate 7.70 N=30 Note: Table entries arc ordinary least squares regression estimates. •=< .10,**=s05, *•* ■=<.01 As we have seen, local and national media coverage of the Republican campaign centered on Bush. Those high coverage levels match up well to his high vote outcomes therefore explaining why we see a relationship. Exposure before an event is not likely to matter to an incumbent. Negative coverage may hurt his support levels somewhat, but in the context of this study, there is no way to determine this point. Perhaps if we compared his vote totals from the 1988 campaign with the vote totals in 1992, we may see a difference in his support. Because the circumstances differ so greatly, testing it is diflScult. 2 6 8 The Democratic campaign, with its greater variation in outcome and coverage for the candidates. Is an easier case to argue. Exposure would be important to candidates, particularly those less well known and lacking financial resources. Moreover, positive coverage (generally in the vein of doing better than your competitors) would contribute to any bandwagon effects that may occur. According to the preliminary regressions above, the Democratic campaign’s outcomes were better predicted by the local media variables than the national media variables. To see how well the intermediary model of presidential primary and caucus vote outcome works under dififering competitive circumstances, it is applied to both the Democratic and Republican campaigns. The bivariate analyses suggested that some variables may not be related to vote outcome. Yet, given the presence of control variables, we may see different results in the multivariate analysis. Therefore, the criterion for inclusion was an indication of a relationship present for either party, therefore the only variable not included is state ideology. The only difference in model application between the two campaigns is the control for incumbent for the Republican campaign. Using ordinaiy least squares linear regression, the estimates for the variables have been generated and are reported in Table 6.10. 269 Table 6.10: Model of Candidate Vote Share: Democratic Campaign Variable B S.E.B Beta Sifc.T Local Coverage Level .22 .09 .248 .0181*** Local Positive Share .16 .07 .229 .0311** National Coverage Level .15 .10 .148 .1321 Candidate Attention to State 10.16 2.95 .264 .0007*** Candidate Spending (thousands) .02 .01 .107 .1217 Home Region 12.60 3.78 .244 .0014*** Home State 17.84 9.01 .146 .0516** (Constant) 2.60 2.42 .2858 R- Squared .70 Adj. R Squared .67 Standard Error 10.99 N=79 Note: Local and National Positive/Negative Diffetcnce, Local Negative and National Negative and positive are dropped from the model due to multicollinearity. Table entries are ordinary least squares regression estimates. *=<.10, **=<.05, ***=<.01 6.3.1 The Democratic and Republican Campaigns In 1992, the Democratic party’s nomination process was marked by many shifts in the media’s candidate focus. They covered first Tsongas, then Clinton, as the frontmnner. Brown, who was literally ignored early in the campaign, became a focal point after several surprising finishes in the middle of the campaign; Many competitive shifts marked the information that voters received during the 1992 campaign as well. Early in the campaign season, George Bush dominated the airwaves and pages of major news outlets. His campaign was in disarray even before it oflRcially began. Then the Democrat’s surprisingly competitive campaign shared the spotlight with a rancorous challenge to the incumbent President from Patrick Buchanan. Then scandal for Clinton and negative campaigning by Clinton (toward Tsongas and vice versa) and the other competitors followed. When 270 Clinton and Bush finally landed their knockout punches, a new player, Ross Perot, was on the scene being pursued by the news media at every turn. Throughout the campaign, but particularly early, the public partook of a roller coaster ride of information. Daily the attention, the coverage directional tone and the issues changed. Nevertheless, did it matter? For the Democrats, it appears that in some ways it did, and in others, it did not. As seen in Table 6.10, the estimates for media impact in the Democratic campaign tell an interesting story. The model can explain almost 67 percent of the variance in the 21 state outcomes examined. The standard error of the estimate suggests that this model can predict a candidate’s vote outcome within 11 percent of the actual vote outcome. The regression results suggest that vote outcomes in the Democratic campaign of 1992 were influenced by local coverage more so than by national coverage. However, problematically, we know that national coverage influences local coverage somewhat, thus national coverage level may have an indirect influence through local coverage that this model does not reflect. However, the direct eflfects indicate that how much coverage the candidate received fi’om the local news media and the degree of positive coverage were positively related to candidate vote outcome. Unfortunately, due to multicollinearity, we cannot estimate the effect of negative coverage levels (both national and local) on vote outcomes, but indications are that the association was positive. That is not to say that this negative coverage made people want to go out and vote for the candidate receiving it. It simply means, that in this campaign, one of the candidates who received negative coverage consistently, was Clinton, the eventual nominee. Despite negative coverage he kept doing 271 better than his opponents. Again, this may be a spurious relationship where frontrunners receive more negative coverage but they continue to do well because they are frontrunners, or it could suggestion that negative character coverage does not matter in certain competitive circumstances. National coverage does not reach statistical significance, but its estimate is in the right direction, although smaller in magnitude that local coverage. Candidate attention has a significant impact on vote outcomes. Devoting one’s personal resources to a state can boost one’s vote by as much as 10 percent. For the Democrat’s in 1992, money does not seem to have had a great direct influence on vote outcome. Being a native or regional candidate, however, could boost one’s vote total significantly. Theoretically, a Democratic candidate who received 10 percent of the local coverage the week before the primary, and 20 percent of the positive coverage, spent $10,000 in the state, and was fi-om the region could expect a vote share of 21 percent. If the candidate was not fi'om the region, his vote share would decline to 8.47 percent. Taking into account the standard error of the estimate, these predictions could range +/-10 percent. As Table 6.11 shows, a control for George Bush, the incumbent has the strongest relationship with vote outcome. Throughout the campaign. Bush would maintain a post- New Hampshire vote outcome of 70 percent or better, while Buchanan struggled to achieve his goal of 30 percent, the magic number determined by New Hampshire’s Republican primary. The model suggests that Bush did better in his home state of Maine (which he did with more than 80 percent, while Buchanan earned less than 4 percent of the 272 vote), adding to his substantial base vote. In addition, both candidates did better in the locales where they could generate local news interest. This is particularly true of Buchanan. The intercept captures Buchanan’s starting point at 20.29 percent. If we examine Buchanan’s coverage in New Hampshire, we can see how the model treats the challenger. Buchanan received 49.40 of the local news coverage in New Hampshire and Table 6.11: Model of Candidate Vote Share: Republican Campaign Variable B S.E. B Beta m .T Local Coverage Level .19 .09 .214 .0601* Local Positive Share .02 .04 .043 .4803 National Coverage Level -.05 .14 -.079 .7036 Candidate Attention to State -.93 3.32 -.019 .7817 Candidate Spending (thousands) .005 .005 .068 .3301 Home Region .379 4.99 .005 .9401 Home State 25.59 9.63 .005 .0143** Incumbent 36.14 7.86 .203 .0000*** (Constant) 20.29 3.14 .0000 R-Squared .95 Adj. R Squared .93 Standard Error 6.21 N=30 Note: Local and National Positive/Negative Difference, Local Negative and National Negative Positive are dropped from the model due to multicollinearity. Table entries are ordinary least squares regression estimates. •=< .10, **=<.05, ***=<.01 78 percent of the positive coverage. He spent $456,866 and much time in the Granite state. He received only 10.7 % of the national coverage that week. Using only the parameters of local coverage, local positive coverage, candidate spending and the intercept, the model predicts 33.52 percent of the vote. Buchanan actually won 37.40 273 percent. For both of these candidates, their variance in the dependent variable was confined to a range somewhat: Bush’s vote share was in the higher range and Buchanan’s was in the lower range consistently. Thus, a dummy variable for incumbent captures this variance. 6.4 Conclusion This chapter set out to compare the influence of two types of media outlets - the national and local news media. While the results produced here are rather preliminary, they shed some light on a topic on which little is known as few comparisons have been made. In 1992 the two parties went through different types of campaigns. The differences in the campaigns had implications for the type of media coverage that would be generated and the influence this coverage exerted on the public. One conclusion that can be drawn fi-om the results of this analysis is that in 1992, the competitive campaign, one that is complex in nature, was more likely to have state-level conditions affect its outcomes. The non-competitive campaign’s outcomes - that of an incumbent with an internal challenge - had greater association with national trends as exemplified by the stronger relationship of national media coverage and vote outcomes for the Republicans. Yet this makes sense. Bush had little necessity of running a typical state-by-state campaign. He was the incumbent president responsible for the well-being of a nation. He did not have to appeal to state constituencies unless Buchanan made a strong challenge (as he did in New Hampshire and Georgia). 274 Generally Bush had to appeal to the nation as a whole, moreover, he still had to lead the country while he campaigned, focusing more attention on his duty as our national leader. The incumbent did not need to gain exposure locally because people knew who he was, and to a certain extent, had an impression of what he had done. Thus, the national news media set the agenda for Bush’s campaign coverage, which the local media mostly reflected, while their coverage of Buchanan, who was active in a number of states, differed somewhat. The degree to which some local outlets covered the Bush campaign differed as well. Some covered the president more than the national media. The Democratic campaign, however, was fought at both levels; state-by-state and nationally. We would expect to see the dual impact of national and local media in a process in which the national media set the expectations at the beginning of the campaign and respond accordingly to each outcome. This response would be influenced by state level conditions, as the campaign progressed. For the candidates exposure is important. Perhaps it is too important. Candidates who can get the attention of the media repeatedly do well. Candidates who do not, do poorly, particularly if they are running a conventional campaign. Whether this process propels the “best” candidates forward is questionable. Candidates with greater financial resources appear viable to the national media out the outset, but often the source of this money is neglected." ^Often the financial support of a particular candidate is state or regionally derived. A large portion of Clinton’s money came fi-om Aricansas. The Federal Election Commission’s requirement that a candidate raise $5,000 in contributions of $250 or less firm twenty states is not sufficient to gauge a truly national candidate. Moreover, $100,000 would not get a candidate very far. So even candidates who meet this test, must have more money in their bank accounts to be considered competitive. 275 While exposure is important, negative exposure, at least in the context of the 1992 nomination campaign, did not affect vote outcomes, except perhaps for the earliest dropouts, and even here the relationship is small. Positive coverage is important, but the nature of positive coverage is generally game-related. Doing well, getting endorsements, rising in the polls and so on are the types of topics addressed in positive coverage. The news media, both nationally and locally, do little comparatively in their positive and negative assessments of issues or governance, except when it involved the incumbent president or a frontruimer. Much of the candidates’ issue coverage is neutral. However, reporters do make choices regarding whether or not to include anti-opponent coverage in their stories - and this is often issue specific or evaluative in nature generally. Whether or not voters discount this information as it comes from a candidate’s opponents is unknown. The outcomes in presidential nomination campaigns are influenced by the news media, both locally and nationally. How and when they are influential may vary. The process by national media influences local media and vice versa are phenomenons worth further pursuit. 276 CHAPTER? NATIONAL AND LOCAL MEDIA AND THE NOMINATION PROCESS This study has answered three general questions about the national and local news media. First, what was the nature of national and local news coverage variation during the 1992 Republican and Democratic presidential nomination campaigns? Second, what explained the variation found in national and local campaign news coverage? Finally, did local and national campaign news coverage have political consequences for the candidates they covered? The findings of the study are reviewed in the next section. A discussion of implications and normative concerns regarding the news media and presidential nomination campaigns conclude the chapter. 7.1 Summary of Findings 7.1.1 National and Local News Media Coverage: Homogeneity Revisited The fact that most national news outlets follow a pattern of allocating more coverage to early contests and “big” contest events is not a new one; however, this study finds evidence for a similar pattern in local news outlets. Examination of four national and elite news media outlets and nine state and metropolitan newspapers over a sbc month period suggested that the national media follow this pattern more closely than the local 276 media. Local news outlets diverge from the national norm due to the attention given to their own contests or, on occasion, to regional activity. The news coverage given to particular candidates was also examined over time. Here more variation was found to exist between national and local coverage levels. Yet, the attention given to candidates by the national and local press was similar in one respect. Both types of outlets gave the incumbent president and/or the out-party’s principal frontmnner the greatest coverage throughout the campaign. Beyond the coverage given to frontmnners in either party, the variation abounds, particularly of the middle-tier candidates. These différences were even more visible at the state level. Regional competition and candidate afiiliation with a state or region can alter the focus to candidates who have not received a great deal of national attention. As to the directional coverage of the candidates, the national media allocated more directional coverage to the frontmnners and their challengers. The local media examined strayed from this pattern somewhat by also focusing directional coverage on the third tier candidates. Moreover, the national press tended toward more negative coverage in general than the local outlets examined in this study. However, as the national media cover national frontrunners more negatively, the local news media are likely to cover “local” frontmnners more negatively. However, these candidates are also more likely to receive substantial positive coverage as well, a point that has not been made in the literature. As to the issue coverage, the examination of news outlets over time suggests that overall, the national media allocate more issue coverage to the campaign. However, when 277 we examine the coverage a local news outlet gives to the campaign immediately before its own primary or caucus, we see far more issue coverage. The local outlets in comparison to their national counterparts meet or exceed the amount of issue coverage. This area has not been examined in the literature on presidential nomination campaigns. Nevertheless, it is an important area given the implications for the quality of voting choice made by the voters in primary and caucus events. While it may not be intentional, in general, the news media’s coverage of the campaign limits the public’s access to types of information about the candidates, particularly later in the campaign. Except for the introductory candidate issue stories that come early in the campaign, most issue coverage was embedded in campaign-activity or horse race stories. Often, issue coverage came during times of heated competition. This may be because candidates used substance (relating to either issues or candidate characteristics — such as age, experience, etc.) to define and contrast themselves with their competition. In 1992, issues such as energy, the environment, and tax cuts prevailed on the Democratic side, while taxes, the budget and trade were key issues for the Republican candidates. Candidates, in their speeches and with their advertisements, can focus attention to particular issues. Few will forget the Buchanan commercial that initiated much news media attention to the issue of public fimding of the arts. His portrayal of Bush as the president who gave financial support to “pornographic” and “immoral” art created a week of discussion on the topic. Even at the local level, where overall we see more substantive pre-event coverage for the out-party campaign and marginally less over the course of the entire campaign than 278 the national media, the issue coverage appeared to be driven by the candidates themselves rather than by the reporters’ analyses and investigative reports. This is less the case for the national news media who appear to follow both a campaign and a responsible citizen model of campaign coverage. This study found that issue substance coverage is rather limited, but present. Frontrunners and incumbents generate more of it generally. Other scholars have found that the presence of an incumbent increases substantive coverage in the general election. While the national media on average presents more issue coverage overall, they appear to do so more at the outset and the latter part of the nomination campaign as the transformation to the general election gets underway. The local media, however, appear to focus more intensely on issues the week or two before their primary or caucus. Yet often, these issues are limited in number compared with the national media’s wider range of issue coverage. Also of note, issue coverage intensified for the local media, primarily for the Democratic campaign, in the week before an important contest. Perhaps this is due to candidate behavior or perhaps the local media feel some obligation to focus on some issue or issues while the candidates are campaigning in their state. Regardless, this is suggested that at least locally, competitive contests are not covered solely with the game in mind, but that local journalists and editors write and select stories that include substance. Perhaps this is why we see the public’s knowledge of the candidates increase over time. The process moves from state to state with the local media (at least while the campaign is competitive) providing a range of information at a time when their information consumers are most likely to pay attention. 279 Scholars have found that voters’ preferences are affected by their perception of candidate viability (Abramson et al., 1992; Bartels, 1988; Brady and Johnston, 1987), but more interesting is the linkage of perceptions of viability and media coverage (Paolino 1996). My study suggests that performance factors influence the national coverage of candidates, which in turn influences local coverage. It may be that voters are responding to the greatest pool of information they have — candidate performance in making their voting decisions since substantive information is generally limited. Overall, the national and local news media do share some similarity of coverage patterns, however, they are not mirror images of each other. In other words, the national media do not determine the nature of local campaign coverage wholly in the presidential nomination context. There is evidence that the local media are more likely to follow the national media in areas of “national” policy-making such as coverage of the executive and congressional branches (Bartels 1996). However, this may not apply to a political process that is inherently state-oriented. Thus, the question of homogeneity is contradicted by this analysis. Previous literature has relied on the assumption of homogeneity to simplify the representation of the news media in empirical analysis of national nominations and elections. And while there are many parallels in local and national coverage, such as focusing on ffontrunners, this does not mean that differences, both great and small, do not exist. These differences may help us understand how vote outcomes vary in contests when the candidates stay the same. It may also help us understand when and why voters become interested in the campaign and thus, participate in it. Furthermore, we may gain some insight into how other candidates respond to their own coverage as well as that of their 280 opponents. Thus, studies of the presidential nomination campaign (and the general election perhaps) intent on explaining the outcomes must consider local media as well as national news media since primaries and caucuses at heart are state events that in their sum create a national nominee. In sum, these findings suggest that the information contexts can vary fi-om locality to locality in addition to varying fi’om a national information context to a local one. If we conceptualize the nomination campaign as a sequential game with each primary or caucus as a decision node, then these findings suggest that at each node the information available to the players (media, candidates, and most important, the voters) will vary in amount and content given the needs and motivations of both the national and local news media plus a host of other conditional factors. 7.1.2 What Moves National and Local News Media Coverage? One of the most interesting discoveries of this study of the 1992 nomination campaigns is that the antecedents of local and national coverage were different. The national media appeared to follow a relatively simple formula for determining who will receive coverage, whereas the local news outlets relied on a mix of factors. The national media focused on the fi"ontrunners and those with increasing viability. Often these two factors coincided. Frontrunners usually had more viability than their opponents, and if they continued to win, their viability estimates increased with each delegate they accrued. Thus, candidates who did well and continued to do well-received attention fi’om the national media. 2 8 1 While this was also true for the local news media to a degree, they were more likely to consider candidate characteristics, candidate resource investment in their state, performance and local competitive factors. Whereas viability and ffontrunner status could explain upwards of 70 percent of the variation in national coverage levels for both campaigns, more information was required to explain local coverage levels. The factors that influenced directional coverage of the candidates in 1992 varied for the national and local media, but more so for the Democratic (out-party) campaign than the Republican (in-party) campaign. Local positive coverage, like local coverage in general, was systematically related to candidate characteristics, performance and resource investment factors. Nevertheless, the positive evaluations of the national media also played a role. Local coverage of the Republican campaign was explained primarily by the positive evaluations of the national media and a variable that represented incumbent president George Bush. National directional coverage of the Democratic campaign, however, was explained by performance variables; momentum, competitive field size, national poll standing, and an indicator that a candidate has accumulated more than half the delegates needed to win the nomination. National coverage of the Republican campaign also corresponded to performance factors: momentum, national poll standing and the halfway indicator. Negative coverage by both the national and local news outlets was by far the most difficult to explain. At the local level, the Democratic candidates’ negative coverage was most influenced by local competitive status. Candidates who are the local ffontrunners 282 were critiqued as severely by the local news media as are national frontrunners by the national media. This complements the literature that holds that frontrunners are treated more harshly than their opponents. However, it expands the theory to local news coverage of both national and local frontrunners. Momentum going into a primary or caucus appeared to decrease the level of negative coverage.' The Republican campaign’s local and national coverage levels were predicted best by indicators that captured George Bush. As a sitting president. Bush, was more likely to receive negative coverage by both local and national outlets. His negative coverage ranged from policy issues such as the economy, trade, taxes and the environment to campaign issues such as poor organization and less than expected outcomes. Understanding the driving political factors behind national and local coverage is an area that has not been explored quantitatively. These preliminary analyses suggest that the attention that the local and national news media allocate to candidates is driven by different types of political factors. This suggests that the differences seen in Chapter 5 are driven not only by the structural features discussed (such as timing of a contest and the importance of the contest) but also for the local and national outlets by candidate-related variables. 'while negative coverage as defined as a quote from an opponent is not included in the analysis due to the requirement that only “objective” news coverage be included in the data, it is worth noting that increases in a candidate’s general level of negative coverage were coincidental to increases in attacks by the opponents. These anti-opponent data will be analyzed in subsequent studies. 283 7.1.3 The Political Impact of Local and National News Media Campaign Coverage In this study’s examination of the political impact of national and local coverage on primary and caucus outcomes I found that local coverage has a greater influence on candidate vote shares than national news coverage for the out-party candidates. Moreover, with the additional influence of local positive coverage, local coverage in general has more than three times the impact of national news coverage. Thus, candidates who received more local coverage, and of this coverage greater positive shares, did better in the state’s primary or caucus. However, the findings in Chapter 6 suggest that national candidate coverage levels influence local coverage levels, so indirectly national news coverage may have influenced the vote outcomes in the states examined. For the Republicans, the model’s estimates suggest that the best predictor of vote outcomes was a control for incumbent president Bush. However, local pre-event coverage does appear to have a positive influence on vote outcomes. This was particularly true in the states that gave Buchanan more coverage. The results of the study suggest a potentially dynamic relationship between the news media, the candidates and the voters with each group having some potential influence on a primary or caucus outcome and also subsequent contest outcomes. Figure 7.1 represents a theoretical dynamic model of primary and caucus vote outcomes. Candidate characteristics such as home state and home region can influence local coverage, while the characteristic of being the fi’ontnmner or incumbent can influence both national and local news coverage of the candidate. Resource allocations of time and money in a state can influence local coverage, while national trends in performance factors 284 influence national coverage primarily. Performance factors, however, can influence a candidates financial situation which can afifect how much attention a candidate can give to a state. National coverage influences local coverage, which in turn influences vote outcomes along with certain candidate characteristics and candidate resource investments. The vote outcomes in a primary or caucus then influence performance factors to be considered during the next round of primaries and caucuses. While I do not examine the dynamic nature of the model in this study, several models are available that allow for an examination. A simultaneous equations model could be examined using LISREL. This would allow for the estimation of indirect effects of key variables. Another approach allows for the estimation of the effect of various actors in a system upon each other. For example, does the national news lead the local news? Do candidates lead the national news? This firamework can be simply represented by a vector autoregression (VAR) model, in which the ciurent value of each variable in a system of interrelated variables is regressed on lagged values of all the variables in the system. The model gives the analyst the freedom on not making judgements about exogeneity. 285 Candidate Characteristics Home State/Region Incumbent/Frontninner Vote Oatcome Local Coverage (Amount& Direction) NatM Coverage (Amount& Direction) ! Resoarce Allocation (Time & Money) Performance Factors i FIGURE 7.1: Dynamic Model of Presidential Nomination Vote Outcomes 7.2 The Question of “Bias” One area of long debate in media studies and political punditry has been the question of bias. This question has usually been posed about the general election campaign coverage, but it has relevance in the nomination campaign as well. If the news media had a liberal bias, the most liberal of candidates would be assumed to receive the most coverage and the most positive coverage at that. Tom Harkin was labeled the “liberal” early in the 1992 Democratic campaign. Yet Harkin received neither the most coverage from the media, nor the most positive coverage, except in his home state. Buchanan, the most conservative candidate, received surprisingly little negative coverage. Thus, a “liberal” media bias as far as coverage amounts or direction of the “news” is not found in this analysis. Perhaps it would be found in editorials. The Manchester Union Leader's front page editorials certainly had an anti-Bush and pro-Buchanan bias. Tsongas received 286 endorsements and complementary coverage in numerous newspaper editorials, however, he did not win the nomination. While ideological bias may not be visible, there is a bias documented by political scientists that may be more influential and thus, potentially questionable; the media pays attention to winners as discussed earlier in this chapter. If there was ever a “bandwagon” effect, it occurs in the national news media. This bandwagon effect is driven by winning and doing better than expected and leads to a general bias in favor of winning candidates. The national media lead the bandwagon, and the local media follow to some extent, but can diverge with local bandwagons of their own. The earlier in the campaign, and the more uncertain the circumstances (the more candidates competing, and the more varied their qualities), the more likely the local media may diverge. After a candidate has reached the halfway point in delegate accumulation, however, any competitors left might as well forget free media coverage. They are old news, and therefore no news at that point. It does not matter if thousands of individuals have yet to make their preferences known. Thus, our democratic system of government is stymied when a large portion of those who might have had a preference had they the information necessary to develop one are all but disenfranchised. The level of national coverage is driven by viability. Incumbents and established frontrunners are given more attention. National positive coverage is driven by momentum, national polls and viability as well. Local coverage, in turn, is driven by national coverage levels, candidate affiliations in the state and a locally-based competitive status value. 287 The game drives the engine of media coverage of the national media, which in turn influences local coverage, but not identically. There is not a one-to-one transmission of attention and content from national to local media, and this may vary by candidate. Yet, both the national and local media may influence outcomes in a state’s primary or caucus. The process then repeats itself. In 1992 it appeared that states that had competitive contests or had one candidate willing to invest a heavy amount of time and other resources (compared with other candidates) had newspapers that were more likely to diverge from the national media’s coverage pattern. One redeeming quality in this process is that the local media appear to pay attention to candidate activities in their state, particularly if a candidate is resonating there and gaining support. Candidates do this by focusing on a mix of local and national concerns and appealing, again, to particular groups within a state - labor in Michigan, ethnic minorities in New York, environmentalists in Oregon, the culinary workers in Las Vegas, the farmers in South Dakota, and so on. Local competitions provide a possibility for candidates to shrug off the specter of little national attention. If they can turn the tide in a state by gaining some traction and then following up with momentum, they can boost their national coverage. This happened to Tsongas early, but he was not able to sustain it. One reason was Super Tuesday. Had that massive roadblock not been there, the Democratic campaign might have been competitive for a longer time. Moreover, it seems that the earlier in the campaign, the more likely a non-frontrunning candidate can gain local attention. As the campaign progresses and becomes dominated by one or two candidates, the likelihood decreases. Under the current frontloaded system, candidates 288 can only give their attention to some states - usually those that are early or important competitively. This practice tends to make people’s votes unequal. New Hampshirites certainly have more say in the process than Nevadans. However, if the process were to spread out over time, giving candidates more opportunities to get traction, generate momentum, and so on, we might see more states playing a meaningful role in the process and we might see more competitive campaigns in general. Whether or not this is something the political parties would like to see is another question. The national and local media’s positive and negative coverage taken together corresponds to whom has done well. So it is not only how much coverage that we see focused on the leaders of the pack, but the direction of this coverage as well. For the national media, the concentration is on who is doing well overall - the viability question. The local media’s coverage reflects this in that they follow the national media’s coverage, but they also consider the local “doing well” criteria as well. The most negative local coverage (over the campaign) went to Harkin, followed by Kerrey, Brown and then Clinton. Tsongas actually had more positive coverage than negative coverage, the greatest shares coming early in the campaign. The order varies for the national media. They gave the most negative coverage to Brown, followed by Clinton, Harkin and Kerrey. Here too, Tsongas received more positive than negative coverage, but only by a small amount. Perot’s coverage differed in that the national media were somewhat critical while the local media were generally positive. I should note that as discussed in Chapter Five, one of the reason we see a deviation at the local level about negative coverage is that often the local fl-ontrunners generate negative coverage, just as the national fi'ontrunners generate more 289 negative coverage. So negative coverage as well as positive coverage is also related to frontrunner bias. The nature of positive and negative coverage bears some discussion. Very often the results of the campaign are not presented neutrally. They are embellished with adverbs and adjectives that give color to the outcomes. Moreover, particularly by the national media, the “news analysis” stories generate the most negative and positive assessments and these often focus on the strategies and expectations of the candidates. This was true for the out-party campaign, and to some extent it is true of in-party campaign as well. However, incumbents must deal with an additional dimension of negative coverage - retrospective policy evaluation and opponent criticism from both the challenger from his party as well as the out-party challengers. The coverage of Bush, the incumbent, was the most similar across outlets. The national media appeared to lead the local media in terms of their incumbent coverage. This corresponds in some way to Bartels’ (1996) finding that the national media, particularly the New York Times, led the local newspapers in their coverage of the executive branch. Whether or not the voting public responds to negative character evaluations is not known with any confidence, although the question merits further investigation. One variable that may play a role in this equation, however, is the relative attractiveness of the candidates. If one’s competition is poor, one’s character flaws may be overlooked if there is something else to balance character - such as strong leadership skills or popular policy stances. Even so, had Clinton not had sufficient money and aggressive fundraisers raising more money, the barrages of negative coverage may have hurt him. In 1988 Gary Hart’s 290 character issue did hurt his campaign. Nevertheless, the conditions under which scandal can afifect outcomes are still unexplored. This study has shown that scandal was not related to local negative coverage, however, it was related to national negative coverage. This also parallel’s Bartels’ (1996) finding that the local newspapers he examined did not cover Whitewater to near the extent that the national news media did. 7.3 Normative Concerns: The Media and Presidential Nominations During the latter part of the 1992 campaign, there was little coverage of the primary or caucus outcomes. In some respects, most assumed the nominee had been chosen. The campaign was no longer competitive. Therefore, not only did media coverage drop, but participation dropped. States that had competitive primaries generally had more local coverage in the aggregate. States that came later in the campaign often had high levels of media coverage during the campaign’s active weeks in February and March, but then dropped off - even when they still had their own event coming up. In fact readers in some states would have had a hard time determining if they had a primary or caucus coming up in their state during the months of May or June because the newspapers simply did not cover them. By then, there was no interest and little participation. This may be one of the reasons political scientists such as Gary Rose (1994) condemn the way the process limits participation. Certainly the news media, both national and local are a part of this process. With ever more primaries and caucuses moving to the top of the calendar, this means quicker resolution of the campaign and little input from contests that come later in 291 the calendar. This has serious implications for the representativeness of the choice of the nominee and the quality of information and potential learning that is generated during the process. Further frontloading may create a situation where the local media’s coverage of the campaign may be altered even more severely. Part of the local media’s incentive to cover the campaign with a local spin is determined by the candidates. If they do not come to the state, it is unlikely that a local debate will occur. If frontloading increases, it will become more diflBcult for candidates to take the time to campaign “locally” in any but the most critical states. We will see more “tarmac” campaigning and fewer local debates. Moreover, candidates who rely on “retail” campaigning in localities to gain support there, do better than expected and thus develop some traction to “push off’ into the next race, will no longer have that luxury. It was difBcult in 1992 and 1996, and it may become more difficult if the frontloading continues. The analysis in Chapter Seven suggested that local coverage could affect a candidate’s vote outcome, particularly in a competitive campaign. Normatively, this is a good sign. The redeeming quality, to a certain degree, is the fact that candidates can rely on state-level conditions to overcome what might be considered national impediments. While it may not always lead to future success, that candidate can spread a message and gamer support and legitimacy for the next campaign. Those who suggest a national primary is the answer to the process’s many faults may not be aware of the deterrent that would be set up for potential candidates. Moreover, a national primary would alter the way in which the local media have covered campaigns. It would seem more likely that if a 292 national primary were conducted, the local media would follow the lead of the national media since the state-level dynamic would be altered severely. Subsequently, primary and caucus frontloading should be addressed and time given between contests so that candidates can turn their momentum in money and their support into success for future contests. While this may extend the campaign somewhat, it would prevent the nomination of a candidate that many partisans feel is not the choice they would wish for the party. However, this ideal contradicts with the need for the party to resolve its nomination in time to prepare for the general election. Yet frontloading contributes to the media’s penchant for simplifying by covering the most viable candidates. This creates a vicious cycle of frontrunner media coverage, followed by frontrunner success, followed by more media coverage (with occasional surprises in outcome) all compacted into a relatively short period. The more delegates at stake early, the more likely a frontrunner will emerge and absorb the bulk of national coverage and some local coverage. This dominance by one candidate viewed as the “viable” one persists in keeping coverage of challengers limited. And this has implications not only for who will be the nominee, but how much influence challengers may have going into the convention. When a candidate reaches the halfway point in delegates needed, the coverage of the campaign switches gears. Challengers who have had their momentum taken away have little opportunity to sway the convention unless the nominee allows them to participate. Thus, this nominating system in conjunction with the news media system can dampen the competitiveness of elections, which are at the heart of democracy. The fact that the local 293 media can sometimes diverge and show attention to those candidates who have received little exposure is a countermeasure since this may help some candidates continue on. 7.4 Conclusion This study of the local and national media and their relationship to the presidential nomination process takes a step in the direction of understanding the role the news media play in a dynamic process. Much needs to be done toward understanding how the media interacts with particular types of campaigns, candidates and the ever changing rules. Understanding how these dynamics may alter the impact of national and local media is a pursuit worth further consideration, particularly when the end result is the nomination of two individuals who will contest for the highest seat of power in the country. The state of the literature in media and politics is at a point of acceleration. The number of studies are growing and more attention is being focused on the news media at the local level. However, the issues of conceptualizing and measuring news coverage need further address. Coverage similarity may be found in some dimensions, but may not in others. Moreover, homogeneity of news coverage may vary by campaign. Patterson’s (1988) study of the 1976 election provided evidence for news coverage homogeneity of both national and local news, however since few studies have examined this question, we cannot say for certain that the results based on 1976 are generalizable to all presidential campaigns. For example, I have demonstrated that if one examined coverage during the week of Super Tuesday, one might find very similar levels of coverage of the campaign and particular candidates. But at other time points in the campaign, the coverage between 294 individual outlets, both national and local may differ. These difiFerences, their influence and the subsequent implications for individual voters, candidates, and the political system as a whole require further investigation. 295 APPENDIX A CANDIDATE LOCAL AND NATIONAL COVERAGE TIME TRENDS 296 1 0 - 6 - ABC 2 - NYT WSJ WP WEEK FIGURE A. I : Brown National Coverage Time Trends 297 NNDP WEEK FIGURE A.2: Brown Local Coverage Time Trends 298 ABC NYT WP WSJ 1 3 5 7 9 11 13 15 17 19 21 2 3 25 27 WEEK FIGURE A 3: Clinton National Coverage Time Trends 299 WEEK FIGURE A 4: Clinton Local Covert^e Time Trends 300 10.0 T 904 8.5- 8.0 7.5- 7.0- 6.5- ! 6.0 - 5.5- 5.0- 4.5 - 4.0- 3.5- 3.0- 2.5- ABC 2. 0 - 1.5- NYT 1. 0 - WP .5- 0.0 .;; WSJ 1 53 7 9 11 13 IS 17 19 21 23 25 27 WEEK FIGURE A.5: Harkin National Coverage Time Trends 301 10 — AJC BND CT 4 - DMR MH MS MUL NNDP SLT 1 3 5 7 9 11 13 15 17 19 21 23 25 27 WEEK FIGURE A.6; Harkin Local Coverage Time Trends 302 10.0- 9.5-; 9.0-I 8.5- I 8.0 - 7.5- 7.0-- 6.5- I 6.0 - 5.5- 5.0- 4.5- 4.0 T 3.5- 3.0- 2.5- ABC 2.0 - NYT 1.5- 1.0 - WP .5- 0.0 ^ WSJ WEEK FIGURE A. 7; Kerrey National Coverage Time Trends 303 10- 9 - 8 "T I 4 I AJC 6 - BND CT 4 - OMR MH 3 - MS MUL NNDP SLT 1 3 5 7 9 11 13 15 17 19 21 23 25 27 WEEK FIGURE A. 8: Kerrey Local Coverage Time Trends 304 6 4 ABC 2 NYT WP 0 WSJ WEEK FIGURE A.9: Tsongas National Coverage Time Trends 305 AJC BND CT DMR MH MS MUL NNDP SLT WEEK FIGURE A. 10: Tsongas Local Coverage Time Trends 306 20 ABC NYT WP / ' y" — \ WSJ 7 7 ÏT 13 15 1 7 19 21 23 25 27 WEEK FIGURE A. 11 ; Bush National Coverage Time Trends 307 AJC BND CT DMR MH MS t'- MUL NNDP SLT WEEK FIGURE A. 12: Bush Local Coverage Time Trends 308 124 1 0 - AJC BND CT □MR MH MS MUL NNDP SLT WEEK FIGURE A. 14: Buchanan Local Coverage Time Trends 309 10^ :î 7 - 6 - 5 - 4i ! I 3 - ! I ABC 2 - NYT / / ' 1 - WP i ; ' oZ - r / - V WSJ 1 3 7 9 11 13 15 17 19 21 2 3 2 5 27 WEEK FIGURE A. 13; Buchanan National Cover^e Time Trends 310 APPENDIX B COVERAGE AREAS AND AMOUNTS BY PARTY OVER TIME PERIODS 311 Table B.1: Substance of the Coverage for the New York Times Period 1: December 1,1991 to Februaiy 18,1992 - Early Days Average Number of words per day: 215 Party Democrats Republicans Total Coverage in words: (17,203) 2676 (15.5%) 14427 (84.5%) Area of Coverage Policy/Issue 277 (103%) 9363 (64.4%) Substance Re: Candidate 684 (25.5%) 494 (3.4%) Campaign 1230 (45.9%) 4463 (30.7%) Horserace 380 (143%) 167 (1-1%) Anti-Opponent Rhetoric 105 (3.9%) 40 (037%) NumbCT Original Stories: 64 Total Stories: 64 Number Wire Stories: Period 2: February 19 to March 10,1992 - Initial Contests Average Number of words per day: 353 Party Democrats Republicans Total Coverage in Words: (7413) 4251 (573%) 3162 (42.6%) Area of Coverage Policy/Issue 162 (3.8%) 385 (123%) Substance Re: Candidate 322 (7.5%) 42 (13%) Campaign 2243 (52.7%) 2336 (73.8%) Horserace 1214 (28.5%) 305 (9.6%) Anti-Opponent Rhetoric 310 (73%) 94 (2.9%) Number Original Stories: 34 Total Stories: 34 Number Wire Stories: 0 Period 3: March 11 to April 7,1992 - Mist Clearing Average Number of words per day: 326 Party Democrats Republicans Independent Total Coverage in Words:(9154) 8051 (87.9%) 1103 (12.1%) 0 (0.0%) Area of Coverage Policy/Issue 639 (7.9%) 72 (6.5%) 0 (0.0%) Substance Re: Candidate 1306 (163%) 23 (2.0%) 0 (0.0%) Campaign 3651 (453%) 791 (71.7%) 0 (0.0%) Horserace 1613 (20.0%) 217 (19.6%) 0 (0.0%) Anti-Opponent Rhetoric 842 (10.4%) 0 (0.0%) 0 (0.0%) NumbCT Original Stories: 39 Total Stories: 39 Number Wire Stories: 0 Period 4: April 8 to June 3,1992- Pre-Convention and General Election Maneuvering Average Number of words per day: 191 Party Democrats Republicans Independent Total Coverage in Words: (10,941) 3500 (31.9%) 4420 (403%) 3021 (27.6%) Area of Coverage Policy/Issue 76 (2.1%) 1525 (34.5%) 444 (14.6%) Substance Re: Candidate 312 (8.9%) 571 (12.9%) 1131 (36.4%) Campaign 1785 (51.%) 2027 (45.8%) 1204 09.8%) Horserace 1074 (30.6%) 230 (53%) 242 (8.0%) Anti-Opponent Rhetoric 253 (73%) 67 (1.5%) 0 (0.0%) Number Original Stories: 51 Total Stories: 51 Number Wire Stories: 0 312 Table R2: Substance of the Coverage for the Washington Post Period 1: December 1 ,1991 to February 18,1992 - Early Days Average Number of words per day: 186^ Party Democrats Republicans Total Coverage in words: (14,8%) 4286 (28.8%) 10,610 (71.2%) Area of Coverage Policy/Issue 224 (5,22%) 6143 (57.9%) Subskince Re: Candidate 2006 (46.8%) 2532 (23.8%) Campaign 1104 (25.7%) 1561 (14.7%) Horserace 628 (14.6%) 295 (2.7%) Anti-Opponent Rhetoric 325 (7.5%) 79 (0.74%) NumbCT Original Stories: 77 Total Stories: 77 Number Wire Stories: 0 Period 2: February 19 to March 10,1992 - Initial Contests Average Number of words per day: 328 Party Democrats Republicans Total Coverage in Words: (6888) 3321 (48.2%) 3567 (51.7%) Area of Coverage Policy/Issue 47 (1.4%) 817 (22.9%) Substance Re: Candidate 123 (3.7%) 161 (4.5%) Campaign 1503 (45,2%) 2189 (60% ) Horserace 689 (20.7%) 261 (73%) Anti-Opponent Rhetoric 859 (25.8%) 139 (3.9%) Number Original Stories: 36 Total Stories: 36 Number Wire Stories: 0 Period 3: March 11 to April 7,1992 - Mist Clearing Average Number of words per day: 301 Party Democrats Republicans Independent Total Coverage in Words:(8415) 5897 (70%) 2252 (26.7%) 266 (33%) Area of Coverage Policy/Issue 219 (3.7%) 652 (28.9%) 0 (0.0%) Substance Re: Candidate 2717 (46%) 464 (20.6%) 180 (67.6%) Campaign 1727 (293%) 921 (40,9%) 86 (33%) Horserace 927 (15.7%) 215 (9.5%) 0 (0.0%) Anti-Opponent Rhetoric 307 (53%) 0 (0.0%) 0 (0.0%) Number Original Stories: 36 Total Stories: 36 Number Wire Stories: 0 Period 4: April 8 to June 3,1992 - Pre-Convention and General Election Maneuvering Average Number of words per day: 172.7 Party Democrats Republicans Independent Total Coverage in Words: (9849) 2755 (27.9%) 4191 (42.5%) 2903 (29.5%) Area of Coverage Policy/Issue 285 (103%) 3096 (73.8%) 775 (26.6%) Substance Re: Candidate 563 (20.4%) 85 (2.0%) 895 (30.8%) Campaign 1224 (44,4%) 880 (20.0%) 904 (31.1%) Horserace 603 (21.8%) 130 (3.1%) 329 (113%) Anti-Opponent Rhetoric 64 (23%) 0 (0.0%) 0 (0.0«/o) ProOpponent Rhetoric 16 (0.58%) 0 (0.0%) 0 (0.0%) Number Original Stories: 53 Total Stories: 53 Number Wire Stories: 0 313 Table BJ: Substance of the Coverage for the Wall Street Journal Period 1: December 1,1991 to February 18,1992 - Earfy Days Average Number of words per day: 117 Party Democrats Republicans Total Coverage in words; (9346) 2277 (13.4%) 7069 (86.6%) Area of Coverage Policy/Issue 305 (13.4%) 5156 (719%) Substance Re: Candidate 747 02.8%) 572 (8.1%) Campaign 761 03.4%) 1304 (18.4%) Horserace 399 (17.5%) 37 (0.5%) Anti-Opponent Rhetoric 65 (2.8%) 0 (0.0%) Number Original Stories: 39 Total Stories: 39 Number Wire Stories: 0 Period 2; February 19 to March 10,1992 - Initial Contests Average Number of words per day: 174.5 Party Democrats Republicans Total Coverage in Words: (3665) 1478 (40.3%) 2187 (59.6%) Area of Coverage Policy/Issue 125 (8.4%) 850 (38.8%) Substance Re: Candidate 245 (16.5%) 0 (0.0%) Campaign 812 (54.9%) 1150 (515%) Horserace 252 (17.0%) 141 (6.4%) Anti-Opponent Rhetoric 44 (2.9%) 46 (2.1%) Number Original Stories: 19 Total Stories: 19 Number Wire Stories: 0 Period 3: March 11 to April 7,1992 - Mist Clearing Average Number of words per day: 150 Party Democrtits Republicans Independent Total Coverage in Words:(4197) 2947 (70J%) 1250 (29.7%) 0 (0.0%) Area of Coverage Policy/Issue 97 (33%) 730 (58.4%) 0 (0.0%) Substance Re: Candidate 707 (23.9%) 23 (1.8%) 0 (0.0%) Campaign 1199 (40.6%) 402 (32.1%) 0 (0.0%) Horserace 728 (24.7%) 89 (7.1%) 0 (0.0%) Anti-Opponent Rhetoric 127 (4.3%) 6 (0.48%) 0 (0.0%) Number Original Stories: 24 Total Stories: 24 Number Wire Stories: 0 Period 4: April 8 to June 3,1992 - Pre-Convention and General Election Maneuvering Average Number of words per day: 108 Party Democrats Republicans Independent Total Coverage in Words: (6157) 2211 (35.9%) 2748 (44.6%) 1198 (19.5%) Area of Coverage Policy/Issue 1101 (49.7%) 2078 (75.6%) 58 (4.8%) Substance Re: Candidate 153 (6.9%) 0 (0.0%) 40 (33%) Campaign 541 (24.4%) 556 (201%) 714 (593%) Horserace 370 (16.7%) 114 (4.1%) 358 (29.8%) Anti-Opponent Rhetoric 46 (2.0%) 0 (0.0%) 28 (23%) Number Original Stories: 38 Total Stories: 38 Number Wire Stories: 0 314 Table &4: Substance of the Coverage for the ABC World News Tonight Period I; December 1,1991 to February 18,1992 - Early Days Average Number of words per day; 432 Party Democrats Republicans Total Coverage in words: (34,622) 10377 (29.9%) 24345 (70.1%) Area of Coverage Policy/Issue 2416 (233%) 15313 (63.9%) Substance Re: Candidate 1853 (17.856) 1426 (5.856) Campaign 4371 (42.1%) 5896 (243%) Horserace 907 (8.7%) 750 (3.156) Anti-Opponent Rhetoric 830 (7.9%) 342 (1.4%) Misc. 0 (0.0%) 318 (13%) Number Original Stories: NA Total Stories: NA Number Wire Stories; NA Period 2: February 19 to March 10,1992 - Initial Contests Average Number of words per day: 778 Party Democrats Republicans Total Coverage in Words: (16,332) 9465 (57.8%) 6887 (42.1%) Area of Coverage Policy/Issue 866 (9.1%) 1214 (17.6%) Substance Re: Candidate 1009 (10.6%) 357 (53%) Campaign 5772 (60.9%) 3480 (503%) Horserace 1165 (123%) 1296 (18.8%) Anti-Opponent Rhetoric 653 (6.8%) 540 (7.8%) Number Original Stories: NA Total Stories: NA Number Wire Stories: NA Period 3: March 11 to April 7,1992 - Mist Clearing Average Number of words per day: 643 Party Democrats Republicans Independent Total Coverage in Words:(18,019) 14330 (793%) 3689 (20.456) 0 (0.056) Area of Coverage Policy/Issue 2177 (153%) 1947 (52.7%) 0 (0.056) Substance Re: Candidate 2317 (16.1%) 81 (2.1%) 0 (0.056) Campaign 7061 (493%) 1408 (38.1%) 0 (0.0%) Horserace 890 (63%) 158 (4.356) 0 (0.0%) Anti-Opponent Rhetoric 1885 (13.156) 95 (2.5%) 0 (0.056) Number Original Stories: NA Total Stories: NA Number Wire Stories: NA Period 4: April 8 to June 3,1992 - Pre-Convention and General Election Manuevering Average Number of words per day: 394 Party Democrats Republicans Independent Total Coverage in Words: (22,480) 9405 (41.8%) 8474 (37.6%) 4601 (20.5%) Area of Coverage Policy/Issue 1689 (17.956) 6410 (75.6%) 494 (10.7%) Substance Re: Candidate 2710 (28.856) 126 (1.4%) 1317 (28.6%) Campaign 3818 (403%) 1685 (19.856) 2338 (50.8%) Horserace 412 (43%) 253 (2.9%) 340 (73%) Anti-Opponent Rhetoric 776 (8.2%) 0 (0.0%) 112 (2.4%) NumbCT Original Stories: NA Total Stories: NA Number Wire Stories; NA 315 Table E5: Substance of the Coverage for the Des Moines Register Period I: December 1 ,1991 to February 18,1992 - Early Days Average Number of words per day: 123 Party Democrats Republicans Total Coverage in words: (9845) 6343 (6 4 .4 % ) 3502 (35.6%) Area of Coverage Policy/Issue 843 (13J%) 2519 (7 1 .9 % ) Substance Re: Candidate 1759 (27.7%) 78 (23%) Campaign 2648 (4 1 .7 % ) 689 (19.6%) Horserace 815 (12.8%) 107 (3.1%) Anti-Opponent Rhetoric 278 (43%) 109 (3.1%) Number Original Stories: 32 Total Stories: 50 Number Wire Stories: 18 Period 2: February 19 to March 10,1992 - Initiai Contests Average Number of words per day: 318 Party Democrats Republicans Total Coverage in Words: (6679) 5041 (7 5 .4 % ) 1638 (24.6%) Area of Coverage Policy/Issue 542 (10.7%) 524 (31.9%) Substance Re: Candidate 363 (73%) 88 (53%) Campaign 2879 (5 7 .1 % ) 721 (44 .0 % ) Horserace 1119 (223%) 305 (18.6%) Anti-Opponent Rhetoric 138 (2.7%) 0 (0.0%) Number Original Stories: 17 Total Stories: 25 Number Wire Stories: 8 Period 3: March 11 to April 7,1992 -Mist Clearing Average Number of words per day: 103 g Party Democrats Republicans Independent Total Coverage in Words:(2909) 1849 (63.6%) 918 (31.5%) 142 (4.9%) Area of Coverage Policy/Issue 81 (4.4%) 520 (56.6%) 0 (0.0%) Substance Re: Candidate 356 (193%) 0 (0.0%) 117 (82.4%) Campaign 683 (36.9%) 209 (22.7%) 25 (17.6%) Horserace 404 (21.8%) 189 (21.0%) 0 (0.0%) Anti-Opponent Rhetoric 265 (143%) 0 (0.0%) 0 (0.0%) Pro-Opponent Rhetoric 60 (33%) 0 (0.0%) 0 (0.0%) Total Locally Written Stories: 7 Total Stories: 16 Number Wire Stories: 9 Period 4: April 8 to June 3, 1992 - Pre-Convention and General Election Maneuvering Average Number of words per day: 71.8 Party Democrats Republicans Independent Total Coverage in Words: (4096) 1062 (25.9%) 1764 (43.1%) 1270 (31%) Area of Coverage Policy/Issue 0 (0.0T4) 1383 (78.4%) 127 (10%) Substance Re: Candidate 435 (40.9%) 27 (1.5%) 91 (73%) Campaign 299 (28.1%) 147 (83%) 1004 (79%) Horserace 274 (25.8%) 166 (9.4%) 48 (3.8%) Anti-Opponent Rhetoric 8 (0.75%) 41 (23%) 0 (0.0%) Pro-Opponent Rhetoric 46 (4.3%) 0 (0.0%) 0 (0.0%) Number Original Stories: 4 Total Stories'20 Number Wire Stories: 16 316 Table &6: Substance of the Coverage for the Manchester Union Leader Period 1: December 1,1991 to February 18,1992 Early Days Average Number of words per day; 146.4 Party Democrats Republicans Total Coverage in Words (11,712) 3957 (33.8%) 7755 (663%) Area of Coverage Policy/Issue 195 (4.92%) 3591 (46%) Substance Re: Candidate 1000 (253%) 330 (43%) Campaign 1890 (47.8%) 2764 (35.6%) Horserace 322 (8.13%) 542 (7.0%) Anti-Opponent Rhetoric 550 (13.9%) 528 (6.8%) Number Original Stories: 53 (563%) Total Stories: 94 Number Wire Stories: 28 (29.8%) Number Editorials: 13(13.8%) Period 2: February 19 to March 10,1992- Initial Contests Average Number of words per day; 139 Party Democrats Republicans Total Coverage in Words (2927) 1336 (45.7%) 1591 (543%) Area of Coverage Policy/Issue 0 (0.0%) 237 (14.9%) Substance Re: Candidate 33 (2.4%) 226 (143 %) Campaign 942 (70.5%) 807 (50.7%) Horserace 289 (21.6) 282 (17.7%) Anti-Opponent Rhetoric 72 (53%) 39 (2.5%) Number Locally Written Stories: 6 (28.6%) Total Stories: 21 Number Wire Stories: 9 (43%) Number Editorials: 6 (28.6%) Period 3: March 11 to April 7,1992 - Mist Clearing Average Number of words per day: 71 Party Democrats Republicans Independent Total Coverage in Words(1980) 1250 (63%) 658 (333%) 72 (3.6%) Area of Coverage Policy/Issue 20 (1.6%) 284 (43.1%) 0 (0.0%) Substance Re: Candidate 369 (29.5%) 54 (83%) 0 (0.0%) Campaign 533 (42.6%) 200 (30.4%) 35 (48.6%) Horserace 281 (22.5%) 81 (123%) 37 (51.3%) Anti-Opponent Rhetoric 47 (3.8%) 39 (13.5%) 0 (0.0»/o) Total Locally Written Stories: 1 Total Stories: 15 Number Wire Stories: 13 Total Editorials: 1 Period 4: April 8 to June 3,1992- Pre-Convention and General Election Maneuvering Average Number of words per day. 37.5 Party Democrats Republicans Independent Total Coverage in Words (2137) 367 (17.1%) 1189 (55.6%) 581 (27.1%) Area of Coverage Policy/Issue: 0 (0%) 480 (403%) 0 (0%.) Substance Re: Candidate: 28 (7.6%) 92 (7.7%) 117 (20.1%) Campaign: 192 (523%) 429 (36.1%) 360 (62%) Horserace: 147 (40.1%) 170 (143%) 104 (17.9%) Anti-Opponent 0 (0%) 0 (0%) 0 (0.0%) Pro-Opponent: 0 (0%) 18 (1.5%) 0 (0.0%) Total Locally Written Stories: 4 Total Editorials: 5 Number Wire Stories: 10 Total Stories: 19 317 Table B.7: Substance of the Coverage for the Bangor Daflv News Period I: December 1,1991 to February 18,1992 Average Number of words per day; 111 Party Democrats Republicans Total Coverage in Words(8850) 386 (4.4%) 8464 (95.6%) Area of Coverage Policy/Issue 24 (62%) 6653 (78.6%) Substance Re: Candidate 61 (15.8%) 339 (4%) Campaign 114 (29.5%) 1194 (14.1%) Horserace 144 (13.1%) 239 (5.6%) Anti-Opponent Rhetoric 121 01.1%) 96 (1.1%) Number Original Stories: 4 Total Stories:46 Number Wire Stories: 42 Period 2: February 19 to March 10,1992 - Initial Contests Average Number of words per day: 244.9 Party Democrats Republicans Total Coverage in Words (5143) 3561 (692%) 1582 00.7%) Area of Coverage Policy/Issue 85 (2.4%) 180 (11.4%) Substance Re: Candidate 260 (73%) 188 (11.9%) Campaign 1595 (44.7%) 634 (40.1%) Horserace 1187 033% ) 367 (23.1%) Anti-Opponent Rhetoric 430 (12.1%) 213 (13.5%) Number Original Stories; 9 Total Stories: 22 Number Wire Stories: 13 Period 3: March II to April 7,1992 - Mist Clearing Average Number of words per day: 122 Party Democrats Republicans Independent Total Coverage in Words (3418) 2535 (74.1%) 571 (16.7%) 312 (9.1%) Area of Coverage Policy/Issue 103 (4%) 218 08.1%) 0 (0%) Substance Re: Candidate 251 (9.9%) 0 (0%) 102 (32.7%) Campaign 1167 (46%) 229 (40%) 78 (25%) Horserace 834 02.9%) 124 (21.7%) 132 (42.3%) Anti-Opponent Rhetoric 162 (6.4%) 0 (0%) 0 (0%) Pro-Opponent Rhetoric 18 (0.7%) 0 (0%) 0 (0%) Total Locally Written Stories: 3 Total Stories: 20 Number Wire Stories: 17 Period 4: April 8 to June 3,1992 - Pre-Convention and General Election Maneuvering Average Number of words per day: 99.6 Party Democrats Republicans Independent Total Coverage in Words (5675) 995 (17.5%) 3140 (553%) 1540 (27.1%) Area of Coverage Policy/Issue; 43 (43%) 2600 (82.8%) 103 (6.7%) Substance Re: Candidate; 249 (25%) 0 (0%) 237 (153%) Campaign: 385 (38.7%) 387 (123%) 853 (55.9%) Horserace; 311 013% ) 153 (4.9%) 223 (14.5%) Anti-Opponent 7 (0.7%) 0 ((M4) 124 (8.1%) Total Locally Written Stories: 4 Total Stories: 26 Number Wire Stories: 22 318 Table B.8: Substance of the Coverage for the Atlanta Journal Constitution Period I: December 1 ,1991 to February 18,1992 Average Number of words per day; 76.8 Party Democrats Republicans Independent Total Coverage in words: (6144) 1459 (23.7%) 4367 (71.0%) 318 (5.1%) Area of Coverage Policy/Issue 82 (5.6%) 3519 (80.5%) 9 (2.8®/.) Substance Re: Candidate 641 (43.9%) 95 (2.2%) 13 (4.0%) Campaign 205 (14.1%) 726 (16.6%) 287 (90.2%) Horserace 389 (26.6%) 11 (0.25%) 9 (2.8%) Anti-Opponent Rhetoric 133 (9.1%) 16 (036%) 0 (0%) Number Original Stories: 19 Total Stories:41 Number Wire Stories: 22 Period 2: February 19 to March 10,1992 - Initial Contests Average Number of words per day: 89.6 Party Democrats Republicans Total Coverage in Words; (1883) 1170 (62.1%) 713 (37.8%) Area of Coverage Policy/Issue 25 (2.1%) 112 (15.7»4) Substance Re: Candidate 25 (2.1%) 0 (0%) Campaign 734 (62.7%) 511 (71.6%) Horserace 280 (23.9%) 67 (93%) Anti-Opponent Rhetoric 106 (9.0%) 23 (33%) Number Original Stories: 10 Total Stories: 11 Number Wire Stories: 1 Period 3: March 11 to April 7,1992 - Mist Clearing Average Number of words per day: 123.1 Party Democrats Republicans Independent Total Coverage in Words: (3446) 2266 (65.7%) 769 (223%) 411 (11.9“/o) Area of Coverage Policy/Issue 0 (0%) 574 (74.6%) 0 (0%) Substance Re: Candidate 744 (32.8%) 0 (0»/o) 0 (0%) Campaign 661 (29.2%) 29 (3.7%) 411 (100%) Horserace 557 (24.6%) 166 (21.6%) 0 (0%) Anti-Opponent Rhetoric 304 (13.4%) 0 (0%) 0 (0%) Total Locally Written Stories: 11 Total Stories; 19 Number Wire Stories: 8 Period 4: April 8 to June 3,1992 - Pre-Convention and General Election Maneuvering Average Number of words per day: 76.9 Party Democrats Republicans Independent Total Coverage in Words: (4385) 791 (18%) 2421 (553%) 1173 (26.7%) Area of Coverage Policy/Issue: 28 (3.5%) 1575 (65.1%) 10 (1.0"/o) Substance Re: Candidate; 252 01.8%) 4 (0.1%) 222 (18.9%) Campaign: 297 073% ) 665 (27.5%) 647 (553%) Horserace: 191 04.1%) 107 (4.4%) 286 (24.4%) Anti-Opponent 23 (2.9%) 70 (2.9%) 8 (0.6%) Total Locally Written Stories: 19 Total Stories: 32 Number Wire Stories: 13 319 Table B.9: Substance of the Coverage for the Salt Lake Tribune Period 1: December 1,1991 to February 18,1992 - Early Days Average Number of words per day. 186 Party Democrats Republicans Total Coverage in words: (14,917) 1644 (11.0%) 13;>73 (89.0%) Area of Coverage Policy/Issue 39 (2.4%) 10,416 (78.4%) Substance Re: Candidate 204 (12.4%) 261 (1.9%) Campaign 875 (53.2%) 2004 (15.0%) Horserace 438 (26.6%) 243 (1.8%) Anti-Opponent Rhetoric 88 (53%) 27 (030%) Miscellaneous 0 (0.0%) 322 (2.4%) Number Original Stories: 55 Total Stories: 59 Number Wire Stories: 4 Period 2: Februaiy 19 to March 10,1992 - Initial Contests Average Number of words per day 193 Party Democrats Republicans Total Coverage in Words: (4061) 2030 (49.9%) 2031 (50.0%) Area of Coverage Policy/Issue 76 (3.7%) 804 (39.5%) Substance Re: Candidate 85 (43%) 24 (13%) Campaign 1004 (49.4%) 559 (27.5%) Horserace 714 (35.1%) 393 (193%) Anti-Opponent Rhetoric 151 (7.4%) 251 (123%) Numbo’ Original Stories: 2 Total Stories: 21 Number Wire Stories: 19 Period 3: March 11 to April 7,1992 - Mist Clearing Average Number of words per day. 166 Party Democrats Republicans Independent Total Coverage in Words:(4650) 1551 (333%) 2927 (62.9%) 172 (3.6%) Area of Coverage Policy/Issue 0 (0.0%) 1952 (54.0%) 0 (0.0«<,) Substance Re: Candidate 268 (173%) 427 (14.5%) 17 (9.8%) Campaign 933 (60.1%) 372 (12.7%) 51 (29.6%) Horserace 328 (21.1%) 176 (6.0%) 87 (50.5%) Anti-Opponent Rhetoric 22 (1-4%) 0 (0.0%) 0 (0.0%) Number Original Stories: 1 Total Stories: 24 Number Wire Stories33 Period 4: April 8 to June 3,1992 - Pre-Convention and General Election Maneuvering Average Number of words per day 83 Party Democrats Republicans Independent Total Coverage in Words: (4754) 1036 (21.7%) 3055 (643%) 663 (13.9%) Area of Coverage Policy/Issue 27 (2.6%) 1972 (64.5%) 16 (2.4%) Substance Re: Candidate 120 (11.6%) 191 (63%) 0 (0.0%) Campaign 452 (43.6%) 655 (21.4%) 584 (88.1%) Horserace 297 (28.6%) 202 (6.6%) 63 (9.5%) Anti-Opponent Rhetoric 125 (12.1%) 35 (1.1%) 0 (0.0%) Pro-Opponent Rhetoric 15 (0.1%) 0 (0.0%) 0 (0.0%) Number Original Stories: 29 Total Stories: 30 Number Wire Stories: 1 320 Table B.10: Substance of the Coverage for the Miami Herald Period 1: December 1,1991 to Februaiy 18,1992 - Earfy Days Average Number of words per day: 66.8 Party Democrats Republicans Total Coverage in Words (5346) 1091 (20.4%) 4255 (79.6%) Area of Coverage Policy/Issue 0 (0%) 3327 (78.2%) Substance Re: Candidate 459 (42.1%) 410 (9.6%) Campaign 434 09.7%) 279 (6.5%) Horserace 144 (13.1%) 239 (5.6%) Anti-Opponent Rhetoric 54 (4.9%) 0 (0%) Numbw Original Stories: 30 Total Stories:39 Number Wire Stories: 9 Period 2: February 19 to March 10,1992 - Initial Contests Average Number of words per day 94.9 Party Democrats Republicans Total Coverage in Words (1993) 790 (39.6%) 1203 (60.4%) Area of Coverage Policy/Issue 43 (5.4%) 444 06.9%) Substance Re; Candidate 28 (3.5%) 98 (8.1%) Campaign 522 (66%) 376 01.2%) Horserace 186 (23.5%) 257 (21.4%) Anti-Opponent Rhetoric II (1.4%) 28 (23%) Number Original Stories: 21 Total Stories; 24 Number Wire Stories: 3 Period 3: March 11 to April 7,1992 - Mist Clearing Average Number of words per day. 70.5 Party Democrats Republicans Independent Total Coverage in Words (1973) 1156 (58.6%) 570 (28.9»"o) 247 (12.5%) Area of Coverage Policy/Issue 11 (1%) 101 (17.7%) 0 (0%) Substance Re; Candidate 222 (192%) 0 (0%) 0 (0%) Campaign 922 (79.8%) 364 (63.9%) 184 (74.5%) Horserace 226 (19.5%) 83 (14.6%) 63 (25.5%) Anti-Opponent Rhetoric 24 (2.1%) 0 (0%) 0 (0%) Total Locally Written Stories: 8 Total Stories: 14 Number Wire Stories: 6 Period 4: April 8 to June 3,1992 Average Number of words per day. 66.6 Party Democrats Republicans Independent Total Coverage in Words (3795) 411 (10.8%) 2905 (76.5%) 479 (12.6%) Area of Coverage Policy/Issue: 91 (22.1%) 1928 (66.4%) 66 (13.8%) Substance Re: Candidate: 0 (0%) 309 (10.6%) 136 (13.8%) Campaign: 206 (50.1%) 558 (192%) 23 (4.8%) Horserace: 94 (22.8%) 110 (3.8%) 77 (16.1%) Anti-Opponent 20 (024%) 0 (0%) 39 (8.1%) Total Locally Written Stories: 20 Total Stories: 26 Number Wire Stories:6 321 Table B.11: Substance of the Coverage for the Chicago Tribune Period 1: December 1,1991 to February 18,1992 Average Number of words per day: 158.9 Party Democrats Republicans Total Coverage in words: (12,716) 3971 (31%) 8475 (69%) Area of Coverage Policy/Issue 233 (6.0%) 5240 (60.0%) Substance Re: Candidate 1754 (44.8%) 841 (9.6%) Campaign 1569 (40.1%) 2059 (24.0%) Horserace 415 (10.6%) 436 (5.0%) Anti-Opponent Rhetoric 0 (0.0%) 169 (2.0%) Number Original Stories: 58 Total Stories:59 Number Wire Stories: 1 Period 2: February 19 to March 10,1992 - Initial Contests Average Number of words per day. 263.4 Party Democrats Republicans Total Coverage in Words: (5532) 3107 (56.2%) 2,425 (43.7%) Area of Coverage Policy/Issue 68 (22%) 369 (152%) Substance Re: Candidate 236 (7.6%) 173 (7.1%) Campaign 1994 (64%) 1225 (50.5%) Horserace 708 (22.8%) 476 (19.6%) Anti-Opponent Rhetoric 101 (32%) 182 (7.5%) Number Original Stories: 27 Total Stories: 28 Number Wire Stories: 1 Period 3: March 11 to April 7,1992 - Mist Clearing Average Number of words per day 239.5 Party Democrats Republicans Independent Total Coverage in Words:(6706) 4287 (63.9%) 2138 (31.9%) 281 (42%) Area of Coverage Policy/Issue 77 (1.8%) 1335 (62.4%) 0 (0.0«/o) Substance Re: Candidate 1380 (322%) 304 (142%) 49 (17.4%) Campaign 1556 (362%) 329 (14.4%) 188 (67.0%) Horserace 945 (22.0%) 170 (7.9%) 44 (15.6%) Anti-Opponent Rhetoric 329 (7.6%) 0 (0.0%) 0 (0.0%) Total Locally Written Stories: 31 Total Stories: 33 Number Wire Stories: 2 Period 4: April 8 to June 3,1992 - Pre-Convention and General Election Maneuvering Average Number of words per day: 91.6 Party Democrats Republicans Independent Total Coverage in Words: (5221) 1629 (312%) 2462 (47.1%) 1130 (21.6%) Area of Coverage Policy/Issue 107 (6.6%) 1256 (51%) 0 (0.0%)Substancc Re: Candidate 115 (7.0%) 96 (3.9%) 0 (0.0%) Campaign 926 (56.8%) 915 (372%) 816 (722%) Horserace 436 (26.7%) 130 (52%) 314 (27.8%) Anti-Opponent Rhetoric 45 (2.7%) 65 (2.6%) 0 (0.0%) Total Locally Written Stories: 23 Total Stories; 25 Number Wire Stories: 2 322 Table B.12: Substance of the Coverage for the Milwaukee Sentinel Period 1: December 1 ,1991 to February 18,1992 - Earfy Days Average Number of words per day: 81 Party Democrats Republicans Total Coverage in words: (6480) 1793 (27.6%) 4687 (72J%) Area of Coverage Policy/Issue 275 (153%) 3543 (75.5%) Substance Re; Candidate 467 (26.0%) 275 (5.8%) Campaign 604 (33.6%) 653 (135%) Horserace 232 (12.9%) 124 (2.6 %) Anti-Opponent Rhetoric 215 (11.9%) 92 (1.9%) Number Original Stories: 6 Total Stories: 30 Number Wire Stones: 24 Period 2: February 19 to March 10,1992 - Initial Contests Average Number of words per day: 94.7 Party Democrats Republicans Total Coverage in Words: (1990) 1235 (62%) 755 (37.9%) Area of Coverage Policy/Issue 21 (1.7%) 63 (8.3%) Substance Re: Candidate 47 (3.8%) 40 (53%) Campaign 769 (62.2%) 513 (67.9%) Horserace 336 (273%) 64 (8.4%) Anti-Opponent Rhetoric 53 (43%) 75 (9.9%) Pro-Opponent Rhetoric 9 (0.72%) 0 (0.0%) Number Original Stories: 0 Total Stories: 10 Number Wire Stories: Period 3: March 11 to April 7, 1992 - Mist Clearing Average Number of words per day: 1163 Party Democrats Republicans Independent Total Coverage in Words:(3257) 1878 (57.6%) 1379 (423%) 0 (0.0%) Area of Coverage Policy/Issue 117 (63%) 605 (43.8%) 0 (0.0%) Substance Re: Candidate 164 (8.7%) 0 (0.0%) 0 (0.0%) Campaign 761 (40.5%) 584 (423%) 0 (0.0%) Horserace 668 (35.5%) 124 (8.9%) 0 (0.0%) Anti-Opponent Rhetoric 168 (8.9%) 66 (4.7%) 0 (0.0%) Number Original Stories: 6 Total Stories: 17 Number Wire Stories: 11 Period 4: April 8 to June 3,1992 - Pre-Convention and General Election Maneuvering Average Number of words per day: 263 Party Democrats Republicans Independent Total Coverage in Words: (1495) 772 (51.6%) 326 (21.8%) 397 (26.5%) Area of Coverage Policy/Issue 171 (22.1%) 155 (47.5%) 0 (0.0%) Substance Re: Candidate 160 (20.7%) 0 (0.0%) 0 (0.0%) Campaign 211 (273%) 42 (12.8%) 384 (96.7%) Horserace 200 (25.9%) 129 (39.5%) 13 (33%) Anti-Opponent Rhetoric 0 (0.0%) 0 (0.0%) 0 (0.0%) Pro-Opponent Rhetoric 30 (3.8%) 0 (0.0%) 0 (0.0%) Number Original Stories: 3 Total Stories: 8 Number Wire Stories: 5 323 Table &13: Substance of the Coverage for the Newport News Dailv Press Period 1: December 1,1991 to February 18,1992 - Early Days Average Number of words per day: 112.2 Party Democrats Republicans Total Coverage in words: (8981) 2209 (24.5%) 6772 (75.4%) Area of Coverage Policy/Issue 345 (15.6%) 5941 (87.7%) Substance Re; Candidate 726 (32.8%) 54 (0.79%) Campaign 815 (36.8%) 530 (7.8%) Horserace 146 (6.6%) 247 (3.6%) Anti-Opponent Rhetoric 188 (8.0%) 0 (Omo) Number Original Stories: 5 Total Stories: 30 Number Wire Stories: 24 Editorials: 1 Period 2: February 19 to March 10,1992 - Initial Contests Average Number of words per day: 106 Party Democrats Republicans Total Coverage in Words: (2245) 870 (38.7%) 1375 (612%) Area of Coverage Policy/Issue 0 (0.0%) 435 Pl.6% ) Substance Re: Candidate 153 (17.5%) 158 (n.4%0 Campaign 352 (40.4%) 346 (25.1%) Horserace 230 (26.4%) 301 (21.8%) Anti-Opponent Rhetoric 135 (15.5%) 135 (9.8%) Number Original Stories: 0 Total Stories: 18 Number Wire Stories: 18 Period 3: March 11 to April 7,1992 - Mist Clearing Average Number of words per day. 178.7 Party Democrats Republicans Independent Total Coverage in Words: (5005) 2634 (52.6%) 1517 (303%) 854 (17.1%) Area of Coverage Policy/Issue 57 (22%) 860 (56.7%) 47 (5.5%) Substance Re: Candidate 808 (30.6%) 20 (13%) 175 (20.5%) Campaign 847 (32.1%) 350 (23.1%) 469 (54.9%) Horserace 617 (23.4%) 181 (11.9%) 163 (19.0%) Anti-Opponent Rhetoric 305 (11.6%) 83 (5.5%) 0 (0.0%) Pro-Opponent Rhetoric 0 (0.0%) 23 (1.5%) 0 (0.0%) Total Locally Written Stories: 2 Total Stories: 32 Number Wire Stories: 30 Period 4: April 8 to June 3,1992 - Pre-Convention and General Election Maneuvering Average Number of words per day; 77.6 Party Democrats Republicans Independent Total Coverage in Words: (4421 ) 851 (192%) 2516 (56.9%) 1054 (23.8%) Area of Coverage Policy/Issue 17 (1.9%) 2046 (813%) 274 (25.9 %) Substance Re: Candidate 65 (7.6%) 0 (0.0%) 288 (273%) Campaign 577 (67.8%) 300 (11.9%) 312 (29.6%) Horserace 192 (22.6%) 170 (6.7%) 147 (13.9%) Anti-Opponent Rhetoric 0 (0.0%) 0 (0.0%) 33 (3.1%) Total Locally Written Stories:3 Total Stories: 51 Number Wire Stories: 48 324 APPENDIX C VARIABLE LIST 325 Over Time Variables Early Contest Week: defined as those primaries and caucuses occurring prior to March 10, 1992. Contest weeks occurring prior to March 10 are coded 1, and all others are coded 0. This variable is used to explain increases in news outlets’ weekly campaign coverage volume over a period of 25 weeks. Critical Contest Week: defined as weeks during which primaries and caucuses occur which have many delegates at stake, are historically important, or have become critical due to competitive nature of the campaign (i.e. South Dakota and Florida in 1992 for the Democratic campaign). Critical contest weeks are code 1, and all others are coded 0. This variable is used to explain increases in news outlets’ weekly campaign coverage volume over a period of 25 weeks. Weekly Campaign Coverage: defined as the percentage of fi-ont page coverage given to the campaign in general relative to total fi'ont page coverage of the news outlet (for newspapers) or the entire evening news broadcast. Pre-event Coverage Variables Local Coverage: defined as a percentage of pre-event (7 and 14 days prior) fi’ont page coverage appearing in a local news outlet given to a candidate relative to front page coverage given to other candidates. This variable ranges fi’om 0 to 100 percent. National Coverage: defined as a percentage of pre-event (7 and 14 days prior)fi’ont page coverage appearing in a national news outlet given to a candidate relative to front page coverage (and entire network evening news coverage) given to other candidates. This variable ranges from 0 to 100 percent. Local Positive Coverage: defined as a percentage of pre-event (7 and 14 days prior) front page positive coverage appearing in the local news outlet given to a candidate relative to the total positive coverage received by all candidates. This variable ranges from 1 to 100 percent. Local Negative Coverage: defined as a percentage of pre-event (7 and 14 days prior) front page negative coverage appearing in the local news outlet given to a candidate relative to the total negative coverage received by all candidates. This variable ranges from 1 to 100 percent. 326 National Positive Coverage: defined as a percentage of pre-event (7 and 14 days prior) front page coverage (and entire network evening news coverage) appearing in the national news outlet given to a candidate relative to the total positive coverage received by all candidates. This variable ranges fi’om 1 to 100 percent. National Negative Coverage: defined as a percentage of pre-event (7 and 14 days prior) fi’ont page negative coverage (and entire network evening news coverage) appearing in the local news outlet given to a candidate relative to the total negative coverage received by all candidates. This variable ranges fiom 1 to 100 percent. Control Variables Candidate attention: defined as the appearance of activity by a candidate and/or that candidate’s surrogates in the state. This variable is operationalized as a dummy variable coded 1 if a candidate and his surrogates devoted substantial time to a state relative to his opponents, and 0 if a candidate did not. Relative spending: defined as a percentage of candidate spending in a state relative to total spending of all active candidates competing in the state. This variable ranges from 0 to 100 percent. Absolute spending: defined as the total spending in dollars in a state by a candidate. Viability: defined as the likelihood that a candidate will win the nomination. This variable is operationalized as the number of delegates accumulated prior to a contest relative to half the number required for nomination. Momentum: defined as a short-term indication of success, i.e. winning a contest or group of contest immediately prior to an upcoming contest. Momentum is operationalized as a weighted dummy variable coded 1 and weighted by the number of contests won if a candidate won a prior contest or group of contests, and zero if not. Home State: defined as a candidate’s home state. Home State is operationalized as a dummy variable coded 1 if the state is considered the home of the candidate and 0 if not. Home Region: defined as a candidate’s home region. Home region is operationalized as a dummy variable coded I if that state borders a candidate’s home state, and 0 if not. Border states that are also the home state of an opponent are not considered Home Region states. 327 Competitive Field: defined as the number of candidates competing. The variable operationalized as the actual number of candidates competing. For the Democratic campaign this variable ranged fi’om 2 to 5. The variable does apply to the Republican campaign which had only two competitors throughout. Local Competitive Status: defined as a candidate being a local fi-ontrunner or runner-up. This variable is operationalized as a dummy variable coded 1 if the candidate is running first or a close second in a state, and zero if not. Incumbency: defined as a candidate being the incumbent president. Incumbency is operationalized as a dummy variable coded 1 if the candidate is an incumbent and 0 if not. Scandal: defined as the occurrence of a breaking scandal associated with a candidate the week prior to a primary or caucus. This variable is operationalized as a dummy variable coded 1 if such a scandal did occur and zero if not. Frontrunner: defined as the status of leading all other candidates with regard to viability, momentum and national poll numbers. This variable is operationalized as a dummy variable coded 1 if a candidate is the fi-ontrunner and 0 if not. National Poll Standing: defined as a candidate’s percentage of support in national polls regarding who individuals would support as the nominee of that party. This variable is operationalized as the percentage of support a candidate received in the national polls. This variable ranges from 0 to 100 percent. Halfway: defined as the point in the campaign when one of the candidates reaches the halfway point in delegate accumulation. Halfway is operationalized as a dummy variable coded I for all candidates when the halfway point is reached, and zero for all cases prior to that time. 328 BIBLIOGRAPHY Aldrich, J. H. 1980. Before the Convention: Strategies and Choices in Presidential Nomination Campaigns. Chicago and London: University of Chicago Press. Ansolabehere, Stephen, Roy Behr, Shanto Iyengar. 1993. The Media Game: American Politics in the Television Age. New York: Macmillan Publishing Company. Arterton, F. Christopher. 1992. Campaign ‘92: Strategies and Tactics, in The Election of 1992: Reports and Interpretations. Gerald M. Pomper, Ed. Chatham, New Jersey: Chatham House. . 1984. Media Politics: The New Strategies of Presidential Campaigns. Lexington, MA: Lexington Books. . 1978. Campaign Organizations Confront the Media-Political Environment. In: James D. Barber, ed. The Race for the Presidency. Englewood Cliffs, N.J.: Prentice-Hall, pp. 32-33. Bartels, Larry M. 1996. Politicians and the Press: Who Leads, Who Follows. Paper presented at the Annual Meeting of the American Political Science Association, San Francisco, September. . 1993. Messages Received: The Political Impact of Media Exposure. American Political Science Review 79: 804-15. . 1988. Presidential Primaries and the Dynamics of Public Choice. Princeton, NJ: Princeton Press. . 1985. Expectations and Preferences in Presidential Nominating Campaigns. American Political Science Review 79: 804-15. Batlin, Robert. 1954. San Francisco Newspapers’ Campaign Coverage: 1896, 1952. Journalism Quarterly 31: 297-303. Berelson, Bernard. 1952. Content Analysis in Communication Research. Glencoe, IL: Free Press. 329 Buchanan, Bruce. 1978. The Presidential Experience: What the Office Does to the Man. Englewood Clififs, New Jersey: Prentice-Hall. Brady, H. E. and R. Johnston. 1987. What’s The Primary Message: Horse Race or Issue Journalism? in Gary Orren and Nelson W. Polsby, eds. Media and Momentum. Chatham, New Jersey: Chatham House. Buell, Emmet H. Jr. 1987. ‘Locals’ and ‘Cosmopolitans’: National, Regional and State Newspaper Coverage of the New Hampshire Primary. In Media and Momentum: The New Hampshire Primary and Nomination Politics, ed. Gary R. Orren and Nelson W. Polsby. Chatham, NJ: Chatham House Publishers, Inc. Chaffee, Steven H., Xinshu Zhao and Glenn Lesfaner. 1994. Political Knowledge and the Campaign Media of 1992. Communication Research 21: 305-324. Clarke, Peter and Susan Evans. 1983. Covering Campaigns: Journalism in Congressional Campaigns. Stanford. CA: Stanford University Press. Cook, F. L., T.R. Tyler, E.G. Goetz, M.T. Gordon, D. Protess, D. R. Layoff and H.H. Molotch. 1983. Media and Agenda Setting: Effects on the Public, Interest Group Leaders and Policy Makers and Policy. Public Opinion Quarterly 47: 16-35. Crouse, Timothy. 1973. The Boys on the Bus. New York: Ballentine. Danielson, Wayne A and John B. Adams. 1961. Completeness of Press Coverage of the 1960 Campaign. Journalism Quarterly 38:441-452. Drew, Dan and David Weaver. 1993.Voter Learning in the 1990 Off-Year Election: Did the Media Matter? Journalism Quarterly 70:356-368. . 1991. Voter Learning in the 1988 Presidential Election: Did the Debates and the Media Matter? Journalism Quarterly 68: 27-37. Entman, Robert M. 1989. How the Media Affect What People Think: An Information Processing Approach. Journal o f Politics 5\\ 347-70. Evarts, Dm and Guido H. Stempel m . 1974. Coverage of the 1972 Campaign by TV, News Magazines and Major Newspapers. Journalism Quarterly 51: 645-652. Graber, Doris. 1994. Do the Media Inform? Yes. In Controversial Issues in Presidential Selection, ed. Gary L. Rose, New York: State University of New York Press. 330 . 1993. Mass Media and American Politics. 4th Edition. Washington, D.C.: Congressional Quarterly Press. . 1988. Processing the News: How People Tame the Information Tide, 2nd edition. New York: Longman. . 1976. Effect of Incumbency on Coverage Patterns in the 1972 Presidential Campaign. Journalism Quarterly 53:499-508. .1971. Press Coverage of Campaign News: The 1968 Presidential Race. Journalism Quarterly 48: 502-512. Gurian, Paul-Henri. 1993. Primaries versus Caucuses: Strategic Consideration of Presidential Candidates. Social Science Quarterly 74: 310-321. and John A. Wolfe. 1992. Regional Primaries and Presidential Campaign Coverage. The Social Science Journal 31: 17-30. . 1991. Less than Expected: An Analysis of Media Coverage of Super Tuesday. Social Science Quarterly 72: 761-851. and Doug Barthlow. 1991. Structural Bias in News Coverage of Presidential Candidates. Unpublished manuscript. Hagen, Michael. 1992. Press Treatment of Presidential Front-Runners. Paper presented at the Midwest Political Science Association, Chicago, Illinois. Hart, Jim A. 1965. Election Campaign Coverage in English and U.S. Daily Newspapers. Journalism Quarterly AT. 213-218. Higbie, Charles E. 1954. Wisconsin Dailies in the 1952 Campaign: Space vs. Display. Journalism Quarterly 31: 56-60. Hofstetter, Richard C. 1981. Content Analysis. In Handbook of Political Communication. Ed. Dan D. Nimmo and Keith R. Sanders. Beverly Hills, CA: Sage, pp. 562-569. . 1976. Bias in the News: Network Television Coverage of the 1972 Election Campaign. Columbus, OH: Ohio State University Press. Houston, Frank. 1996. The Virtual Trail. Columbia Journalism Review. January/February, pp. 26-28. 331 Iyengar, Shanto and Stephen Ansolabehere. 1995. Going Negative: How Political Advertisements Shrink and Polarize the Electorate. New York: Free Press. and Donald R_ Kinder. 1987. News That Matters. Chicago: University of Chicago Press. and D. Peters, and Donald R. Kinder. 1982. Experimental demonstrations of the Not-So-Minimal Consequences of Television News Programs. American Political Science Review 76: 848-858. Joslyn, Richard. 1984. Mass Media and Elections. Reading, MA: Addison-Wesley. Just, Marion R., Aim M. Crigler, Dean E. Alger, Timothy E. Cook, Montague Keen and Darrell M. West. 1996. Crosstalk. Chicago: University of Chicago Press. Katz, Elihu and Paul P. Lazarsfeld. 1955. Personal Influence. The Part Plaved bv People in the Flow of Mass Communications. New York: The Free Press. Keeter, Scott and Cliff Zukin. 1983. Uniformed Choice: The Failure of the New Presidential Nominating System. New York: Praeger. Kessel, John Howard. 1988. Presidential Campaign Politics: Coalition Strategies and Citizen Response. 3rd edition. Homewood, IL: Dorsey Press. Kinder, Donald R. and David. O. Sears. 1985. In Gamer Lindzey and Elliot Aronson, eds.. Public Opinion and Political Action in the Handbook of Political Socialization Volume n . New York: Random House. Klapper, J. T. 1960. The Effects of Mass Communication. Clencoe, Illinois: The Free Press. Krippendorf, Klaus. 1980. Content Analysis: An Introduction to its Methodology. Beverly Hills, CA: Sage. Larson, Patricia. 1996. Media Coverage of the White House in Transition. Unpublished Manuscript. University of Georgia. Lazarsfeld, Paul F., Bernard Berelson, and Hazel Gaudet. 1948. The People's Choice. (2nd Ed.) New York: Columbia University Press. Leroy, David J. And Christopher H. Sterling. 1973. Mass News: Practices, controversies and alternatives. Englewood Cliffs, New Jersey: Prentice-Hall, Inc. 332 Lippman, Walter. 1922. Public Opinion. New York; Free Press. Loevy, Robert. 1995. The Flawed Path To The Presidency 1992: Unfairness and Inequality in the Presidential Selection Process. Albany, NY: State University of New York Press. Malaney, Gary D. and Terry F. Buss. 1979. AP Wire Reports vs. CBS TV News Coverage of a Presidential Campaign. Journalism Quarterly 56: 602-610. Markham, James W. and Guido H. Stempel in. 1957. One-Party Press Charge Suffers Firm Setback in Pennsylvania Study. The Q uill April: 13-21. . 1957. Analysis of Techniques in Measuring Press Performance. Journalism Quarterly 3^: 187-190. Marshall, Thomas R. 1981. Presidential Primaries in a Reform Age. New York: Praeger. Matthews, Donald R. 1978. Winnowing. In James David Barber, ed.. Race for the Presidency. Englewood Cliffs, New Jersey: Prentice Hall. McCombs, M.E. and D. Shaw. 1972. The Agenda-setting Function of the Mass Media. Public Opinion Quarterly 36: 176-87. McGillivray, Alice V. 1992. Presidential Primaries and Caucuses 1992: A Handbook of Election Statistics. (Washington D C.: Congressional Quarterly Press) McGuire, W. J. 1985. Attitudes and Attitude Change. In E. Aronson and G. Lindzey, eds. Vol. 1, 3rd ed. New York: Random House. McKuen, M. 1984. Exposure to Information, Brief Integration, and Individual Responsiveness to Agenda Change. American Political Science Review 78: 372- 91. Meadow, Robert G. 1973. Cross-Media Comparison of Coverage of the 1972 Presidential Campaign. Journalism Quarterly 50: 482-488. Mendelsohn, Matthew. 1996. The Media and Interpersonal Communications: The Priming of Issues, Leaders and Party Identification. Journal o f Politics, Vol. 58. No. 1, pp. 112-25. Neuman, W. R., and A. C Fryling. 1985. Patterns of Political Cognition. In Sidney Kraus and R.M. Perloflf eds.. Mass Media and Political Thought. Beverly Hills: Sage. 333 Page, Benjamin I., Robert Y. Shapiro, and G. R. Dempsey. 1987. What Moves Public Opinion? American Political Science Review 81; 23-43. Paolina, Philip. 1996. Perceptions of Candidate Viability: Media Eflfects During the Presidential Nomination Process. Political Methodology Working Paper. Patterson, Thomas E. 1993. In I-800-President: The Report on the Twentieth Century Fund Task Force on Television and the Campaign o New f 1992. York: Twentieth Century Fund. . 1988. The Mass Media Election: How Americans Choose Their President 3rd. edition. New York: Preager Publishers. and Richard Davis. 1985. The Media Campaign: Struggle for the Agenda. In Micheal Nelson, ed.. The Election of 1984. Washington: Congressional Quarterly Press. and Robert D. McClure. 1976. The Unseeing Eye: The Myth of Television Power in National Elections. New York: G.P. Putnam’s Sons. Patterson, Thomas E. and Robert D. McClure. 1974. Political Advertising: Voter Reaction to Televised Political Commercials. Princeton, New Jersey: Citizen’s Research Foundation. Pavlik, John and Andras Szanto. 1994. Multiple-Method Research: The Case of the 1992 Presidential Campaign. Special Report for the Freedom Forum Media Studies Center, Columbia University, New York, New York. Peterson, Thomas. 1956. The Social Responsibility Theory of the Press, in F. Siebert, Thomas Peterson and William Schramm, ols. Four Theories of the Press. Urbana, Illinois. Polsby, Nelson W. 1981. The Democratic Nomination. In The American Elections of 1980. ed. Austin Ranney. (Washington and London: American Enterprise Institute), pgs. 37-60. Repass, Donald E. and Steven H. Chaflfee. 1968. Administrative vs. Campaign Coverage of Two Presidents in Eight Partisan Dailies. Journalism Quarterly 45: 528-531. Riffe, Daniel, Stephen Lacy, Jason Nagovan and Larry Burkum. 1995. The Effectiveness of Simple and Stratified Random Sampling in Broadcast News Content Analysis. Journalism and Mass Communication Quarterly 73: 159-168. 334 , Charles F. Aust, and Stephen Lacy. 1993. The Effectiveness of Random, Consecutive Day and Constructed Week Sampling in Newspaper Content hna!iys\s. Journalism Quarterly 10: 133-139. Robinson, Michael. J. and Margaret A. Sheehan. 1983. Over the Wire and on TV: CBS and UPI in Campaign ‘80. New York, NY: Russell Sage. Rose, Richard 1974. Comparability in Electoral Studies. In Richard Rose, ed.. Electoral Behavior: A Comparative Handbook. New York: Free Press. Rosenstiel, Tom. (1994) Strange Bedfellows: How Television and the Presidential Candidates Changed American Politics. 1992. New York: Hyperion. Stempel, Guido H. in . 1955. Increasing Reliability in Content Analysis. Journalism Q uarterly 32: 449-455. . 1961. The Prestige Press Covers the 1960 Campaign.Journalism Quarterly 62: 157-163. and John W. Windhauser. 1984. The Prestige Press Revisited: Coverage of the 1980 Presidential Campaign. Journalism Quarterly 61: 49-55 Stimson, J. A. 1991. Public Opinion in America: Moods. Cvcles and Swings. San Francisco: Westview Press. Weaver, David and Dan Drew. 1995. Voter Learning in the 1992 Presidential Election: Did the “Nontraditional” Media and Debates Matter? Journalism and Mass Communication Quarterly, 72: 7-17. , Doris Graber, M. E. McCombs and C.H. Eyal. 1981. Media Agenda Setting in a Presidential Election. New York: Praeger. Zaller, John. 1992. The Nature and Origins of Mass Opinion. Cambridge/New York: Cambridge University Press. Zhao, Xinshu and Glen L Bleske. 1995. Measurement Eflfects in Comparing Voter Learning From Television News and Campaign Advertisements. Journalism and Mass Communication Quarterly 72: 72-83. , and Steven H. Chaflfee. 1995. Campaign Advertisements versus Television News as Sources of Political Issue Information. Public Opinion Quarterly 59: 41- 65. 335