FDR to Clinton, Mueller to ?: a Field Essay on Presidential Approval Author(S): Paul Gronke and Brian Newman Source: Political Research Quarterly, Vol

University of Utah

FDR to Clinton, Mueller to ?: A Field Essay on Presidential Approval Author(s): Paul Gronke and Brian Newman Source: Political Research Quarterly, Vol. 56, No. 4 (Dec., 2003), pp. 501-512 Published by: Sage Publications, Inc. on behalf of the University of Utah Stable URL: http://www.jstor.org/stable/3219810 Accessed: 21/06/2010 14:03

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=sage.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

University of Utah and Sage Publications, Inc. are collaborating with JSTOR to digitize, preserve and extend access to Political Research Quarterly.

http://www.jstor.org FIELDESSAY

FDR to Clinton, Mueller to ?: A Field Essay on Presidential Approval

PAUL GRONKE, REEDCOLLEGE BRIAN NEWMAN, DUKEUNIVERSITY

Since the 1930s, polling organizations have asked Americans whether they "approveor disapprove of the job [the incumbent]is doing as president."In the early 1970s, John Muellerstarted an academicindustry by askingwhat drivesthese evaluations.American politics and the tools availableto examineit have changeddra- maticallysince then, inspiringa burst of researchon presidentialapproval in the 1990s. We reviewthis new body of literature, arguing that it builds on but differs importantly from earlier approval studies. Since Mueller's writing,scholars have expandedhis relativelysimple model, takingaccount of presidents'goals and personal characteristics,other political actors, the ubiquitousmedia, and an inattentivepublic. We describethree waves of research,focusing on the most recentwave. We suggestthat history,along with new intellectualcurrents, data,and methodshave enabledeach wave to incorporatemore of political,social, and psychologicalreality. Finally,we identifythe issues most likely to motivatepresidential approval research for the next ten years.

In the 1930s, the Gallup organization began asking study of the Americanpresidency, Richard Neustadt (1980: Americans "do you approve or disapprove of the way 81, n. 9) arguedthat reportsof these ratings"are very widely [the incumbent] is handling his job as president?"Since read in Washington"and are "widelytaken to approximate then, the question has been asked "with tenacious regular- reality."Higher approvalratings tend to pay off electorally, ity" (Mueller 1973: 196). In the early 1970s, John Mueller both for the president and for his party in Congress (1970, 1973) sparked something of a political science (Gronke, Koch, and Wilson 2003; Newman and Ostrom movement when he treatedthe Gallupapproval ratings as a 2002; Sigelman 1979) and also affectthe president'spolicy- and success in dependent variable. Coalitions of minorities, rallying making goals, legislativestrategy, promoting aroundthe flag, economic decline, and war, Muellerargued, his agenda.2 drove changes in approvalover time. The American political universe has undergone impor- Since then, scholarly studies of presidential approval tant changes since the early 1970s, and the presidency has has have been almost equally tenacious and regular.In fact, a not been immune. The office of the president endured conservative estimate finds over 70 books, articles, and two major crises. RichardNixon resigned, almost certain to chapters that attempt to explain approval ratings.' The be impeached, and left office with 24 percent approval. attention is well placed. In the era of "the public relations William Jefferson Clinton was impeached, yet ended his presidency"(Brace and Hinckley 1993: 382), approvalrat- second term with a 64 percent approvalrating, the highest on In has ings play a critical role in presidentialpolitics. The presi- final rating record. addition, presidentialapproval dent's performance in this "new referendum"(Brace and become increasingly volatile. One President Bush experi- after first Gulf Hinckley 1992: 18) is a key to understandingpresidential enced almost universal acclaim the War, power in the postwar era (Neustadt 1980). In his classic only to watch this support whither by over 50 percentage points in a matter of months. A second PresidentBush was

The actualnumber is certainlyhigher. We included only a subset of these studies in the bibliography.We apologizein advancefor the many books, 2 These claims are established in the voluminous literatureon elections, articles,and workingpapers that we have omitteddue to spacerestrictions. presidentialapproval, and presidentialleadership. Four recent volumes NOTE: An earlierversion of this articlewas presentedat the 2000 Annual include Brody (1991), Braceand Hinckley (1992), Kernell (1997), and Meeting of the AmericanPolitical Science Association, Washing- King and Ragsdale(1988). On how popularityaffects legislative strategy, ton, D.C. The authors thank Matt Baum, RichardBrody, Harold see Canes-Wroneand Shotts (2002); on legislative success, see Canes- Clarke,Jeffrey Cohen, RobertEisinger, Constantine Spiliotes, and Wrone and de Marchi (2002). Additional treatments include Kernell anonymous reviewers for their comments and Carrie Liken and (1997); Peterson (1990); Rivers and Rose (1985); Rohde and Simon Jennifer Merolla for their research assistance. This research was (1985). Many argue that those effects are variableand not always sub- supported in part by the National Science Foundation SBR- stantial (e.g., Bond and Fleisher 1990; Cohen 1997). We consider the 9730854, the NSF Research Experiences for Undergraduates debate over approval'sinfluence on Americanpolitics a significanttopic, (REU)program, and the Stillman-DrakeFund at Reed College. one that makes studying approvalas a dependent variablemeaningful. However,we cannot do this topic justice in the space providedand focus PoliticalResearch Quarterly, Vol. 56, No. 4 (December 2003): pp. 501-512 on literaturetaking presidentialapproval as the dependent variable.

501 502 POLITICAL RESEARCH QUARTERLY catapulted to 90 percent approval just after the terrorist the quarter century since. Despite his sparse and self- attacks of September 11th and retained remarkablyhigh described "irreverent"approach, Mueller anticipatedmany approvalfor more than a year afterwards.Further, the "ava- of the ways his model would be extended, refined, and lanche of opinion polls" (Jacobsand Shapiro2000, xi) pro- enriched.Analytically, he tried to account for the seemingly vides an almost daily rendering of the president'sapproval inevitable decline in approval over each presidentialterm. rating. During the Lewinsky scandal, for instance, John He arguedthat this was most likely the result of a "coalition Zaller(1998) examined more than 30 differentpolls over a of minorities"that builds during an administrationas the 20 day period, all of which dealt with Clinton'sjob per- president is forced to act on controversialissues, conse- formance. Presidents'private polling has also dramatically quently alienatinggroups of real or potentialsupporters. He increased, as has scholarly attention to these activities operationalizedthe "coalitionof minorities"thesis simply as (Jacobsand Shapiro2000). time in office. Second, he noticed that approval increases In the academicworld, new analyticaltools have shaped during foreign crises, the "rally round the flag" effect.4 approvalstudies. Longer time series have provided greater Muellerwas the first,but certainlynot the last, to realizethat leverage on comparisons across administrations. Rich measuringrally events is fraughtwith difficultyTo resist the datasetsderived from presidentialarchives have provided a temptation to find a rally point to match every bump in far more detailed treatmentof presidentialagendas. Increas- approval, fitting the model to sample data, he adopted a ingly sophisticated analytical methods have been intro- priori criteria for rally events, a recommendationthat has duced and applied to these new data sources, as well as pro- been followed irregularlysince.5 Third, Muellerargued that vided a fresh look at the traditionalGallup series. "aneconomy in slump harmsa president'spopularity, but an Scholarly scrutiny of the presidency and presidential economy that is improvingdoes not seem to help his rating" approvalhas intensifiedin response to these developments (1973: 215).6 Finally,given his interest in public opinion and especiallyin reactionto the GeorgeH. W Bushand Clin- during wartime, he included dummy variables for the ton presidencies.What has not occurred,however, is a com- KoreanWar and Vietnamconflict. Ultimately,he found that prehensiveoverview and criticalexamination of the literature all of these variablessignificantly affected approval.7 during this same period. We provide this here. We trace the When Mueller initially estimated his model, only the development of the study of presidentialapproval over the time variable was statistically or substantively significant. last 30 years, from Mueller through the present. We argue The expected relationships with war, the economy, and that by and large, students of approvalhave taken Mueller's other rally events appeared only once he added a dummy self-described"rather austere representation of a presumably variable for each administration(or a presidency-specific complex process"and put it into a richer context, taking time counter). This simple finding turns out to be a funda- account of presidents' goals and personal characteristics, mental issue in the study of approvalratings and in presi- other political actors, the ubiquitous media, and a public dency research generally: how do we resolve the tension made up of generallyinattentive citizens (1973: 217-18). We between elegantmodels that capturesystematic forces active describethree waves of research,but focus on the most recent across presidencies and the harsh empirical realities of wave, which has emergedover the past decade.3We suggest importantidiosyncratic features that characterizeindividual that history,new intellectualcurrents from the broaderpublic presidents? Mueller'sposition on this debate is clear: he opinion field, and new data and methods have enabled each claimed that "anyanalysis of presidentialpopularity cannot rely wave to take more of political,social, and psychologicalreal- entirelyon the [substantive]variables.. ., but must also incor- ity into account. Finally,we identify the questions and puz- porateparameters designed to allowfor the specialcharacter of zles which are most likely to motivatepresidential approval each administration"(1973: 222, italics in original). He researchfor the next ten years. embracedwhat he considered the more empiricallyplausi- ble model, taking into account idiosyncratic and unmea- THE THREE WAVES OF APPROVALRESEARCH sured effects of each presidency with the lament "so much for beauty"(ibid.). TheFirst Wave-Mueller and Reactions

The first wave of researchconsists of Mueller'swork and 4 Muellercites KennethWaltz as the originatorof the idea that the public immediatereactions to it. Mueller'sinitial article(1970) and will "rallybehind their chief executive"(1973: 208) during crises, but he his elaborationon it in his book (1973: chs. 9-10) identified is the first to our knowledge to use the term "rallyround the flag"to from a well-known Civil much of the that has studies in describe it. As best we can tell, the term comes agenda occupied approval War song. 5 Events had to be international,to involve the U.S. and particularlythe president directly,and to be "specific, dramatic, and sharply focused" 3 Naturally,breaking the literatureinto three waves oversimplifiesreality (1973: 209). and draws sharp boundaries where they do not exist. Many pieces fit 6 However, he honestly notes that coding his economic variable so that imperfectlyin this structure,anticipating new trends or hearkeningback only increasesin unemploymentand not decreasescould affectapproval to earlierquestions and methods. We offer this view of the literaturenot was "theonly way the data can be made to come out 'right"'(1973: 215). so much as a mutuallyexclusive and exhaustivecategorical analysis, but 7 Mueller found that approvaldeclined during the KoreanWar, but not as a bit of intellectualhistory and an organizingframework. significantlyduring the Vietnamconflict. FDR TO CLINTON,CLiNTON,MUELLER TOTO? ? 503

Although his final model explained 86 percent of the argued, "as a variable has no inherent meaning" and is variancein approval,Mueller pointed to severalpuzzles and simply a convenient statistical shorthand for real world paths for future research.First, he suggested that the rally forces, such as the economy, wars, Watergate,and interna- variablecould be expanded (1973: 238). He experimented tional events (508). After operationalizingthese factors in with a domestic rally variable, a variabletapping scandals, ways that "reflectmore reasonable assumptions about the and splitting events into positive and negative categories. world," he found approval responds to events and condi- Although he found these experiments unfruitful, he sug- tions ratherthan the marchof time (521). Similarly,Monroe gested that "more precise social, political, and economic (1978) argued that Stimson underestimatedthe effects of indicatorscan be sought"(1970: 33). Second, he suggested economic variablesvia the use of a time counter. She found that greaterattention to individual presidents might prove that when inflation and military expendituresare included beneficial.Eisenhower's approval did not fit the coalition-of- in the model, and their effects estimated in appropriately minorities thesis and Mueller conjectured that something nuanced ways, approvalreacts to these "realworld" indica- about him in particularmay account for the finding. He tors. Both articles tied approvalto elements of the political argued, tongue in cheek, that "if a president wants to leave and economic context and sought to measure and estimate office a popular man, he should either (1) be Dwight Eisen- their effectson approvalmore realistically,efforts that would hower, or (2) resign the day afterinauguration" (1973: 233). continue in later waves. He suggested that taking account of individual presidents' attributes,like their style, integrity,competence, and "per- TheSecond Wave-Advancements in Specification sonal warmth,"might explain the Eisenhowerphenomenon and Estimation and other differences across administrations(1973: 233). He also suggested that future research examine different The second wave of research,published in the early to economic indicators,group differences,other officeholders, mid-1980s, took up many of the themes outlined in the first and executives in other countries.8 Finally, he points out wave, attempting to specify more realistically the links that serial correlationplagues his analyses, noting that the between the economy,political events, and approval.Schol- Durbin-Watsonstatistic "reachedat best only about .70" ars used more sophisticated time series techniques in an (1973: 233, n. 28). Impressively,subsequent researchhas effort to determine whether the effects Mueller and others taken up all of these puzzles, problems, and suggestions. observed were real or just artifacts.They tried to determine The initial reactions to Mueller'swork fill out the rest of how long rally events (MacKuen1983) and economic con- the first wave. The two main reactions both addressed ditions affect approval (e.g., Norpoth and Yantek 1983). Mueller's"coalitions of minorities"thesis, or more specifi- These efforts spurred methodological debates as scholars cally, what it means to include a time counter in a time- offered differentmodel specifications and disagreed about series context. Stimson (1976) agreed that approval was the durationof effects, appropriatelag structures,functional trended, but argued that it drifted downward quadratically forms, and estimation techniques. Not surprisingly, the rather than linearly.Thus, approval immediately declines, period saw a significant advance in the technical nature of but eventually bottoms out and near the end of the term the field and many took note of the substantiveimplications approval rises again (although not to its original level). of different statistical approaches (see Ostrom and Smith Stimson gave this decline a differentsubstantive interpreta- 1992 and Beck 1991 for summaries of different methods tion. He argued that it results from relativelyuninformed and their implications).10 citizens having exaggerated expectations of the president Arguments about the duration of various effects were after a successful election campaign, and their inevitable importantbecause the question holds substantive implica- disappointmentafter the president took office. The end-of- tions. If effects have a relativelyshort duration, approvalat term boost, Stimson believed, simply indicates another any given time depends mostly on recent conditions, a build up of expectationsas the reelectioncampaign begins.9 "whathave you done for me lately"assessment ratherthan In sharp contrast, Kernell(1978: 521) rejected the view a long-term summary evaluation. If effects are more that time dominates approval and argued instead for the durable, past conditions contribute to overall evaluations, impact of "real events and conditions." Time, Kemell making it more difficult to change approval and to use approvalat any give time as an indicatorof the public'sreac- tion to recent actions (MacKuen1983). In addition, scholars in the second wave to 8 We will not review the extensiveliterature on in responded executiveapproval other Mueller'scall for a more nuanced treatmentof events. countries,but point to Paldamand Nannestad(1994) for a review.Several Many studies examineapproval of Congress(e.g., Durr,Gilmour, and Wolbrecht examined the effects of election campaigns, speeches and 1997), the SupremeCourt (e.g., Caldeiraand Gibson 1992), state legislatures (Patterson,Ripley and Quinlan 1992), and governorsand senators (see a special issue of State Politics& PolicyQuarterly edited by Niemi, 10 Beyle, and Sigelman(vol.2, no.3, Fall 2002) for six articles). It is interesting to note that approval models have often been at the 9 Mueller outlined almost identical disillusionment reasoning for the methodologicalfrontier in political science. Although immediatelydis- decay of approvalover time, but ultimately argued that the coalition of carded in favor of more advancedmethods, Mueller (1970) points out minorities explanationfit the data best. that his use of multiple regressionwas relativelyrare in political science. 504 POLITICALRESEARCH QUARTERLY other public appearances, domestic crises, significant The ThirdWave-Media, Elites,Individuals, domestic policy accomplishments, scandals and other and a HeterogeneousPublic events involving the president personally,such as the assas- sination attempt against Reagan(see especially Ostrom and By the late 1980s and early 1990s, presidentialapproval Simon 1985 and MacKuen1983). They found that domes- studies experienceda renaissance,largely in response to the tic events affect approval and that events could boost or "new look" approach to public opinion (Sniderman1993). depress approval.ll The new look emphasizes the explicit efforts of politicians, Two important theoretical developments mark the parties, and other elites to shape and change political atti- second wave. In both cases, Mueller'sinitial ideas were put tudes and behavior. Reflecting the new look perspective, into richer political context. First, scholars began to take scholars thought beyond the usual theoreticalsuspects, and account of presidents' incentives to gain and maintain considered how the media stand as an intervening force approval. Although presidents may have limited control between presidentsand the public, and how elite discourse over the economy, they have both the power and the incen- about events and conditions shape public attitudes. This tive to create dramatic symbolic events such as speeches, most recent wave of researchalso wrestles with heterogene- trips, legislative proposals, or even international engage- ity, explicitly modeling group- and individual-level varia- ments that boost approval(Ostrom and Job 1986; Ragsdale tions in approval. 1984; but see Simon and Ostrom 1989). By paying more AlthoughBrody and Page (1975) cited media coverageas explicit attention to the president'sincentive structureand a significantinfluence on approval,this insight was seldom the public'sattraction to drama, second wave studies made exploited until recently,when scholars emphasized media's these models significantlymore political and realistic. capacity to shape individual reactions to events and eco- Second, some scholars began to think more explicitly nomic conditions, therebyaffecting aggregate approval (e.g., about approvalat the individual-level.For example, Kerell Nadeau et al. 1999; Goidel, Shields, and Peffley 1997; Mutz and Hibbs (1981) and Hibbs, Rivers,and Vasilitos(1982a, 1992, 1994; West 1991). Brody (1991) provided the clear- 1982b) wrestledwith the consequencesof the way approval est statementof the way media content shapes approvalrat- ratings are typically conceptualizedand constructed. They ings. He constructed a media-based reinterpretationof took seriouslythe fact that approvalratings are aggregationsof empirical regularities like honeymoon periods and rally individualopinions rather than the opinion of an aggregate. events (see also Callaghanand Virtanen1993), arguingthat Although data limitations often constrained second wave public reactions to events and new presidents were best studies to analyzingaggregate data ratherthan modeling the understood via a two-step process, as events are first inter- individual-levelresponses directly, some scholarsbuilt explic- preted and evaluatedamong opinion elites, then these inter- itly micro-level theories and tested their implications for pretationsand assessmentsare transmittedto the public via aggregateapproval (e.g., Ostrom and Simon 1985; but see the mass media. For example, Brody argued that honey- Kinder1981 and Tedin 1986 for two analysesof individual- moon periods result from an overwhelminglypositive bal- level data). Conceptualizingapproval as an aggregationof ance of media coverage at the beginning of a term as most individual opinions led some analysts to consider whether elites tend to offer little criticism, withholding judgment differentpartisan or class groups employed a differentmix of until the president begins to take controversial actions. considerationswhen evaluatingthe president (Tedin 1986; Once this happens, opponents offer criticism,media cover- Hibbs, Rivers,and Vasilitos1982a). Regardless,the distinc- age becomes more negative, and public support falls as a tive featureof the second wave was an elaborationof the sta- result. He explained the rally phenomenon similarly. tistical technology used to estimate aggregate models of One psychological process by which media coverage approval.This wave culminatedwith severaldefinitive state- influences public attitudes-priming-has also become an ments, MacKuen's(1983) treatment of "politicaldrama," important part of the approval literature. According to Hibbs, Rivers,and Vasilitos'(1982a, 1982b) analysis of the Miller and Krosnick (2000: 301), "primingoccurs when economic underpinnings of approval, and Ostrom and media attention to an issue causes people to place special Simon's(1985) comprehensivemodel of approvaland leg- weight on it when constructingevaluations of overallpresi- islativesuccess. Littlemore needed to be said. dential job performance."The notion of priming rests on By the mid-1980s, then, the presidentialapproval field the claim that when individualsmake a choice, they rely on had reached something of a standstill. Continued progress salient and accessible information. This tends to be infor- was being made, particularlyas carefully conceived data mation emphasized"by the prevailingeconomic, social, and collection effortsand increasinglysophisticated attention to political conditions of the time,"which in turn are generally design and method enriched our knowledge of the determi- portrayedvia mass media (Krosnickand Kinder 1990: 500; nants of approval.However, Mueller's original claims largely see also Iyengarand Kinder 1987). drove the intellectualagenda. Mediacoverage can change the ingredientsof presidential support by making some considerations more salient or accessiblethan others (Iyengarand Kinder1987) and salient more other considera- 11Competing event coding schemes, however,became a continuing point considerationsaffect approval than of controversy,which we discuss below. tions do (Edwards,Mitchell, and Welch 1995). For example, FDR TO CLINTON, MUELLER TOTO? ? 505

Krosnickand Kinder(1990) found that opinions about for- the mass public into their models, including individuals' eign affairshad greaterinfluence on approvalin the wake of own assessmentsof the political and economic environment Iran-Contrarevelations. Similarly,Krosnick and Brannon ratherthan objectiveindicators of those environments.This (1993) found that evaluationsof Bush'shandling of foreign marks a significant departure from the first two waves, affairshad greaterinfluence on overall evaluationsafter the which relied on aggregateand objective measures. Several Gulf War than they had previously(see also Goidel, Shields, studies, including Brody's,unpack the public's reaction to and Peffley 1997; Peffley,Langley, and Goidel 1995). rally events, examining exactly who rallies-what types of The primingliterature raises an importantquestion: does people change from disapproval to approval. They find the public hold constant expectations of presidents over much more than blind patriotismis behind rallies.12In fact, time? In statistical terms, does one model account for panel studies with waves sandwiching a rally event approvalat all times? Mueller initially raised this question (Edwardsand Swenson 1997; Peffley,Langley, and Goidel and concluded that models must take the idiosyncrasiesof 1995) and time series analyses disaggregatedby partisan- each president into account. However, Mueller,like almost ship and education (Baum 2002; James and Rioux 1998) all afterhim, estimateda single coefficientfor events and the find that individual-level predispositions and attention to economy,assuming that their effects did not vary over time. media coverage of events shape attitudes toward the event Some argue explicitly that the public holds stable expecta- and the president.13 tions across presidencies. Ostrom and Simon (1985: 336) Many have taken a similar approach to study the link argue that "all presidents are expected to maintain peace, between economic conditions and approval. Rather than prosperity,domestic tranquility,and both the authorityand measuring aggregateresponses to objective economic indi- integrity of the office" (emphasis added). On the other cators, studies examined perceptions of the economy and hand, the priming literaturefinds that they can vary with their links to presidential approval (e.g., Norpoth 1996; current conditions. New estimation techniques that allow Clarkeand Stewart 1994; MacKuen,Erikson, and Stimson for time varying coefficients (Wood 2000) or data sets that 1992). Many of the major disputes in this period shift from pool cross sections over time (Gronke 1999) may help solve the correctstatistical specification of the effectsof the econ- this puzzle. omy to the correctpsychological foundations of the way the History also raises this question on occasion. For exam- public thinks about the economy. Most notably, scholars ple, despite enduring "the worst recession since the Great sparredover whether individualsare "peasantsor bankers," Depression," a foray into Beirut that cost 265 American relying on retrospectiveor prospective assessments of eco- lives, a host of scandals involving appointees, and the Iran- nomic conditions when evaluatingthe president (Norpoth ContraAffair, Reagan's approval remained high, resultingin 1996; Clarke and Stewart 1994; MacKuen, Erikson, and claims that "heis The Man in the TeflonSuit; nothing sticks Stimson 1992, 1996). to him" (Weisman 1984: 39). It seemed that the conven- These studies posited differentviews of how individuals tional wisdom regardingpublic approvalno longer applied. connect their perceptions of the economy to evaluationsof Similarly,flying in the face of researchfinding that approval the president. One side arguesthat the public, in the aggre- declines in response to scandal (e.g., Norpoth 1996; Rags- gate, acts remarkablysophisticated, punishing or rewarding dale 1987; Ostromand Simon 1985), Bill Clinton'sapproval presidentsbased on its expectationsof futureeconomic per- remained high during and after the Lewinsky scandal and formance (MacKuen, Erikson, and Stimson 1992). impeachment, again suggesting that he was evaluated by Although a sophisticated public defies mountains of differentcriteria. The spectacularrise and fall of Bush the research (Delli Carpini and Keeter 1996), MacKuen,Erik- elder's approval after the Gulf War and his son's dramatic son, and Stimson argue that the public lets economic and long lastingboost afterSeptember 11 also seem to chal- experts make forecasts,which media sources broadcast,and lenge general explanation. Can a single model account for the attentive portion of the public responds, adjusting all these presidencies? approvalin line with expectations.On the other side, schol- Remarkably,the answer seems to be "yes."The same fac- ars argue that the public evaluates current and past condi- tors that accounted for variations in approval from 1948 tions, punishing or rewarding the president accordingly through 1980 continue to operateboth before and after.For (Norpoth 1996; Clarke and Stewart 1994). As Norpoth example, Ostromand Simon (1989) found that Reaganwas (1996: 777) put it, "therequirements for that kind of behav- not made of Teflon after all. The elder Bush'sapproval fol- ior are far more modest regardingboth the averagecitizens' lowed expected patterns (Clarke, Rapkin, and Stewart capabilities and motivations to follow politics." Although 1994). Clinton's approval fits well within the findings of extant research(Newman 2002). Baum and Kerell (2001) even found that the standardset of indicators account for 12 See Edwards(1983) and Hibbs, Rivers, and Vasililtos(1982a) for ear- public approvalof FranklinRoosevelt. At least some factors lier examinationsof partisandifferences in response to Watergateand consistently affect approvalacross a wide variety of condi- rally events. 13 In addition to James and Rioux tions and presidencies. disaggregatingapproval by partisanship, (1998) furtherenrich the study of the rally phenomenon by incorpo- Beyond paying greater attention to media and elites, rating insights from international relations theory and taking into scholars in the third wave explicitly incorporatedaspects of account the costs of reactionsto internationalcrises. 506 POLITICALRESEARCH QUARTERLY the precise natureof the links between economic elites, eco- sources of those differencesand how these differencesrelate nomic news, mass perceptionsof the economy, and evalua- to the politicaland social positions that these groups occupy tions of the president remain under debate (e.g., Nadeau et in Americansociety (Newman 2003). Groupdifferences can al. 1999), explicitly specifying these links is a significant result from two sources, differencesin the levelsof the inde- step forward.All sides of the debateoffer a richerview of the pendent variables(e.g. blacks may perceive,or experience,a American public than that offered in early studies of differenteconomy than whites) or differencesin the weights approval.Rather than models of a public that almost mech- that they attach to variables,or both. For example, in their anistically responds to economic conditions, these works individual-level studies of approval by men and women, offer more explicit models of the public's interaction with Gilens (1988) and Newman (2003) conclude differencesin the economic and media environment. attitudes(levels of independentvariables) explained much of In addition to taking perceptionsof rally events and the the "approvalgender gap." However, Gilens (1988) also economy into account, several studies incorporatedmicro- found thatwomen weighed theirmore liberaldefense spend- foundations in a more direct way-they estimated individ- ing views more heavilywhen evaluatingthe President(in this ual-level models. Although studying approval at the indi- case, Reagan) than did men. Furthermore,Clarke et al. vidual-level differs significantly from the aggregate, (2000) found that approvalamong men responds more to discovering that individual-level analyses largely corrobo- pocketbook considerations,while approval among women rate the findings of aggregate-levelstudies lends credence to responds more to sociotropicassessments. our general understanding of approval. Individual-level Other studies similarly find that independent variables studies consistentlyconclude that economic conditions and have differenteffects across differentgroups. For example, perceptions powerfully affect approval(e.g., Gronke 1999; Ostrom and Simon (1988), using three panel waves in Gilens 1988; Ostrom and Simon 1988), as do assessments 1980, found that news coverage of the Iran Hostage Crisis of the way presidentshave handled significantinternational drove Democrats'approval, inflation was the strongestinflu- and domestic crises, i.e., rally events (e.g., Greene 2001; ence on Republicans' assessments, and unemployment Edwards and Swenson 1997; Peffley,Langley, and Goidel influenced independents. Others have found differential 1995). We expect continued efforts to compare and bridge reactions across income and partisan groups (Baum and individual and aggregatemodels of approval. Kernell2001; Ragsdale1987), while Dawson (1994) found Beyond corroboratingaggregate-level findings, however, that AfricanAmericans respond more to economic condi- individual-levelstudies have pointed to importantfactors that tions, particularlyunemployment, than do whites. Future affectapproval, which aggregatestudies cannot isolate. First, work should continue to examine subgroup differencesin these studies arebetter able to take predispositionslike parti- the weights that groups attach to various economic and sanshipinto account (Fischle2000). The unsurprisingeffects political considerations. of partisanshiphave importantimplications. The weakening Second, focusing on differences across the public can of partisanties among the public, ties that stabilizeindividu- help to isolate stability and change among the public. Not als' evaluations, may have made aggregateapproval more everyone in the public is equally likely to change his or her volatile over time and individualevaluations more uncertain evaluation (Kernell and Hibbs 1981). Focusing on who at any given time (Gronkeand Brehm2002; Gronke 1999). changes and why will provide insight into approval at the Second, individual-levelstudies find that policy attitudes aggregate level. For example, Tatalovich and Gitelson affectapproval (e.g., Gronke1999; Gilens 1988; see Erikson, (1990) found that out-partisanswho initially support the MacKuen,and Stimson 2002 for a discussion of policy atti- president fall into disapproval over time, leading to a tudes and approvalat the aggregatelevel). Third,individual- decline in approvalover the term.14 Further,Tedin's (1986) level studies point to the significantrole assessmentsof the analysis of panel data found that out-partisanswere the president'scompetence and integrityplay in overall evalua- most unstable in their opinions (see also Ostromand Simon tions (e.g., Newman 2003; Greene2001). 1988; Ragsdale1987). Individuallevel theorizingalso highlightssignificant het- Other approval studies focused on whether the most erogeneity within the public, raising the possibility that sophisticated (Miller and Krosnick 2000; Krosnick and variableshave differenteffects on differentsegments of the Brannon 1993) or the least sophisticated (Krosnick and population. This is hardly a new idea, as Mueller disaggre- Kinder 1990; Tedin 1986) are most likely to alter their eval- gated approvalby partisanshipin his book (1973: ch. 10; uations of the president, and what the implications are for see also Monroeand Laughlin 1983 and Hibbs, Rivers,and the quality of the public'sjudgments. These studies have Vasilitos1982a). However,there seems to be renewed atten- primarilyexplored whether primingis more powerfulat dif- tion to differences across the public as recent work has ferent levels of political knowledge. Some found that the examined heterogeneity across gender, racial, and party least knowledgeable were most susceptible to priming groups, across informationlevels, and at the level of individual attitudes. differencesin three Examininggroup approvalprovides 14 Again, Mueller outlined almost identical reasoning for the decay of important insights that aggregate studies easily overlook. approvalover time, but argued that the coalition of minoritiesexplana- First, studies examining group differencescan point to the tion fit the data best. FDR TO CLINTON, MUELLER TOTO? ? 507

(Krosnick and Kinder 1990; Iyengar et al. 1984), while and Roper archives). Several individual-levelsurveys have others found no knowledge-baseddifferences (Iyengar and built up considerable time series. The National Election Kinder 1987). Most recently,Krosnick and Brannon(1993) Studies began to ask a presidentialapproval item in 1972. and Millerand Krosnick(2000) found that priming is most Unlike the Gallup survey,the NES allowed respondents to likely among those with greater stores of political knowl- "strongly"approve and disapprovefrom 1980 on, providing edge. While previous work suggested that political novices a scale with twice as many response categories.In addition, were helpless against the tide of media coverage, more NYT/CBSpolls, administeredroughly on a monthly basis, recent work argues that considerable sophistication is have included a presidentialapproval item since 1976 and required to "interpret,store, and later retrieve and make are available.These data have not been tapped to their full inferences from news stories" (Miller and Krosnick 2000: potential to study presidentialapproval.16 312). The responsiveness of presidential approval to elite Fourth,methodological advances-the growingpopular- and media priming, the relative stability or instability of ity of experimental and quasi-experimental research approval compared to other political opinions, and differ- designs, the increasingsophistication of time series analysis, ences between sophisticatedand less sophisticatedrespon- and the development and popularization of discrete dents remain open questions. Nevertheless, raising these response models-have enabled scholars to model questions has enriched the study of presidentialapproval. processes more precisely and have strengthenedour inferences about cause and effect. Experimentaltechniques have TheDevelopment of the ThreeWaves led to greater attention to the individual-level processes underpinning evaluations, helping us understand public We believe four factors drove this broad trend toward reactions to rally events and the assignment of rewardand richer, more contextualized models, and we suspect that blame for economic change.17Time series advances have many of these factors will continue to shape the research allowed scholars to test when and for how long independ- agenda in the future. First, as alreadymentioned, events of ent variables affect approval.l8 Discrete response models the past 30 years forced scholars to extend, elaborate,and (e.g., binomial, ordered, and heteroskedastic probit and challengeconventional wisdom. For example, GeorgeH. W logit) make individual-level modeling of approval feasible Bush'srapid rise and decline in the polls during and after (Gronke 1999). Students of presidentialapproval have a far the Gulf War prompted studies of priming (e.g., Goidel, more varied set of data with which they can explore various Shields, and Peffley 1997; Krosnickand Brannon1993) and questions and a dramatically more sophisticated toolkit volatility (Gronke and Brehm 2002; Brace and Hinckley with which to test their ideas. 1992). Likewise,many scrambledto make sense of Clinton's rising approvalin the face of scandal (e.g., Shah et al. 2002; PRESIDENTIALAPPROVAL AND DEMOCRATICGOVERNANCE Fischle 2000; Miller 1999; Rozell and Wilcox 1999; Zaller 1998).15The simple passageof history will continue to force These innovations have been far more than data-fitting the literatureforward. exercises. Presidential approval research speaks to larger Second, changes in the field of public opinion-the emergenceof the "newlook"-encouraged scholars to con- 16 sider political competition, elite discourse, media coverage, Approval data are readily availableover the worldwide web and elec- and heterogeneityin their studies of presidentialapproval, tronic databases. Variouswebsites like pollingreport.comoffer aggre- be thus pointing the way toward the individual-level and gate-levelresults. The "pollsand surveys"section of Lexis-Nexiscan used to accumulate a number of survey observations, although group-centricanalyses of cited above. large approval at the level. recently announced a feature,avail- advancesin and only aggregate Gallup Third, rapid methodologicaltechniques able to subscribersto the "TuesdayBriefing," which will allow down- the accumulationof richer data sources undoubtedly stim- loading of individual level survey data. The Roper Center's"Poll" serv- ulated, and were stimulatedby, these trends in opinion and ice has long been a way to access individual Gallup studies, although of can be approval research. The first wave's focus on aggregate the cost to assemble a large number individual polls prohibi- was in due to the of a suffi- tive. Scholarsand students affiliatedwith the Inter University Consor- approval large part availability tium for Political and Social Researchcan obtain NES surveys, as well and consistent time series. ciently long, systematic, Gallup as media polls (ABC,NY Times). Individuallevel Harrissurveys, from While some disaggregateddata were available as early as the 1960s-1990s, are availableat the Universityof North Carolina. Mueller's(1973) writing, disaggregatedtime series data and 17 For instance, Krosnick and Brannon (1993) took advantage of well- to a cross-sectional studies only recently became available. placed panel surveys before and after the Gulf War develop quasi- the time demo- experimentaldesign to assess the way the political environmentshapes Today, Gallup series, disaggregatedby major the criteriaof evaluation.Other studies and Kinder 1987; and is obtainablein both (e.g., Iyengar graphics partyidentification, easily Miller and Krosnick 2000) set up explicitly experimental designs to printed (Ragsdale 1998) and electronic form (the Gallup assess the degree to which media coverage may prime the public to evaluatethe presidentin particularways, or how the Stateof the Union boosts approval(Druckman and Holmes 2003). 18 E.g., Box-Steffensmeierand Smith (1998, 1996); DeBoef and Granato 15 Many of these pieces fit nicely into the third wave, focusing on individ- (1997); see also a special edition of ElectoralStudies (Lebo 2000) and ual level changes, emphasizing media coverage and elite debate, and volume 4 of PoliticalAnalysis (Freeman 1992), which summarizesvari- examining differencesin reactionsacross groups. ous techniques. 508 POLITICALRESEARCH QUARTERLY

questions rooted in democratic theory. Does the president However, these features also hold darker prospects. have any power to improvehis standing?The answerto this Faced with impossible expectations, presidents may be left question, positive or negative, has implications for democ- little choice but to create the image that they are fulfilling racy In a perfect world, if higher approval means greater expectations, using the powers of office to stage symbolic power, the public would rewardthe president for meeting events, public relations derbies to effect favorablepercep- realistic and appropriateexpectations. Then the president tions, resulting in nothing more than an "image-is-every- would have an incentive to meet those expectations and thing presidency"(Waterman, St. Clair, and Wright 1999; would benefit from doing so. However, if the public holds see also Brace and Hinckley 1992). In the extreme, some the presidentaccountable for outcomes that he has little con- fear the possibility that the president may use his power in trol over, or has unrealisticallyhigh expectations, then the the internationalsphere in hopes of creatinga rally event to president'spower depends on the whims of the economy and divert attention from other less favorable considerations, history,setting up the presidency for almost inevitable fail- trying to induce the so-called "wag the dog" effect (see ure. On the other hand, if presidentshave too muchcontrol James and Rioux 1998 for an overview of literatureon the over approval,they could boost it without meeting appro- "diversionaryhypothesis"). priateexpectations, manipulating their way to power. In the end, it appearsthat presidentsare neither entirely Scholarshave tried to assess the president'sability to con- at the mercy of outside forces nor in total command of their trol approvalsince the firstwave's debate over the decline of approval.As usual, the truth is somewhere in the middle. support. Mueller,who arguedthat presidentsface the "inex- Sorting out when and to what extent the president shapes orabledescent" of their approval,offered a bleak view of the approvaland under what conditions the public resists pres- presidency,an officealmost doomed to failure(Mueller 1973: idential efforts to boost approval will provide insight into 213). Similarly,Stimson's analysis led him to suspect that the natureof presidentialpower and the characterand qual- "presidentialapproval may be almost wholly independentof ity of accountabilityin the Americansystem. the President'sbehavior in office, a function largely of inevitableforces associated with time"(1976: 1). The only way QUESTION AND PUZZLESFOR THE FUTURE presidentscan shape approvalis by runningfor reelectionand "using their office to control events, particularlysymbolic Although the general flow toward greatercontextualiza- events"to boost approval(1978: 10, emphasisin original). tion has taught us much, this trajectoryhas left important For his part, Kernell contended that presidents are questions unanswered. First, who or what determines the judged on the basis of "outcomes in the form of current criteriathe public uses to evaluatepresidents? The first two events and conditions"rather than what the presidentactu- waves took these criteria as largely exogenous, while the ally does in office (1978: 515). Kernellargued that holding priming literatureof the third wave suggests that they are presidents responsible for outcomes "should prompt the endogenous to the political, economic, social, and media system'smost powerful actor to engage in problem solving" environments.Extant literature does not tell us much about (ibid.). Others, however, would later argue that holding the the conditions under which the public will resist priming, president responsible for outcomes he can do little to con- but there appearto be some. For example, in his analysisof trol leads inevitablyto public disillusionmentand cynicism, the Lewinsky scandal, Zaller (1998: 185) argued that the and ultimately,presidential failure (e.g., Waterman,Wright, public ultimately returns to "bottom line politics" and is and St. Clair 1999; Lowi 1985). concerned with "politicalsubstance." As he put it, "however Studies in the third wave offer more hope for the presi- poorly informed, psychologically driven, and 'mass medi- dent, finding that perceptionsof the political and economic ated'public opinion may be, it is capableof recognizingand environment,not just objective indicators, drive approval. focusing on its own conception of what matters"(1998: Perceptionsdo not always perfectlyfollow objectiveindica- 186). Still, the bases of expectationsabout presidentialper- tors (e.g., Nadeau et al. 1999), and can be cultivatedmore formanceremain surprisingly unexplored (see, for example, easily than objectiveindicators can be altered.Furthermore, Hibbing and Theiss-Morse's1995 penetrating analysis of the priming literaturesuggests that presidentsmay have an Congressionalapproval). even subtlertool. Ratherthan changingcitizens' minds, pres- Second, while research to date has demonstrated the idents may be able to change how they make them up. If importanceof the two pillars of presidentialapproval-the presidentscan shape media coverage,which they sometimes economy and foreign affairs-little researchhas considered succeed in doing (e.g., Edwardsand Wood 1999), they may whether and how the changing internationalenvironment be able to affect approvalindirectly.19 Thus, public expecta- may affect presidentialpolitics, presidentialapproval, and tions may be met more easily than previouslythought. ultimately the power of the presidency In the economic realm, globalization and economic interdependence has reducedthe abilityof the presidentto controlmacroeconomic forces the ever had such control in the first 19 (if president For example, just days after revelations of the Lewinsky scandal, Bill If Americaneconomic conditions are affected Clinton tried to refocus the nation's attention his State of the place). deeply during of Union address, not once mentioning the scandal, a strategy that may by a regionalrecession, trading disputes, or the crash a for- have worked to boost his approval(e.g., Zaller 1998). eign stock market, will the president be held accountable? FDR TO CLINTON,MUELLER TOTO? ? 509

Perhapsfuture models of approvalwill have to include inter- synthesis will prove difficult but is essential for progress national,as well as national,economic conditions. towarda generalmodel of approval. Another revolutionarychange has taken place in foreign Fifth, most studies have ignored problems of simultane- affairs.In the last decade, since the end of the Cold War,the ity bias that often creep into analyses of presidential U.S. militaryhas been engaged in frequent,low-level inter- approval and presidential governance. Analyses of single national conflicts. Under the conventional model, these wave surveys certainly suffer from this as individuals may should result in a series of "rallyevents." Foreign affairs, first decide whether they approveof the president'sjob per- however, have always had something of a contradictory formance,and then use thatjudgment to inform their other impact on approval. Internationalconflict is supposed to responses. In addition, some have arguedthat presidentsare boost approval,even if just temporarily,while ongoing con- more likely to score legislativevictories when their approval flict tends to undermine it. This provides little insight into ratings are high, and then these same accomplishments how we should expect the public to react to militaryactiv- enhance their approvalratings (see Ostromand Simon 1985 ity in the Gulf War, Somalia, Haiti, or at the time of this for an example of dealingwith this issue).20From a very dif- writing, possibly Iraq. Does the public receive these differ- ferent direction,Jacobs and Shapiro (2000) argue that the ently than peacekeeping missions in Kosovo or Bosnia, or popular view of the presidency and the polls is backwards. an ongoing, undefined war on terrorism?There are some Superficially,it seems like presidents are slaves to public indications that the public does see some nuance in foreign opinion, but Jacobs and Shapiro demonstratethat, in fact, affairs.The public, both at the mass and elite levels, have presidentshave a specific political agenda, and use the polls displayedfar less consensus with respect to foreignaffairs in as a guide to achieving that agenda. Often thought to be a the post Cold War era (Holsti 1996). Thus, presidential data issue, these examples show that simultaneity is more action in the international sphere may produce more broadly an issue of research design and causal interpreta- divided responsesthan has historicallybeen the case. To our tions. To date, no simple solution presents itself. It seems knowledge, there have been few scholarlyattempts to exam- that carefulconceptual and theoreticalthinking, along with ine the impact of these changes on presidential approval, using a varietyof data sources and analyticaltechniques, are beyond simply coding these conflicts as foreign policy the best strategiesavailable. "events"(e.g., Gronkeand Brehm2002; Braceand Hinckley Finally, we think the field would benefit from greater 1993). If nothing else, these changes in the economy and unity in data and measures. We have reviewed a large foreignaffairs open up a rich new avenue of researchfor the number of studies and very few of them draw upon the next ten years. same set of survey data or event series. Each new scholar Third, while we are considering revolutions, future seemingly builds the dataset anew. Selection of events is researchmust be sensitive to the changing media universe. especially inconsistent across studies.2' We realize that dif- Mass media are undergoingtremendous change, as newspa- ferent substantive questions may call for different opera- pers merge and disappear,cable television stations prolifer- tional approaches, but we echo Mueller'sinitial call for ate, and the Internet continues to revolutionize the way developing better measures to capture political realities. A informationis disseminatedand viewed. The rapid diffusion unified event series would save significant labor, ease the of the Starr Reportprovided just a glimpse of this future: temptationto use events to boost model fit, and limit suspi- informationthat disseminatesat "warpspeed" during the 24- cions that others have done so. Further,a unified and avail- hour news cycle (Kovach and Rosenstiel 1999). How this able dataset, containing Gallup, Harris, and New York may affect the president'sability to marshalpublic opinion Times polls, would be a boon to the field. remains unclear. Some have begun to tackle this question, In the end, we think much has been gained by 30 years finding that the growth of cable television has diminished of pushing, refining, and contextualizing the basic argu- the president'sability to dominate the news (Baumand Ker- ments Mueller advanced. Our understandingof how and nell 1999). Furthermore, trust in media has decreased, why the economy,events, time, individualcitizens, and indi- potentiallyaltering its affectson approval.Miller and Kros- vidual presidentsaffect approvalis more rich, realistic,and nick (2000) found that trusted media sources were more comprehensive. Future research must continue to deepen effective in priming viewers (see also Druckman 2001). our understanding,bringing together insights that currently Futureresearch must take these changes into account. remain somewhat in isolation. Doing so will lead to greater Fourth,most studies have contextualizedapproval in different ways, resulting in a patchwork of increasinglyrich 20 This issue is moot if does not affect the pres- models that do not fit togethereasily. We have surveyedstud- particular approvalreally ident'spower in congress (e.g., Bond and Fleisher 1990). ies of and some 21 individual-,group-, aggregate-levelapproval, Comparingtwo careful operationalrules for identifying events (Brace of which focus on economics, otherson dramaticevents, and and Hinckley 1992; Ostromand Simon 1985; Ostromand Smith 1992) others on media effects. Future researchwill have to bring is illustrative.Over the 1953-1988 period that each study examined, togetherthe variousfeatures of media behavior,the economy, the Ostrom, Simon, and Smith strategy identified 122 events and the Braceand identified 98. 42 eventswere events, drama, and individuals' Hinckley strategy Only identified presidential predispositions both less than half the events Further,17 of the and it will have to deal by strategies, of identified. long-termexpectations. Furthermore, 42 commonly identified events were coded as affectingapproval in dif- with significantheterogeneity across the public. This work of ferentmonths. 510 POLITICALRESEARCH QUARTERLY understanding of that "unfocused" but politically important Dawson, MichaelC. 1994. Behindthe Mule: Race and Classin African- Princeton Press. question of whether individuals approve or disapprove of the AmericanPolitics. Princeton, NJ: University and Granato. 1997. Data president's performance in office (Neustadt 1980: 81). DeBoef, Suzanna, Jim "Near-Integrated and the Analysis of PoliticalRelationships." American Journal of PoliticalScience 41 (April):619-40 REFERENCES Delli Carpini,Michael X., and Scott Keeter.1996. WhatAmericans Knowabout Politicsand Why it Matters.New Haven, CT: Yale Baum, Matthew A. 2002. "The Constituent Foundations of the UniversityPress. Rally-Round-the-Flag Phenomenon." International Studies Druckman,James. 2001. "On the Limits of FramingEffects: Who Quarterly46: 263-98. Can Frame?"Journal of Politics63: 1041-66. Baum, Matthew,and Samuel Kerell. 1999. "HasCable Ended the Druckman,James N., and Justin W Holmes. 2003. "ThePolitics of Golden Age of PresidentialTelevision?" American Political Sci- Presidential Approval: How Presidential Rhetoric Affects enceReview 93 (January):99-114. Approval."Presented at the 2003 Annual Meeting of the Mid- .2001. "Economic Class and Popular Support for west PoliticalScience Association, Chicago. FranklinRoosevelt in War and Peace."Public Opinion Quarterly Durr,Robert H., John B. Gilmour,and ChristinaWolbrecht. 1997. 65: 198-229. "ExplainingCongressional Approval." American Journal of Polit- Beck, Nathaniel. 1991. "ComparingDynamic Specifications:The ical Science41: 175-207. Case of PresidentialApproval." Political Analysis 3: 51-88. Edwards, George C. 1983. The PublicPresidency. New York: St. Bond, Jon R., and RichardFleisher. 1990. ThePresident in the Leg- Martin'sPress. islativeArena. Chicago: University of Chicago Press. Edwards,George C. III, William Mitchell, and Reed Welch. 1995. Box-Steffensmeier,Janet M., and Renee M. Smith. 1996. "The "ExplainingPresidential Approval: The Significance of Issue Dynamics of AggregatePartisanship." American Political Science Salience."American Journal of PoliticalScience 39: 108-34. Review90 (September):567-80. Edwards,George C. III, and Tami Swenson. 1997. "Who Rallies? _ . 1998. "InvestigatingPolitical Dynamics Using Fractional The Anatomyof a RallyEvent." Journal of Politics59: 200-12. IntegrationMethods." American Journal of PoliticalScience 42 Edwards,George C. III, and B. Dan Wood. 1999. "WhoInfluences (April):661-89. Whom? The Presidentand the Public Agenda."American Polit- Brace,Paul, and BarbaraHinckley. 1992. Followthe Leader:Opin- ical ScienceReview 93: 327-44. ion Pollsand the ModernPresidents. New York:Basic Books. Erikson, Robert S., Michael MacKuen, and James A. Stimson. . 1993. "Presidential Activities from Truman through 2002. TheMacro Polity. New York:Cambridge University Press. Reagan:Timing and Impact."Journal of Politics55 (2): 382-98. Erikson, Robert S., Gerald C. Wright, and John P Mclver. 1993. Brody, RichardA. 1991. Assessingthe President:The Media, Elite StatehouseDemocracy: Public Opinion and Policyin the American Opinion,and PublicSupport. Stanford, CA: StanfordUniversity States.New York:Cambridge University Press. Press. Fischle, Mark. 2000. "MassResponse to the Lewinsky Scandal: Brody,Richard A., and Benjamin I. Page. 1975. "The Impact of MotivatedReasoning or BayesianUpdating?" Political Psychol- Events on Presidential Popularity: The Johnson and Nixon ogy 21: 135-59. Administrations."In Aaron Wildavsky,ed., Perspectiveson the Freeman,John R., ed. 1992. PoliticalAnalysis, vol. 4. Ann Arbor: Presidency,pp. 136-47. Boston, MA:Little, Brown. Universityof MichiganPress. Caldeira,Gregory A., and James L. Gibson. 1992. "TheEtiology of Gilens, Martin. 1988. "Genderand Support for Reagan:A Com- Public Support for the Supreme Court."American Journal of prehensiveModel of PresidentialApproval." AmericanJournal of PoliticalScience 36: 635-64. PoliticalScience 32: 19-49. Callaghan,Karen J., and Simo Virtanen.1993. "RevisedModels of Goidel, RobertK., Todd G. Shields, and MarkPeffley. 1997. "Prim- the 'Rally Phenomenon':The Case of the Carter Presidency." ing Theory and RASModels: Toward an IntegratedPerspective Journalof Politics55: 756-64. of Media Influence."American Politics Quarterly 35: 287-318. Canes-Wrone, Brandice, and Scott DeMarchi. 2002. "Presiden- Greene,Steven. 2001. "TheRole of CharacterAssessments in Pres- tial Approval and Legislative Success." Journal of Politics idential Approval."American Politics Research 29: 196-210. 64: 491-509. Gronke, Paul W 1999. "Policies, Prototypes, and Presidential Canes-Wrone,Brandice, and KennethW Shotts. 2002. "TheCon- Approval."Paper prepared for the AnnualMeeting of the Amer- ditional Nature of PresidentialResponsiveness to Public Opin- ican PoliticalScience Association,Atlanta. ion." Paper presented at the Annual Meeting of the Midwest Gronke,Paul, and John Brehm.2002. "History,Heterogeneity, and PoliticalScience Association, Chicago. PresidentialApproval." Electoral Studies 21: 425-52. Clarke, Harolde D., Jon Rapkin, and MarianneC. Stewart. 1994. Gronke,Paul, Jeffrey Koch, and J. MatthewWilson. 2003. "Follow "APresident Out of Work."BritishJournal of PoliticalScience 24: the Leader?Presidential Approval, Perceived Presidential Sup- 292-305. port, and Representatives'Electoral Fortunes." Journal of Politics Clarke, Harold D., and Marianne Stewart. 1994. "Prospections, 65: 785-808. Retrospections,and Rationality:The 'Bankers'Model of Presi- Hibbs, Douglas A. Jr., R. Douglas Rivers, and Nicholas Vasilatos. dential Approval Reconsidered."American Journal of Political 1982a. "TheDynamics of PoliticalSupport for AmericanPresi- Science38: 1104-23. dents Among Occupational and Partisan Groups."American Clarke, Harold D., MarianneC. Stewart, Michael Ault, and Euel Journalof PoliticalScience 26: 312-32. Elliott. 2000. "Men,Women and the Dynamics of Presidential 1982b. "On the Demand for Economic Outcomes: Macro- Approval."Working Paper. University of North Texas. economic Performance and Mass Political Support in the Cohen,Jeffrey E. 1997. PresidentialResponsiveness and PublicPolicy- United States, Great Britain,and Germany."Journal of Politics, Making.Ann Arbor:University of MichiganPress. 44: 426-62. FDR TO CLINTON,MUELLER TOTO? ? 511

Hibbing, John R., and ElizabethTheiss-Morse. 1995. Congressas Mutz, Diana C. 1992. "MassMedia and the Depoliticization of PublicEnemy: Public Attitudes Towards American Public Institu- Personal Experiences." AmericanJournal of Political Science tions.New York:Cambridge University Press. 36: 483-508. Holsti, Ole. 1996. PublicOpinion and AmericanForeign Policy. Ann .1994. "ContextualizingPersonal Experience:The Role of Arbor:University of MichiganPress. the Mass Media."Journal of Politics56: 689-714. Iyengar,Shanto, and Donald R. Kinder. 1987. News that Matters. Nadeau, Richard,Richard G. Niemi, David P Fan, and Timothy Chicago:University of Chicago Press. Amato. 1999. "Elite Economic Forecasts, Economic News, Jacobs, LawrenceR., and RobertY. Shapiro.2000. PoliticiansDon't Mass EconomicJudgements, and PresidentialApproval." Jour- Pander:Political Manipulation and the Lossof DemocraticRespon- nal of Politics61: 109-35. siveness.Chicago: University of Chicago Press. Neustadt, Richard. 1980. PresidentialPower, 2nd ed. New York: James, Patrick, and Rioux, Jean Sebastian. 1998. "International Wiley Crises and Linkage Politics: The Experiences of the United Newman, Brian.2002. "BillClinton's Approval Ratings: The More States, 1953-1994." PoliticalResearch Quarterly 51: 781-812. Things Change the More They Stay the Same." Presidential Kernell,Samuel. 1978. "ExplainingPresidential Popularity. Ameri- ResearchQuarterly 55: 781-804. can PoliticalScience Review 72 (2) June: 506-22. .2003. Standardsof Judgment:The Foundationsof Presidents' . 1997. Going Public: New Strategies of Presidential Leadership, PopularSupport. Ph.D. Dissertation,Duke University 3rd ed. Washington,DC: CongressionalQuarterly Press. Newman, Brian,and CharlesW Ostrom,Jr. 2002. "ExplainingSeat Kernell, Samuel, and Douglas A. Hibbs, Jr. 1981. "A Critical Changesin the US House of Representatives:1950-1998." Leg- Threshold Model of Presidential Popularity."In Douglas A. islativeStudies Quarterly 27: 383-405. Hibbs, Jr. and Heino Fassbender,eds., ContemporaryPolitical Norpoth, Helmut. 1996. "Presidentsand the ProspectiveVoter." Economy:Studies on the Interdependenceof Politicsand Economics, Journalof Politics58: 776-92. pp. 49-71. New York:North-Holland Publishing Company. Norpoth, Helmut, and Thom Yantek.1983. "MacroeconomicCon- Kinder,Donald R. 1981. "Presidents,Prosperity, and Public Opin- ditions and Fluctuation of PresidentialPopularity: The Ques- ion." PublicOpinion Quarterly 45: 1-21. tion of LaggedEffects." American Journal of PoliticalScience 27: King, Gary,and Lyn Ragsdale. 1988. TheElusive Executive. Wash- 785-807. ington, DC: CongressionalQuarterly Press. Ostrom, CharlesW., Jr. and BrianL. Job. 1986. "ThePresident and Kovach, Bill, and Tom Rosenstiel. 1999. WarpSpeed. New York: the PoliticalUse of Force."American Political Science Review 80: CenturyFoundation Press. 541-66. Krosnick,Jon A., and LauraA. Brannon.1993. "TheImpact of the Ostrom, CharlesW, Jr. and Dennis M. Simon. 1985. "Promiseand Gulf War on the Ingredientsof PresidentialEvaluations: Multi- Performance:A Dynamic Model of Presidential Popularity." dimensionalEffects of PoliticalInvolvement." American Political AmericanPolitical Science Review 79: 334-358. ScienceReview 87: 963-75. .1988. "ThePresident's Public." American Journal of Political Krosnick,Jon A., and Donald Kinder.1990. "Alteringthe Founda- Science32: 1096-1119. tions of Support for the PresidentThrough Priming." American 1989. "The Man in the Teflon Suit? The Environmental PoliticalScience Review 84: 497-512. Connection, Political Drama, and Popular Support in the Lowi, Theodore. 1985. ThePersonal President. Ithaca, NY: Cornell ReaganPresidency" Public Opinion Quarterly 53: 353-87. UniversityPress. Ostrom, CharlesW, Jr., and Renee M. Smith. 1992. "ErrorCor- MacKuen,Michael. 1983. "PoliticalDrama, Economic Conditions, rection, Attitude Persistence,and ExecutiveRewards and Pun- and the Dynamics of PresidentialPopularity." American Journal ishments:A BehavioralModel of PresidentialApproval," Politi- of PoliticalScience 27: 165-92. cal Analysis4: 127-83. MacKuen, Michael, Robert S. Erikson, and James A. Stimson. Paldam,Martin, and Peter Nannestad. 1994. "TheVP Function:A 1992. "Peasantsor Bankers?The AmericanElectorate and the Surveyof the Literatureon Vote and PopularityFunctions After U.S. Economy"American Political Science Review 86: 596-611. 25 Years."Public Choice 79: 213-45. _ . 1996. "Presidents and the Prospective Year: Comment." Patterson,Samuel C., RandallB. Ripley,and Stephen V Quinlan. Journalof Politics58: 793-801. 1992. Citizens'Orientations Toward Legislatures: Congress and Miller,Arthur H. 1999. "Sex, Politics, and Public Opinion: What the State Legislatures."Western Political Quarterly 45: 315-38. Political Scientists Really Learned from the Clinton-Lewinsky Peffley,Mark, Ronald E. Langley,and RobertKirby Goidel. 1995. Scandal."PS: Political Science and Politics32: 721-29. "PublicResponses to the PresidentialUse of MilitaryForce: A Miller, Joanne M., and Jon A. Krosnick. 2000. "News Media Panel Analysis."Political Behavior 17: 307-37. Impacton the Ingredientsof PresidentialEvaluation: Politically Peterson, MarkA. 1990. LegislatingTogether: The WhiteHouse and Knowledgeable Citizens are Guided by a Trusted Source." CapitolHillfrom Eisenhower to Reagan.Cambridge, MA: Harvard AmericanJournal of PoliticalScience 44: 295-309. UniversityPress. Monroe, Kristen R. 1978. "EconomicInfluences on Presidential Ragsdale,Lyn. 1984. "The Politics of PresidentialSpeechmaking, Popularity."Public Opinion Quarterly 42: 360-69. 1949-1980." AmericanPolitical Science Review 78: 971-84. Monroe, KristenR., and Dona MetcalfLaughlin. 1983. "Economic .1987. "PresidentialSpeechmaking and the Public Audi- Influences on PresidentialPopularity Among Key Political and ence: Individual Presidents and Group Attitudes."Journal of Socioeconomic Groups: A Review of the Evidence and Some Politics49: 704-36. New Findings."Political Behavior 5: 309-45. .1998. Vital Statisticson the PresidencyWashington, DC: Mueller,John. 1970. "PresidentialPopularity from Truman to John- CongressionalQuarterly Press. son." TheAmerican Political Science Review 64 (1) March:18-34. Rivers,Douglas, and Nancy L. Rose. 1985. "Passingthe President's . 1973. War, Presidents, and Public Opinion. Lanham, MD: Program:Public Opinion and PresidentialInfluence in Con- UniversityPress of America. gress."American Journal of PoliticalScience 29: 183-96. 512 POLITICAL RESEARCH QUARTERLY

Rohde, David W, and Dennis M. Simon. 1985. "Presidential Tedin, Kent. 1986. "Changeand Stabilityin PresidentialPopular- Vetoes and CongressionalResponse: A Study in Institutional ity at the IndividualLevel." Public Opinion Quarterly 40: 1-21. Conflict."American Journal of PoliticalScience 29: 397-427. Waterman,Richard W, RobertWright, and GilbertSt. Clair.1999. Rozell, MarkJ., and Clyde Wilcox, eds. 1999. "Scandaland Gov- TheImage-is-Everything Presidency: Dilemmas in AmericanLead- ernment:Current and FutureImplications of the Clinton Presi- ership.Boulder, CO: Westview Press. dency A Symposium"PS: Political Science and Politics32: 538-61. Weisman,Steven R. 1984. "Canthe MagicPrevail?" New YorkTimes Shah, DhavanV, MarkD. Watts, David Domke, and David P Fan. Magazine29 April, pp. 38-56. 2002. "NewsFraming and Cueing of Issue Regimes:Explaining West, DarrellM. 1991. "Televisionand PresidentialPopularity in Clinton'sPublic Approvalin Spite of Scandal."Public Opinion America."British Journal of PoliticalScience 21: 199-214. Quarterly66: 339-70. Wood, B. Dan. 2000. "WeakTheories and ParameterInstability: Sigelman, Lee. 1979. "PresidentialPopularity and Presidential Using Flexible Least Squares to Take Time VaryingRelation- Elections."Public Opinion Quarterly 43: 532-34. ships Seriously."AmericanJournal of PoliticalScience 44: 603-18. Simon, Dennis M., and CharlesW Ostrom,Jr. 1989. "TheImpact Zaller,John R. 1992. TheNature and Originsof Mass Opinion.New of Televised Speeches and Foreign Travel on Presidential York:Cambridge University Press. Approval."Public Opinion Quarterly 53: 58-82. . 1998. "MonicaLewinsky's Contribution to Political Sci- Sniderman,Paul. 1993. The New Lookin PublicOpinion Research. ence." PS:Political Science and Politics33: 182-89. In Ada Finifter,ed., PoliticalScience: The State of the DisciplineII. Washington,DC: AmericanPolitical Science Association. Stimson,James A. 1976. "PublicSupport for AmericanPresidents." Received: March 2003 PublicOpinion Quarterly 40: 1-21. 11, 2003 Tatalovich,Raymond, and Alan R. Gitelson. 1990. "PoliticalParty Accepted for Publication: March 18, Linkagesto PresidentialPopularity: Assessing the 'Coalitionof [email protected] Minorities'Thesis." Journal of Politics52: 234-42. [email protected]