The War at Home: Antiwar Protests and Congressional Voting, 1965 to 1973 Author(s): Doug McAdam and Yang Su Source: American Sociological Review, Vol. 67, No. 5 (Oct., 2002), pp. 696-721 Published by: American Sociological Association Stable URL: http://www.jstor.org/stable/3088914 Accessed: 09/08/2009 16:11

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=asa.

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 organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access to American Sociological Review.

http://www.jstor.org THE WAR AT HOME: ANTIWAR PROTESTS AND CONGRESSIONAL VOTING, 1965 TO 1973

DOUG MCADAM YANG SU Stanford University University of California, Irvine

Time-series analysis is used to assess the relationship between antiwar protests and congressional voting on war-related roll calls during the Vietnam era. Using protest event data coded from The New York Times and counts of roll-call votes generated from congressional voting data, we test for three specific mechanisms: disruptive protest, signaling, and public opinion shift. Extreme forms of disruptive protest are hypothesized as having a direct positive effect on congressional voting. Lohmann's signaling model posits exactly the opposite relationship between protest and policy. Especially extreme protests are expected to have a negative effect on both the pace and pro-peace direction of congressional action. Conversely, large (and more mod- erate) protests are expected to have a positive effect on House and Senate voting. The final mechanism, public opinion shift, depicts the relationship as indirect, with protest encouraging public opinion change, which, in turn, encourages increasingly favorable congressional voting. The results are somewhat mixed with respect to all three mechanisms, but suggest an interesting general pattern. The most extreme or threatening forms of protest (e.g., those featuring violence by demonstrators and/or property damage) simultaneously increase pro-peace voting while depressing the overall pace of congressional action. The reverse is true for more persuasive forms of protest (e.g., large demonstrations), which appear to increase the pace of voting while depressing the likelihood of pro-peace outcomes.

A lthough the study of power and poli- in point. Over the past two decades, political tics is central to both political science sociologists have made social movements a and sociology, sharp disciplinary divisions central object of study. Political scientists, on persist with regard to a host of topics that fall the other hand, have paid comparatively little under these general headings. The study of attention to social movements. No doubt social movements affords an interesting case much of this discrepancy is due to the central importance accorded formal political institu- tions within political science. Sociology, by Direct correspondence to Doug McAdam, De- partment of Sociology, Stanford University, Stanford, CA 94305 ([email protected]). received two collaborative research grants (SBR- This article is part of a larger collaborative re- 9709356, SES-991296), and Penn State one search project investigating the dynamics of col- (SES-9911431). We also acknowledge the gradu- lective protest. With four principal investiga- ate students who have given so much of their tors-Doug McAdam, John McCarthy, Susan time and energy to the project. The key graduate Olzak, and Sarah Soule-and three research students at each site include: Jennifer Earl, Alan sites-Arizona, Penn State, and Stanford-the Schussman, and Nella Van Dyke (Arizona); project has benefited from six separate National Assata Richard and Andrew Martin (Penn State); Science Foundation grants. The work at Arizona and Maya Beasley, Kyra Greene, Yang Su, Noah has been supportedby two collaborative research Uhrig, and Simon Weffer (Stanford). Finally, we awards (SBR-9709337 and SBR-9709356) and a thank Andrew Martin, John McCarthy, Assata young investigator award to Sarah Soule (SES- Richard, Sarah Soule, and Susan Olzak for their 9874000). In addition, Stanford University has helpful comments on early drafts of the article.

696 AMERICAN SOCIOLOGICALREVIEW, 2002, VOL. 67 (OCTOBER:696-72 1) THE POLITICAL IMPACT OF ANTIWAR PROTESTS 697 contrast, has long been interested in informal Burstein's (1979, 1998; Burstein and or emergent social and political processes. Freudenburg 1978) and Lohmann's (1993) But we suspect there is another influence at work is especially promising in this regard, work here. From the outset, social-movement as both theorize the complex relationships scholars within sociology appear to have as- between protest activity, public opinion, and sumed the efficacy of movements as vehicles policy outcomes. We draw heavily on their of social and political change, whereas po- work while also seeking to extend it. Spe- litical scientists have tended to view social cifically, we identify three mechanisms- movements as politically ineffectual, stress- disruptive protest, signaling, and public ing instead the role of elections and public opinion shift-that we believe may help ac- opinion as the main popularmechanisms me- count for the variable impact social move- diating policy shifts. ments have on policy processes. It is time, in our view, to broaden the study Besides this theoretical contribution, the of social movements by incorporating in- empirical focus of our work bears mention sights from both disciplines. Some of this is as well. We seek to understand the predic- already happening. For example, sociolo- tive mechanisms that account for social- gists have taken to systematically assessing movement impacts in one of the most sig- the impact of social movements on specific nificant, yet curiously understudied, move- policy outcomes, rather than simply assum- ments in recent U.S. history. Although the ing effects. Still, with few exceptions, soci- Vietnam era antiwar movement remains one ologists continue to evince little interest in of the most intense, large-scale, and divisive the popular policy mechanisms of concern to movements in American history, it has been political scientists. The disregard for public almost totally ignored by social-movement opinion is especially regrettable in our view. scholars. One of the few exceptions is More specifically, the links between protest Burstein and Freudenburg's(1978) article on activity, shifts in public opinion, and policy the impact (or lack there of) of antiwar dem- change should be of central concern to so- onstrations on Senate voting on Vietnam cial-movement scholars. War-relatedmeasures. Alas, the Burstein and If sociologists assume the impact of move- Freudenburg article did not inspire other ments without testing for effects, political similar efforts. Instead, it remains a singular scientists, with few exceptions (Fording piece topically and ahead of its time in its 1997; Lohmann 1993; Rochon 1998; Tarrow creative use of systematic quantitative data 1998), continue to make the opposite error: to address the issue of movement outcomes. assuming the ineffectuality of movements. This lack of systematic research on the anti- Some of this rests on a stereotypic-and war struggle by social-movement scholars is somewhat outdated-view of social move- in marked contrast to the voluminous litera- ments as small collections of marginal out- tures on the civil rights struggle and moder siders, easily ignored by policy elites. But women's movement. So while extending our some movements are neither small (e.g., understanding of the dynamics of'social- women's, environment, civil rights, etc.) nor movement outcomes, we hope as well to properly conceived of as outside of main- shed empirical light on the dynamics of con- stream institutions. The work of Katzenstein tention associated with this neglected social (1990, 1998), McCann (1994), Meyer and movement. We begin by offering a brief his- Tarrow (1997), and others represents an im- tory of the Vietnam War and the domestic portant corrective to the simplified insider/ struggle it set in motion. outsider view of formal institutions and so- cial movements. The institution- increasing VIETNAM: THE WAR ABROAD alization of the social-movement form in the AND AT HOME United States and other Western democra- cies makes it all the more important for po- Our research focuses on antiwar protests, litical scientists to take social movements Congressional voting, and public opinion seriously as a potential influence on policy during the crucial period, 1965 to 1973. We processes (McAdam 1998; McCarthy and focus on these years because they nicely de- McPhail 1997; Meyer and Tarrow 1997). marcate the active period of domestic con- 698 AMERICAN SOCIOLOGICAL REVIEW tention over U.S. involvement in Vietnam. presidential race occasion any electoral pro- Although the Tonkin Gulf Resolution that test against Johnson's Vietnam policy. On led to a dramatic escalation in American in- the contrary,Johnson defeated his opponent, volvement in Vietnam was passed by Con- Barry Goldwater,in one of the most lopsided gress on August 4, 1964, a visible antiwar races in presidential history. But if antiwar movement only coalesced the following dissent lay dormant through 1964, the situa- year. And even though the United States re- tion changed markedly the following year. mained deeply enmeshed in Vietnam until As troop levels rose rapidly and combat the fall of Saigon in 1975, domestic conten- deaths grew apace of those levels, so too did tion over that involvement droppedto almost public attention and opposition to the war. none following the signing of the Paris By May of 1965 the war had replaced civil Peace Accords in 1973. Hence the temporal rights as the "most important problem con- focus for our research. fronting the country" (Gallup 1972). And In popular imagination, America's Viet- while a sizable majority of the American nam misadventure roughly corresponds to public still expressed support for the war ef- the years of our research. In actual fact, U.S. fort, a burgeoning antiwar movement rooted involvement in Vietnam spans nearly the en- on campus and in established peace groups tire Cold War era. U.S. concern over Viet- expanded rapidly. nam (then known as French Indochina) first By the middle of 1966, 55 percent of the surfaced in 1948-1949 when a strong Com- American public had come to regard the war munist-inspired nationalist movement arose as the country's most pressing problem to challenge the existing French colonial re- (Gallup 1972). Although still in the minor- gime. Unlike similar rebellions that devel- ity, the percentage that had come to view the oped in the same period throughout South- war as a "mistake" had risen to 35 percent. east Asia, the Vietnamese movement proved Reflecting these trends, antiwar protests more enduring, ultimately inflicting on the grew ever larger and more disruptive. The French the twin humiliations of crushing October 1967 March on the Pentagon repre- military defeat (at Dien Bien Phu in May sented a new and disturbinghigh-water mark 1954) and a negotiated peace settlement that in antiwar protest, featuring perhaps 20,000 marked the end of French rule in Vietnam. participants, repeated clashes with police, Under the terms of that settlement, Vietnam and some 647 arrests and 47 injuries requir- was to be unified through a popular election ing hospitalization (DeBenedetti 1990). that was to take place in the summer of The Tet Offensive in February,1968 deep- 1956. Fearing the outcome of the contest, the ened the domestic crisis still further by un- United States worked with allies to block the dermining public confidence in the govern- election, leaving the fate of the country un- ment's handling of the war, and especially resolved and requiring ever greater commit- its continued rosy assurances that the war ments of American financial, diplomatic, was "winnable." Having consistently por- and military resources to maintain the ille- trayed the enemy as limited in its military gal arrangement.These efforts continued un- capacity, the coordinated assaults that der three Presidents-Eisenhower, Kennedy, marked Tet, and for a time threatened sev- and Johnson-until the still-controversial eral major South Vietnamese cities, soured Tonkin Gulf incident in August 1964 al- public opinion all the more. So much so that lowed Johnson the freer hand he sought to Lyndon Johnson was forced to withdraw expand America's military involvement in from the 1968 presidential race on the the country. strength of growing antiwar sentiment. The Tonkin Gulf Resolution marked a de- Ironically, it was that ultimate Cold warrior, cisive turning point in America's long Viet- Richard Nixon, who was to be the ultimate nam saga. It did not, however, immediately beneficiary of this sentiment as he defeated engender significant opposition to the Johnson's Vice President, HubertHumphrey, stepped-up U.S. war effort. Indeed, on Au- in the November election. gust 4, the Resolution passed the House Nixon's antiwar "honeymoon" proved unanimously and occasioned only two dis- short-lived, however. The nationally coordi- senting votes in the Senate. Nor did the fall nated Moratorium days held on October 15 THE POLITICAL IMPACT OF ANTIWAR PROTESTS 699 and November 15, 1969 proved to be the plications here. The first concerns the possi- largest single actions mounted by the anti- bility of mounting collective action. If ratio- war movement. That same November, anti- nal actors refrain from such action, then it war sentiment among the general public should be nearly impossible to organize a peaked as well, with 55 percent of those movement in the first place. The second im- polled pronouncing the war a "mistake" plication has to do with those who might (Boettcher 1985:446). The spring of 1970 take part in a movement, given its demon- brought heightened outrage, as leaks to the strated irrationality.If rational actors can be press laid bare Nixon's secret bombing cam- counted on to refrain from such action, then paign against suspected Vietcong bases in only those with nonrational motives (e.g., neutral Cambodia. Student reaction was im- extreme ideologues) are likely to gravitate to mediate and intense. Protest peaked follow- social movements, and this will generally ing the May 4 killing of four antiwar pro- keep the movements small and politically testors at Kent State University, as countless impotent. colleges and universities shut down or oth- The latter implication of the rational erwise suspended normal activities in defer- choice perspective shades into the second ence to the deaths and the broader domestic body of work alluded to above. This is the crisis occasioned by the war. It was some- rich, if amorphous, work in political science thing of a surprise, then, when campus anti- that stresses the strategic preference of war protest waned markedly following the elected policymakers for broad centrist poli- reopening of campuses in the fall of 1970. cies that can attract majority support. As Nixon announced further troop reduc- Downs's (1957) work on the'"median voter" tions (even while maintaining his massive is both emblematic and influential in this re- air campaign) and stepped up his efforts to gard. When coupled with research showing forge a negotiated peace agreement with that movement activists are typically far Hanoi, antiwar activity declined still further. more ideological and extreme in their policy By the time the Paris Peace Accords were preferences than the population at large signed in January of 1973, the movement (Finkel, Muller, and Opp 1989; Nie and was largely moribund. Verba 1975), the implication is clear: Their characteristic nonrepresentativeness renders social movements as a MOVEMENT OUTCOMES/ inconsequential po- litical force. POLICY CHANGE The study of social movements emerged We began by highlighting the very different as a significant subfield within sociology conclusions reached by political scientists during the 1980s and has flourished ever and sociologists concerning the impact of since. For most of this period, the basic as- social movements on the policy process. sumption of sociological analysts of social With notable exceptions (e.g., Lipsky 1968), movements tended in the opposite direction the conventional view in political science from their colleagues in political science. has been one that attributes little influence Far from assuming the policy irrelevance of to social movements as vehicles of policy movements, sociologists asserted the impact change. At least two dominant lines of re- of movements without, however, typically search and theory in the field have con- subjecting this assumption to systematic em- verged to support this conclusion. The first pirical tests. is the voluminous literature spawned by In recent years this has changed, and while Olson's (1965) book, The Logic of Collec- more work is still needed on the topic, a dis- tive Action. Concluding that it is irrational cernible literature on "movement outcomes" for someone to engage in costly collective has begun to emerge in sociology (Amenta, action when they cannot be denied the ben- Carruthers,and Zylan 1992; Andrews 1997, efits of such action, Olson and his many dis- 2001; Banaszak 1996; Burstein 1979, 1993, ciples have fashioned a powerful perspective 1998; Burstein and Freudenburg 1978; But- that strongly implies the ineffectiveness of ton 1989; Cress and Snow 2000; Earl 2000; social movements as a force for policy Gamson [1975] 1990; Giugni 1998; change. There are actually two distinct im- McAdam 1989; McCammon et al. 2001; 700 AMERICAN SOCIOLOGICAL REVIEW

Mirowsky and Ross 1981; Schumaker 1975; tion of the war means we have picked for Soule et al. 1999; Steedly and Foley 1979; study an outcome of demonstrable signifi- Tarrow 1993). While certainly a positive de- cance. The undeclared nature of the Vietnam velopment, the focus of this burgeoning lit- conflict may have undercut Congress's con- erature remains elementary. The main ques- stitutionally mandated war-making power. tion motivating this work has been "Do so- Nonetheless, Congress remained the central cial movements matter?" That is, scholars policy arena within which contentious de- have simply sought to assess the impact (or bate over the prosecution of the war was lack thereof) of specific movements on par- waged. We distinguish two dimensions of ticular outcomes. Taken together, these vari- Congressional action. The first is the pace ous studies confirm the potential of social of action as measured by the monthly count movements to serve, under certain circum- of House and Senate roll calls on war-related stances, as effective vehicles of policy or measures. The second dimension is the di- other social change. That said, we know al- rection or valence of the action, defined as most nothing about the "how" of the ques- the proportionof votes cast for the pro-peace tion. Having found that some movements position(s) in any given month. appearto produce significant change effects, Below we hypothesize three different identifying the factors that account for the mechanisms through which antiwar protests variability of those outcomes becomes the may affect the pace and valence of Congres- central analytic task. Accordingly, we take sional action. These mechanisms reflect our up three mechanisms that have been pro- interest in differentiating two general ways posed as keys to understandingthe variable in which public protest may shape the policy outcomes of social movements. responses of state actors. These two ways are threat/disruption and persuasion. At its democratic asserts the central SEARCHING FOR MEDIATING core, theory of as the MECHANISMS importance persuasion primary mechanism of policy change. Juxtaposed to McAdam, Tarrow, and Tilly (2001) have this view is a well-developed theme in so- called for a move away from static structural cial movement studies that equates the effec- models to a search for the dynamic mecha- tiveness of social movements with their abil- nisms (and concatenated processes) that ity to achieve bargaining leverage through shape contentious politics. By mechanisms, the disruption (or threatened disruption) of the authorsmean "a delimited class of events public order. Needless to say, these two gen- that alter relations among specified elements eral emphases/accounts are in tension By in identical or closely similar ways over a deploying the following three mechanisms variety of situations" (McAdam et al. and various subsidiary hypotheses, we hope 2001:11). The implication of this approach to learn more about the role that persuasion for the study of movement outcomes should and threat/disruptionplay in mediating the be clear: Besides establishing an empirical impact of social movements. relationship between movement activity and We turn now to those mechanisms-dis- a given outcome, analysts should also attend ruptive protest, signaling, and public opin- as well to the identification and measure- ion shift-and test for their influence in pre- ment of whatever mix of mechanisms appear dicting the relationship between antiwarpro- to mediate the activity/outcome relationship. tests and Congressional voting on Vietnam Our outcome of interest is House and Sen- War-related measures. ate roll-call votes on Vietnam War-related issues. The choice of roll calls is for salutary DISRUPTIVE PROTEST two reasons. First, the close temporal con- nection between antiwar demonstrationsand A recurrentdebate in the literature concerns the roll calls makes the establishment of a the tactical effectiveness of disruptive ver- plausible causal connection a lot easier than sus more moderate forms of movement ac- if we were trying to assess the long-term im- tion. Starting with Lipsky's (1968) classic pact of a given movement. Second, the cen- work on "protest as a resource," many ana- tral importance of these votes to the prosecu- lysts have endorsed the general idea that THE POLITICAL IMPACT OF ANTIWAR PROTESTS 701 movement success typically depends on the work by Taft and Ross (1969) on labor un- ability of challenging groups to create rest in the United States and Snyder and "negative inducements to elite bargaining" Kelly (1976) on strikes in Italy. through the disruption of public order and Similarly mixed results characterize the the threat such disruption poses to the real- "urban disorders" literature. Some studies ization of elite interests (Astin et al.1975; appear to show moderately positive effects McAdam 1982; Tarrow 1998; Tilly, Tilly, at the city level for those urban areas that and Tilly 1975). This basic tenet underlies experienced rioting (Colby 1982; Eisinger McAdam's (1983) work on"'tactical innova- 1973; Jennings, 1979). Others appear to tion" as well as the influential argument ad- show negative effects (cf. Welch 1975). The vanced by Piven and Cloward (1993). For bulk of studies, however, assert no effect or McAdam, strategic innovation is not effec- mixed results that do not allow for a defini- tive in and of itself, but only to the extent tive answer to the question (Berkowitz 1974; that the introduction of a new tactic occa- Button 1978; Feagin and Hahn 1973; Kelly sions renewed disruptions of public order and Snyder 1980; Sears and McConahay that compel action by authorities. For Piven 1973). and Cloward, rapid expansions in welfare So our survey of existing empirical work rolls should generally be seen as ameliora- has turnedup mixed or inconclusive findings tive, if cooptative, responses to violence on on the relationship between social-move- the part of the urban poor. In a more con- ment disruption and policy outcomes. These temporaryvein, Cress and Snow (2000) find results could well be attributable to other that, in combination with other factors, dis- important contextual factors, as Button ruptive protest helps explain the success of (1978) and others argue (e.g., Cress and mobilization of the homeless in a number of Snow 2000). But there may be an important the cities they studied. methodological factor contributing to the Additional support for the mediating role mixed results as well. Quite simply, these of disruption/violence in specifying the re- various studies adhere to no single opera- lationship between movement action and so- tional definition of "disruption."In fact, in cial change comes from Gamson ([1975] most cases, there is no explicit operation- 1990). In his most controversial finding, alization mentioned in connection with the Gamson shows that groups which deployed concept. But one can imagine protest events either "violence" or "nonviolent constraints" being "disruptive" in a variety of ways. A tended to enjoy a higher rate of success than disruptive protest might feature violent tac- their more moderate or restrained counter- tics by demonstrators,or propertydamage as parts (Gamson [1975]1990: chap. 6). Be- a byproduct of a large, yet generally peace- cause of the univariate character of ful protest march; or injuries to protesters Gamson's analyses, several critics dis- resulting from overzealous police actions, counted his results, often singling out the and so on. In thinking through the various violence effect as a main suspect in this re- dimensions of public protests, we identified gard. However, on reanalysis, several inves- four features that might be thought of as tigators confirmed the general thrust of contributing to the overall "disruptive" in- Gamson's original finding and interpretation tensity of a protest event. These features are: (Mirowsky and Ross 1981; Steedly and (1) the use of violent tactics by demonstra- Foley 1979). tors, (2) the use of violence by law enforce- One can also glean some support for the ment personnel, (3) property damage as a re- disruption/violence thesis from two other lit- sult of the protest, and (4) injuries resulting eratures: those on strikes/strike outcomes from the protest. and on the "urbandisorders" that rocked the In distinguishing these various dimen- United States in the 1960s. With respect to sions, we aim to move beyond the undiffer- strikes, Shorter and Tilly (1974) found a entiated conception and assessment of dis- positive correlation between the use of vio- ruption. We seek, instead, to explore the lence and strike outcomes in their study of variable impact of different components of French strike activity. Several other studies, public protest events. We do so by positing however, yielded opposite results, including two contrasting pairs of the above protest 702 AMERICAN SOCIOLOGICAL REVIEW dimensions. In the first, we differentiate litical action" in which elected officials use violence by demonstratorsfrom violence by mass political activity to better understand police. The second pairing-property dam- the policy preferences of the electorate. age versus injuries-mirrors the first, with While Lohmann is at least as interested in property damage resulting mostly from ex- using her model to explain the incentives that treme action by demonstrators, and injuries allow individual activists to overcome the (overwhelmingly to demonstrators) owing "free-riderproblem," she also seeks to char- to the aggressive policing of protest. These acterize the functional benefits to elected of- pairings can also be tied to the general dis- ficials of attending to the information con- tinction between threat and persuasion. Al- veyed by mass political activity. She begins though none of the four dimensions match with a familiar characterization of elected clearly to the notion of persuasion, violence officials: "The political leader wishes to by demonstrators and property damage are make a decision that is advantageous for a reasonable proxies for threat-both repre- majority of the population.... A leader fully sent extreme and/or violent forms of protest informed about the state of the world would by movement groups. Drawing on the tradi- choose a policy preferred by a majority un- tional movement literature, our first hypoth- der the true state of the world" (Lohmann esis posits a significant impact of these 1993:320). But how does a leader discern the more threatening dimensions of protest "true state of the world"? Lohmann's insight events. is to recognize that public protest may serve as an informative to 1: Protest events vio- signal policymakers Hypothesis featuring about the state of the world. But how to read lence demonstrators and/or by property those For one will be related to an increase in signals? Lohmann, important damage cue comes from the numberof "activist mod- House and Senate and an in- voting erates" involved in action. In her crease in the results of those protest pro-peace "one of the model is that votes. words, implication the political leader shifts policy if the politi- Hypothesis 1 rests on a simple view of the cal action turnout exceeds a critical thresh- relationship between disruption and Con- old" (Lohmann 1993:322). This leads to our gressional action; that is, the more extreme third hypothesis. or violent the forms of action, the more and more favorable the Congressional response. Hypothesis 3: There will be a significant, But the movement literature suggests a positive relationship between size of second, more qualitative link between dis- protest event and the pace and pro-peace ruption and state response. In particular, valence of House and Senate roll calls. studies of the civil rights movement suggest That is, large events will be related to that it is not disruption per se, but disruption more roll calls and an increased likeli- characterized by violence directed against hood of pro-peace outcomes. the movement that is especially productive of favorable government response (McAdam Beyond the "size of the protest move- 1982, 1983). This account suggests a second ment" (Lohmann 1993:319), Lohmann of- way in which disruption may be linked to fers few specifics on how elected officials state action. read public protest. But she suggests another meaningful feature of mass political activity Hypothesis 2: Protest events featuring vio- when she writes that "the political leader lence by law enforcement officials and/ discounts the observed turnoutfor extremist or injuries to demonstrators will in- political action and shifts policy [only] if the crease the pace and pro-peace valence estimated number of activist moderates ex- of House and Senate roll calls. ceeds a critical threshold" (p. 319, italics added). If elected officials are motivated to read for whatever of SIGNALING public protests signs majoritarianpolicy sentiment they may re- Lohmann (1993:319) proposes a "signaling veal, it makes sense-following Lohmann- model of informative and manipulative po- to hypothesize a negative effect of protests THE POLITICAL IMPACT OF ANTIWAR PROTESTS 703 characterized by the use of especially vio- empirical evidence supporting a reasonably lent or "extremist" tactics. Although strong positive association between opinion Lohmann's stress is on the revealed "policy shift and policy change. For instance, in preferences"of activists, it seems reasonable their exhaustive survey of the opinion/policy to suggest that extremist tactics (opera- link, Page and Shapiro (1983:175) find tionalized here as violence by demonstrators "considerableevidence that public opinion is and/or property damage) would be inter- often a proximate cause of policy." More re- preted by policymakers as synonymous with cently, Stimson et al. (1995) have assessed extreme, and therefore nonmajoritarian, the dynamic responsiveness of all branches policy positions. Use of such tactics are thus of government to shifts in public opinion likely to motivate elected officials to signal over the period 1956-1993, and, allowing opposition to the protesters by voting to sus- for variable responsiveness by period and in- tain the policy status quo. stitution, they "find [strong support] that policy responds dynamically to public opin- Hypothesis 4: Protest events featuring vio- ion change" (p. 543). lence by demonstrators and/or property All of this suggests a less direct link be- damage will be related to a decrease in tween protest, public opinion, and policy House and Senate voting and a decrease change than the one imagined by Lohmann. in the pro-peace results of those votes. Here protest does not work directly-as a signal-to change the policies of leaders, but Note that Hypothesis 4 runs exactly rather does so only indirectly by first shift- counter to the functional view of disruption ing public opinion in the direction of move- sketched in Hypothesis 1. That is, ratherthan ment goals. Once opinion has shifted in this viewing extreme and/or violent forms of way, it then acts, in the manner consistent protest as strategically effective, Hypothesis with the aforementioned research (and clas- 4 posits that certain forms of disruption sig- sical democratic theory) to alter the policy nal extremist views, and thus work to the preferences of those public officials who are detriment of the movement. These opposite subject to electoral pressures. This is the predictions can be traced back to the funda- third and final mechanism-public opinion mental distinction between threat and per- shift-alluded to above. suasion. Hypothesis 1 assumes the strategic Alas, as noted previously, political soci- effectiveness of threat, while Lohmann's no- ologists have evinced far less interest in pub- tion of signaling is consistent with the stress lic opinion than have political scientists. Es- on persuasion in traditional democratic pecially regrettable is the almost total igno- theory. The general notion of persuasion also rance of the topic among social movement underlies our third and final mechanism: scholars.' Indeed, the absence of attention public opinion shift. makes it somewhat difficult to formulate hy- potheses regarding the relationship between and PUBLIC OPINION SHIFT protest activity public opinion. Drawing on sketchy scholarly evidence, we offer the The link between public opinion and policy following two provisional hypotheses: outcomes has interested political scientists 5: Protest events that feature vio- (and to a lesser extent, economists) for years Hypothesis (Arnold 1990; Arrow 1963; Bartels 1991; lence by police and/or injuries to dem- onstrators will be related to Dahl 1956; Erikson, Wright, and McIver positively increases in the 1993; Monroe 1998; Page and Shapiro 1983; public opinion against Sen 1970; Stimson, MacKuen, and Erikson war. 1995). Viewing elected officials as rational Hypothesis 6: The simple pace of protest actors intent on staying in office, these events will be positively related to in- scholars have hypothesized that politicians creases in public opposition to the war will generally modify their own policy pref- as measured by public opinion polls. erences to fit shifting public opinion in an effort to retain electoral support. Theory Burstein(1998) is conspicuousamong social aside, several broad studies have yielded movementscholars in doing workon the topic. 704 AMERICAN SOCIOLOGICAL REVIEW

Hypothesis 5 rests on what might be and France over much of the nineteenth and termed the "functional victimization" view early twentieth centuries. articulated by scholars of the civil rights Event research has now developed into movement. Indeed, that account posits the one of the methodological staples of social identical two-step process imagined here. movement researchers. Indeed, many of the That is, violence against demonstrators led classic empirical works in the field make use to increased public attention and support for of the technique (Andrews 1997; Burstein insurgents, which, in turn, promptedincreas- and Freudenburg 1978; Duyvendak 1995; ingly favorable government action in rela- Jenkins 1985; Jenkins and Perrow 1977; tion to the movement. Hypothesis 6 draws Koopmans 1993, 1995; Kriesi et al. 1995; exclusively on Schuman's (1972) empirical McAdam 1982, 1983; Olzak 1987, 1989, work on the changing nature of public opin- 1992; Tarrow 1989). The larger project of ion about the Vietnam War in relation to the which this study is a part is in the main tra- antiwar movement. Schuman shows that in- dition of protest event research. In design- creasing protests led to both growing oppo- ing the project, however, the principal inves- sition to the war and a decided public opin- tigators-who , in addition to the first au- ion backlash against the movement. thor, include John McCarthy, Susan Olzak Our final hypothesis concerns the link be- and Sarah Soule-sought to improve on the tween public opinion and House and Senate approach by adding a few methodological voting. The paucity of research on protest innovations of their own. and public opinion made it difficult to for- To be fair, protest event research has been mulate the previous two hypotheses, but the criticized on a number of grounds by social vast literature on the link between public movement and other analysts (Danzger opinion and policy outcomes makes this last 1975; Davenport 2001; Franzosi 1987; hypothesis easy to state: McCarthy, McPhail, and Smith 1996; Mueller 1997; Oliver and Myers 1999; Hypothesis 7: Increased public opinion Snyder and Kelly 1977). At the heart of against the war will be positively related these critiques is skepticism that newspa- to increases in the pace of Congres- pers represent anything like an unbiased sional voting and the likelihood of pro- source of information on collective action peace outcomes from the votes. events. Much of this skepticism is informed by systematic evidence attesting to various forms of bias in accounts of PROTEST EVENT RESEARCH newspaper pro- test events. For example, McCarthy et al. The term protest event research refers to (1996), show that size, location, and vio- empirical work on social movements that lence predict newspaper coverage of protest utilizes content coding of newspaper ac- events. In their study of collective action in counts of movement events to study the dy- Madison, Wisconsin, Oliver and Myers namics of contention involving movement (1999) find several sources of bias in the and other actors (i.e., countermovements, coverage of "public events." They find that state actors, etc.). While some social scien- large events, or those involving "conflict," tists had made use of a rudimentaryform of are more likely to be covered by local events analysis to study the patterning, cor- newspapers. In addition, business-spon- relates, and consequences of the "urbandis- sored events and those taking place in more orders" that took place in the United States central locations in Madison were more in the 1960s (cf. Lieberson and Silverman likely to be reported in the papers. 1965; Spilerman 1970, 1976), Tilly and vari- Davenport's (2001) study of Black Panther ous colleagues pioneered the method as a activity in the San Francisco Area in the tool for social movement research (Shorter late 1960s/early 1970s bears mention as and Tilly 1974; Tilly et al. 1975). Making well. Rather than predicting coverage on systematic use of newspaper and other pub- the basis of various factors (e.g., size, vio- lished sources of "collective action events," lence, etc.), Davenport simply tracks Pan- Tilly and his colleagues mapped the ebb and ther activity using five different newspa- flow of popular protest in Italy, Germany, pers, the The New YorkTimes, The Oakland THE POLITICAL IMPACT OF ANTIWAR PROTESTS 705

Tribune, The Berkeley Barb, the Black Pan- same assumption. On the contrary, when ther Intercommunal News Service, and the events are deployed as an independent vari- Sun Reporter.2 What is noteworthy in able, the method honors the central insight Davenport's data is the absence of any real that informs the various critiques reviewed convergence in the temporal distribution of above. That is, rather than treat newspapers events produced by the five newspapers. as a passive channel of communication, the What this marked divergence in activity use of events as a predictor variable ac- patterns results from is not clear. Davenport knowledges the active role of the print me- does not attempt to resolve the inconsis- dia as an influence on the dynamics of con- tency by attributing greater accuracy to one tention. or more sources over the others. Instead, as The empirical case at issue here can be his provocative title suggests, these various used to illustrate our point. The principal sources of data represent a real-life example dependent variable in our analysis is Con- of the "Rashomon Effect." Like the wit- gressional voting. If we were to try to iden- nesses in the movie of the same name, tify the main sources of influence on votes Davenport's papers represent distinctive by individual Senators or members of Con- takes on a complex phenomenon rather than gress, what would the prime candidates be? consistent windows on an underlying pat- It might be nice to believe that individual tern. The methodological implications of conscience would be decisive in this regard, these findings are sobering to say the least. but most Congressional analysts have long We take this evidence of newspaper bias asserted the primacy of two more self-inter- very seriously, and needless to say, it poses ested influences. One is party loyalty, and serious challenges to those who would use the other is the perceived "electoral returns" such data to study social movement or other to a given pattern of voting. The latter fac- public collective action. But the challenges tor involves aligning one's voting record to vary considerably depending on the specific the preferences of whatever electoral con- nature of the event research. For instance, stituency the Senator or member of Con- comparing the results of "national" versus gress sees as key to his or her reelection. more local studies of protest activity would But how do these elected officials monitor seem to support a straightforward conclu- electoral preference? Today, regular track- sion: All things equal, the more local the fo- ing polls and focus-group analyses have be- cus of attention, the more credible the use come key to this effort. But neither tech- of newspapers as a source of event data. nique was widely used in the Vietnam War The other strong conclusion that we draw era. Instead, in trying to gauge how the war from the multiple critiques of the method is was "playing in Peoria," Senators and that studies focusing on events as a depen- members of Congress would have relied dent variable are far more questionable than primarily on letters from constituents in those treating those same events as an inde- their home states, irregular public opinion pendent, or predictor, variable. To treat polls, and media reports of pro-war and an- events as an outcome variable would seem tiwar demonstrations. And no newspaper to require that the analyst view events as a probably played a more central role in these credible proxy for some underlying tempo- assessment efforts than The New York ral pattern of action, an assumption strongly Times. Indeed, the criticisms of the paper challenged by the various studies we re- by scores of politicians during these viewed above. To study the impact of years-Nixon, Johnson, and Agnew are events-more accurately the impact of re- only the most prominent to come to mind- ported events-on some later outcome vari- only serves to make the point. In its cover- able (e.g., public opinion, state action, etc.) age of reported events, no less than in its does not oblige the analyst to make this editorial policy, the Times was an active in- fluence on public discourse and policy- 2 the war. our efforts ThePanther Speaks was the official newspa- making during Indeed, per of the Black PantherParty during the period to assess the impact of antiwar protests on in question.The Berkeley Barb was a prominent Congressional voting depends on the medi- left "underground"paper during the sameperiod. ating influence of the Times. 706 AMERICAN SOCIOLOGICAL REVIEW

THE STUDY third and final qualifying feature of the event concerns the identity of the initiating actors. Since 1997, the aforementioned consortium We accept two types of actions: those initi- of social-movement researchers have been ated by named social movement organiza- involved in a massive data collection effort tions (SMO), or those by any "actor per- designed to better understand the causes, forming in a manner outside that actor's of- consequences, and changing nature of pub- ficial capacity to express a grievance, claim, lic protest in the United States in the post- belief, or opinion" (Uhrig and Van Dyke World War II period. To date, this effort has 1996:2). The intent here was to demarcate a centered on assembling time-series data sets terrain of protest action distinct from other on public protest events and a host of theo- more conventional forms of claim-making or retically relevant covariates for the period other political activity. 1960-1980. Eventually, however, we hope to Although not identical to the coding con- extend these data sets both backward and ventions employed by other event research- forward in time to encompass the full 50- ers, these definitions are nonetheless of a year period, from 1950 to 2000. piece with earlier work in the tradition. Even while employing these definitions, however, we treat events than PRO lf. i EVENTS differently many past researchers. Traditionally, event researchers As with all protest-event research, the em- have utilized simple counts of events to mea- pirical centerpiece of the project has in- sure movement activity. Our data will allow volved the ongoing identification and cod- us to do that as well, but in addition to these ing of newspaper accounts of public protest event counts, however, we will want to as- events. The New YorkTimes has functioned sess the impact of movement activity by at- as the newspaper of record as we have tending to various dimensions or features of worked to assemble the 20-year event time protest events. series (1960-1980) for which we now have coded and cleaned data. We will how- not, DISRUPTIVE EVENTS ever, make use of all of this data in this ar- ticle. Instead, we will rely only on events As noted above, a good many analysts have from 1965 to 1973 and only for one issue argued that social movements become a sig- area: peace. nificant force for social change only when We define a public protest event as any they succeed in generating "bargaining le- event or action "in which individuals collec- verage" through the disruption of normal so- tively make a claim or express a grievance cial or political routines. But simple event on behalf of a social movement organization counts tell us next to nothing about the dis- or [aggrieved] social category" (Uhrig and ruptive intensity of the aggregated actions. Van Dyke 1996:1). Our operational defini- To get at this theoretically important aspect tion of a protest event specifies three "defin- of movement actions, we code all of our ing features" that the event must exhibit to events along the following four dichotomous be included in the data set. First, the act must dimensions4: (1) injuries-were there any involve a group of people rather than a injuries reported in connection with the single individual. Second, it must involve event? (2) violence by demonstrators-did "contentious claim-making." That is, at the heart of the event there must be, implicitly in pro-waras in antiwarprotests during this pe- or explicitly, a demand for either a change riod,the paucityof the formerhas led us to focus in society or an avowed desire to resist a exclusivelyon antiwarprotests. 4 We intendedto a sixth di- proposed change. In the case of the Vietnam originally employ this would mean that would mension in this part of our research. The sixth War, protests dimension focused on the of need to be motivated either a desire to presence/absence by deaths in connection with the event. Assuming the nature of U.S. change war-making policy one values the generally nonviolent character of or to resist efforts to change that policy.3 The American politics, the good news is that too few events featured deaths (three by actual count) to 3 But while we are theoretically as interested allow us to employ this dimension in the study. THE POLITICAL IMPACT OF ANTIWAR PROTESTS 707 the event feature the use of violence by dem- compiled by Rosenthal and Poole (1991) onstrators? (3) violence by police-did the entitled "United States Congressional Roll event feature the use of violence by police? Call Voting Records, 1789-1987: Reformat- and (4) property damage-did the event re- ted Data."5The data set includes every roll- sult in property damage? call vote taken in the Senate or House be- In addition to these four dimensions, we tween 1789 and 1987. Each entry in the differentiate protest events based on the data set includes information about the date number of movement participants involved. the vote was taken, the sponsor of the vote, We identify those events with more than and the number of yes and no votes re- 10,000 participants and those with less than corded for the measure. The entry also in- 1,000 participants. Our use of these five di- cluded a brief summary of the measure that mensions (the four above plus size) is allowed us to code the valence of the vote's straightforward. Besides employing simple outcome. To extract from this massive data event counts, we want to determine whether set the votes relevant to our case, we first the predictive power of our models is in- used the issue codes devised by Poole and creased when we substitute any one or vari- Rosenthal (1997). To ensure that we did not ous combinations of these dimensions for miss relevant measures, we included in our the event counts. The theoretical signifi- initial list of "candidate votes" all roll calls cance of these substitutions is worth high- between 1965 and 1973 that fell into any lighting. If, as many have argued, move- one of the following four issue codes: ments owe their force as social change ve- "VietnamWar," "Selective Service," "Peace hicles to their disruptive capacity, then these Movements/Pacifism/Anti-Military," and dimensions-especially those capturing espe- "Communists/Communism/Un-American cially extreme or violent movement action Activities." We then winnowed the large (e.g., violence by demonstrators, property pool of candidate events by carefully read- damage)-ought to produce stronger predic- ing the summary of each one and including tive effects than models employing simple in the final data set only those votes that event counts. bore directly either on (1) the prosecution of the war, or (2) the workings and legisla- tive status of the Selective Service THE DEPENDENT VARIABLE: HOUSE AND System. Votes that bore a more relation- SENATE ROLL-CALL VOTES tangential ship to the Vietnam conflict (e.g., those We derive our two main dependent variables having to do with the repression of domes- from all House and Senate roll-callvotes on tic movements or U.S. relations with Com- Vietnam War-related measures between munist countries such as China and the So- 1965 and 1973. We operationalize pace as viet Union) were excluded from the final the monthly counts of all relevant roll calls, data set. and valence as the proportion of legislators These procedures yielded a final data set voting the peace position in a given month. of 236 votes, 80 taken in the House and 156 In employing these two dependent variables, in the Senate. For each vote, we recorded the we seek to distinguish between two criti- date, the distribution of yes/no votes, and, in cally important forms of political influence: most cases, the valence of the outcome. agenda setting-or "agenda responsive- Forty-seven of the 236 votes yielded no clear ness," as Schumaker (1975) termed it-on valence and were therefore coded as miss- the one hand, and-policy change on the ing on this dimension. other. Too often, in our view, movement 5 analysts adopt some version of policy We gratefully acknowledge the generosity change as the only salient metric for assess- shown us by Howard Rosenthal and Keith Poole. Besides their data set avail- ing impact, while ignoring the fact that making extraordinary is both a achieve- able to us-a courtesy that is generally available agenda setting significant to the also were ment in its own and a for scholarly community-they gra- right prerequisite cious enough to answer numerous questions, policy change. which allowed us to better adapt our method- These two roll-call measures were gener- ological procedures to the ones they used in com- ated using a unique and invaluable data set piling their data set. 708 AMERICAN SOCIOLOGICAL REVIEW

Table 1. Descriptive Statistics of Variables in Models Predicting Vietnam War-Related Congressional Roll-Call Voting 1965 to 1973 Observations Variable (Months) Mean S.D. Min. Max. Data Source CongressionalRoll-Call Vot Roll-call counts 2.19 5.43 0 45 Rosenthaland Poole (1991); Poole and Rosenthal(1997) Pro-peacevotes 42 40.73 21.32 .23 93 Rosenthaland Poole (1991); Poole and Rosenthal(1997) Anti-VietnamWar Protest Events, 1965 to 1973 All events 97 10.61 13.69 0 90 New YorkTimes (1965-1973)

Types of Events: Events with more 97 .38 .89 0 4 New YorkTimes (1965-1973) than 10,000 participants Events with violence 97 .96 2.20 0 15 New YorkTimes (1965-1973) by demonstrators Events with violence 97 1.00 2.20 0 14 New YorkTimes (1965-1973) by police Events with injuries 97 .75 1.60 0 9 New YorkTimes (1965-1973) Events with property 97 .64 1.55 0 10 New YorkTimes (1965-1973) damage Other Covariates U.S. militarydeaths in 97 533.95 514.86 0 2,413 VietnamWar MemorialFund Vietnam Call-ups in Selective 97 15,494.74 10,778.74 1,666.(67 35,366.67 Selective Service System Service (1967-1974) Conscientiousobjectors 97 2,850.66 1,205.19 1,616.(00 5,456.5 Selective Service System (1967-1974) Percentageof public 22 51.69 14.82 26.338 76.99 Gallup International,Inc. believing war is (1965-1973) a mistake Publishedarticles 97 73.52 42.01 3 171 Reader's Guide to Periodical relatedto Vietnam Literature

Period Dummy Variables: January1965 to October 1967 [Referencecategory] (PentagonMarch) (tl) November 1967 to 0 1 December 1968 (t2) January1969 (Nixon election) 0 1 to May 1970 (Kent State) (t3) June 1970 to December 1973 (t4) 0 1

Our measure of pace is simply the that month (e.g., "yes" votes if the framing monthly counts of these roll-call votes. Our of the roll call is pro-peace; "no" votes if the valence measure is a bit more complicated. framing of the roll call is pro-war) and then We construct it in two steps. First, using divide the total by the number of roll calls in only those individual votes cast by members the month. In order to have a compatible of Congress for which the valence is clear, measure for both chambers, we scale the we sum the total of pro-peace votes cast in pro-peace votes in the House to fit with the THE POLITICAL IMPACT OF ANTIWAR PROTESTS 709 metric used for the Senate (from 0-435 to 0- measure of shifting U.S. involvement in 100). The descriptive statistics for these two Vietnam. We draw information from a se- dependent variables are shown in Table 1. ries of publications entitled Semi-Annual Report of the Director of Selective Service Government Office, OTHER COVARIATES (U.S. Printing July 1967-December 1973). The report gives us Besides broadening our analysis of events to call-up numbers only for every half a year, include an assessment of the impact of dis- so we assign the mean of the semiannual ruptive events on the pace and outcome of count in our unit month. It is also weighted Congressional action, the study is marked by by 1/100 to provide reader-friendly coeffi- another, if not quite so distinctive, innova- cients for the regression models. tion. We refer to the general project's com- CONSCIENTIOUS OBJECTORS. Conscien- mitment to assemble a full slate of time-se- tious objectors were those individuals who ries covariates required for a systematic em- registered for a deferment or exemption pirical assessment of movement outcomes. from the draft "on the grounds of religious These covariates include over-time measures training and belief' (U.S. Government Print- of public opinion change, mass media cov- ing Office, 1967:9). In the Vietnam War era, erage of the war, and the pace of the Viet- such registrationfor deferment or exemption nam War (e.g., combat deaths, troop call- became a widespread tactic of protest and re- ups, etc.); the pace of the war being thought sistance against the war. The mean semian- of as the objective pressure or "demand"for nual count is assigned to each month and is protest against the war. Before turning to our weighted by 1/100 in the models. results, we take up each of these covariates PUBLIC OPINION. Among various polls in turn. (Table 1 shows the descriptive sta- released during the Vietnam conflict, the tistics for these covariates.) question that was asked most frequently and U.S. MILITARY DEATHS IN VIETNAM. We persistently is the following one by Gallup: use the monthly count of military deaths in "In view of developments since we entered Vietnam to measure the pace of the war. It is the fighting in Vietnam, do you think the reasonable to hypothesize that the more U.S. made a mistake in sending troops to deaths, the stronger the opposition to the war fight in Vietnam?" (Gallup International, among the general public and U.S. policy- Inc. 1965-1973). There are 22 data points in makers alike. Such effects need to be con- the period of interest. Because the data show trolled in our examination of the impact of a strong linear trend over time (R2 = .83), protests on congressional action. We rely on the percentages are regressed on time to pro- statistics provided by an organization called vide estimates for months when the question the Vietnam WarMemorial Fund. From their was not asked. For months for which obser- website (in the searchable section of "Virtual vations are available, they are used; for other Wall") we obtain death statistics for the 108 months, regression estimates (OLS) are months between 1965 and 1973.6 In the re- used. This treatment of public opinion as an gression model, we change the unit from per independent variable follows the approachto person to per 100 person in order to obtain this issue by Burstein and Freudenburg reader-friendly coefficients. (1978:117). In our analysis, however, we CALL-UPS BY THE SELECTIVE SERVICE. also treat public opinion as a dependent vari- Call-ups are the number of young men able to see whether protests have an impact drafted by the U.S. military at various on public opinion. In those models, we in- points during the study period. This is a clude only the 22 data points mentioned above. In order to account for the intrinsic linear trend, we use as a dependent variable 6 We accessedthe websitein December2000; the difference between the observed the statistics could because of percent- change slightly and the value estimated from the above newly available data on Vietnam deaths. We ages shouldalso cautionthat we do not knowwhether mentioned regression line. these numbersare the same as the numbersof NUMBER OF ARTICLES PUBLISHED IN deathsaccessible to the public duringthe Viet- PERIODICALS. To measure media attention namconflict. to the Vietnam Warissue, we count the num- 710 AMERICAN SOCIOLOGICAL REVIEW ber of related articles published in the holds; autocorrelation is not significant for month. Under the alphabetically ordered our outcome measures. Substantively, this headings, each volume of the Readers' means that the likelihood of congressional Guide to Periodical Literature (1965-1974) voting in a month is not contingent upon lists the title, publishing date, and name of how many roll calls have occurred in previ- the periodical for all the articles published ous months. The second issue is whether the in the previous year. We photocopy the outcome variable has an intrinsic trend that pages under the heading "Vietnam" and may not be explained away by the indepen- count the number of published articles in the dent variables. If we examine the shape of 108 months of our period.7 It is weighted by the temporal distribution of roll calls, there 1/100 to yield more accessible coefficients is a discernable patternof period effects (see in our models. Figure 1): There appears to be an increase in Congressional action early in 1967, follow- a of and an- METHOD: TIME-SERIES ANALYSIS WITH ing two-year period inaction; other of late in ZERO-INFLATED POISSON REGRESSION period heightened activity 1970 that does not subside until 1973. It is The result of our data collection effort is a empirically unknown whether such a pattern time-series data set spanning the 108 months can be accounted for by the protest events of 1965 through 1973 and consisting of and our other covariates. With this consider- monthly counts for most of the variables de- ation in mind, we introduce categorical vari- fined above. In most of our models we lag ables dividing the 108 months into four pe- observations in the previous month to pre- riods (in turn coded into three dummy vari- dict the likelihood of congressional voting in ables). We use three historical events as cut- the month following. ting points: the October 1967 Pentagon Before we choose appropriate models to March, Richard Nixon's January ascension generate estimates, however, we discuss to the White House, and the May 1970 kill- three issues germane to an understandingof ings at Kent State. In doing so, our model outcome variables in time-series data will allow the rate of likely roll-call voting (Wonnacott and Wonnacott 1990). The first to differ in the four periods. is autocorrelation-whether the observa- We use zero-inflated Poisson regression tions in different units are correlated to one (ZIP) models to generate estimates for our another. By conducting a Durbin-Watson predictions. While event-count data in prin- test, we find the assumption of independence ciple are suitable for Poisson regression, there are two problems with using regular 7 Thereare two reasonsthat such countsmay Poisson regression in our case. First, our not reflectthe exact numberof Vietnamwar-re- outcome variables violate a key assumption lated articles.First, some articlesrelated to the for using poison regression: The variance war may be classifiedunder headings other than must equal the mean. As shown in Table 1, "Vietnam,"such as "defense,""medical relief the variances of the three outcome variables work,"and so on. The task would become un- are greater than the mean, a problem called manageableif we triedto searchfor possibleen- tries under all related This overdispersion. Second, there is a substan- possibly headings. tial number of months in which our may resultin an undercountof relevantarticles. depen- On the otherhand, the articlesunder the heading dent variables will have zero value; as a re- "Vietnam"are not always directlyrelated to the sult regular Poisson regression will war. Some entries are about the geographical, underpredict the chances of being a zero cultural,and political backgroundof Vietnam. count. ZIP models are designed to adjust the We include all the entriesunder "Vietnam" be- difference between the mean and variance, we find difficultto set to cause it criteria gauge as well as to add more predictions of zero how close an is relatedto the war read- entry by values on the outcome variable (Greene ing its title. This may result in an overcountof articles.Because these two sourcesof errorare 1997; Long 1997). ex- counterto one another,we are confidentthat the The general form of the models can be measureis a broadlycredible proxy for temporal pressed as the following equation: variationin mediaattention paid to the Vietnam War. Li = exp(xiP), (1) THE POLITICAL IMPACT OF ANTIWAR PROTESTS 711

100 Protestevents 90 ..-- Roll-callvotes 80

. 70' c > 60.60 '5 50 Q) . E 40 z 30 20 10

rtII kArtjJJA- -- % a w ~ . -~ ~0-r w - ~W ~1"~ - -- 3L A^UL~A-- MM--W =M . .. Ai In I - - - - - I l l I . t) ED N- O D a 0 > cY Ce ID

Figure 1. Temporal Distribution of Protest Events and Congressional Roll-Call Votes, 1965 to 1973 where pi is the predicted count of roll calls ports the same models applied to the net di- in ith month; xi is a set of values of our rectional outcome of the votes (e.g., "pro- covariates in the given month; and p is a peace" or "pro-war"). vector of parametersto be estimated. Maxi- With Hypotheses 1 and 2 we distinguish mum-likelihood estimates are used. That is, two very different ways in which disruptive ,1 is a set of parameters that maximizes the protest may shape the pace and direction of likelihood of the prediction and whose func- Congressional roll calls. The first hypothesis tion is based on the two-stage probability rests on the notion that social movements function of the ZIP model (see Long derive their effectiveness by posing a disrup- 1997:243-45). tive threat to the established order. Accord- ingly, the more extreme the event-as mea- sured the of demonstrator vio- RESULTS by presence lence and/or property damage-the stronger We turn now to the results of our various the predictive relationship with our two de- analyses. Most of our findings are presented pendent variables. The idea is simple: in two tables: Table 2 shows various models Through extreme disruptionsof public order, predicting the pace of Congressional voting; otherwise powerless groups are able to com- Table 3 shows the same models predicting pel favorable state action motivated by the the direction or valence of House and Sen- need to restore "business as usual." The re- ate roll calls. We organize our results, how- sults, as reported in Models 2 and 3 in these ever, not according to these two dependent two tables, support interesting, but opposite, variables but rather in relation to the three conclusions. More disruptive events do in- mechanisms and related hypotheses dis- deed predict a rise in the proportion of pro- cussed above. peace votes, but these same events bear a significant negative relationship to the over- all of Increased DISRUPTIVEPROTEST pace Congressional voting. disruption may have inclined members of We begin by assessing the relationship be- Congress to vote pro-peace, but in this case tween various measures and dimensions of at least, it also seems to have depressed the disruptive protest and the pace and direction rate at which the House and Senate consid- of House and Senate voting on war-related ered such measures. roll calls. Table 2 reports the results of vari- Hypothesis 2 posits a very different view ous models predicting the monthly count of of the link between disruption and govern- roll-call votes (e.g., pace), and Table 3 re- ment action. Here the spur to increased and 712 AMERICAN SOCIOLOGICAL REVIEW

Table 2. Maximum-Likelihood Estimates of Poisson Regression Models Predicting the Monthly Count of War-Related Congressional Roll Calls, by Protest Events

IndependentVariable Model 1 Model 2 Model 4 Model 3 Model 5 Model 6 Intercept -5.600** -3.873** -4.811** -3.015** -2.239* -3.019* (1.183) (1.109) (1.285) (1.132) (1.082) (1.240) Protest Events All events .016 -.003 .018 -.018* -.030 -.008 (.015) (.037) (.013) (.009) (.016) (.015) Violence by demonstrators -.122 -.101 (.069) (.080) Violence by police -.157** .122** (.036) (.040) Propertydamage -.469** -.355** (.122) (.130) Injury .770** .613** (.085) (.091) More than 10,000 .613** .566** .362** participants (.078) (.078) (.082) Other Covariates U.S. militarydeaths -.028 -.020 .080 -.012 .018 .062 in Vietnam (.037) (.037) (.041) (.041) (.041) (.042) Call-ups in Selective .003* .000 -.002 .002 .000 .001 Service (.001) (.002) (.002) (.002) (.002) (.002) Conscientiousobjectors .032** .023* .025** .015 .010 .012 (.009) (.009) (.009) (.010) (.010) (.010) Percentageof public .087** .068** .083** .048* .040 .052* believing war is mistake (.023) (.022) (.023) (.023) (.022) (.023) Publishedarticles related 1.016** 1.023** -.151 .502 .618 -.152 to Vietnam (.290) (.295) (.338) (.320) (.324) (.342)

Period Dummy Variables:a t2 -.946 -1.051* -3.856** -.996 -1.182* -2.766** (.503) (.521) (.559) (.538) (.552) (.584) t3 -1.362* -1.037 -1.732* -2.093** -1.627** -1.598** (.589) (.595) (.568) (.592) (.595) (.564) 4 -.767 -.777 -.814 .174 .113 .039 (.838) (.860) (.816) (.879) (.901) (.869)

-2 Log-likelihood 564.00 541.02 465.12 500.93 489.76 446.68 (d.f.) (9) (11) (11) (10) (12) (12) Pseudo R2 .21 .25 .35 .30 .32 .38 Note: Numbersin parenthesesare standarderrors; number of observations= 97. a See Table 1 for definitions of period variables. *p < .05 **p< .01 (two-tailed tests) more favorable state action comes not from demonstrators-to the pace and direction of disruption per se but from the apparent vic- House and Senate voting. Reflecting a very timization of demonstrators. We test this different mechanism, the results of these idea by looking at the relationship of two analyses yield findings inconsistent with kinds of events-those that feature violence those reported above. Models 2 and 3 in by police and those resulting in injuries to Table 2 provide strong support for the vic- THE POLITICAL IMPACT OF ANTIWAR PROTESTS 713

Table 3. Maximum Likelihood Estimates of Poisson Regression Models Predicting the Monthly Pro- Peace Votes in Congress, by Protest Events

IndependentVariable Model 1 Model 2 Model 4 Model 3 Model 5 Model 6

Intercept 3.089** 3.173** 3.190** 2.539** 2.851** 2.577** (.358) (.364) (.359) (.385) (.395) (.385) Protest Events All events .006 .013* -.011 .012** .014* -.005 (.003) (.006) (.006) (.003) (.006) (.006) More than 10,000 -.148** -.082* -.176** participants (.034) (.037) (.035) Violence by .055* .049* demonstrators (.002) (.022) Violence by police -.092** -.073** (.018) (.019) Injury .085** .097** (.031) (.034) Propertydamage .095** .113** (.036) (.032) Other Covariates U.S. militarydeaths -.077** -.083** -.064** -.080** -.083** -.061** in Vietnam (.012) (.011) (.013) (.011) (.011) (.013) Call-ups in .002* .003** .003** .002** .003** .002** Selective Service (.001) (.000) (.000) (.001) (.001) (.000) Conscientiousobjectors .000 -.002 -.001 .005 .001 .001 (.003) (.004) (.003) (.004) (.004) (.004) Publishedarticles -.070 -.083 -.081 .082 .009 .030 related to Vietnam (.140) (.149) (.176) (.146) (.156) (.177) Percentageof public believing war is mistake .007 -.000 .005 .018* .007 .016* (.008) (.007) (.008) (.008) (.008) (.008)

Period Dummy Variables:a t2 .641** .910** .364** .573** .812** .138 (.150) (.160) (.194) (.145) (.162) (.203) t3 .363 .517* .274 .376 .495* .219 (.199) (.204) (.211) (.199) (.206) (.212) t4 -.028 .488 .116 -.316 .237 -.261 (.305) (.315) (.313) (.313) (.335) (.317)

-2 Log-likelihood (d.f.) 585.17 556.19 574.24 570.36 551.34 548.48 (9) (11) (11) (10) (12) (12) Pseudo R2 .15 .20 .18 .18 .21 .21

Note: Numbersin parenthesesare standarderrors; number of observations= 42. a See Table 1 for definitions of period variables. p <.05 **p< .01 (two-tailed tests) timization thesis in relation to the pace of however, report mixed results as regards the Congressional action. So events that feature victimization thesis. The strongest finding in either police violence or injuries to demon- this regard concerns events featuring police strators are, in fact, highly predictive of violence: Far from aiding the movement, po- House and Senate roll calls on war-related lice violence is negatively related to pro- measures. The same two models in Table 3, peace outcomes. 7 1 4 AMERICAN SOCIOLOGICAL REVIEW

SIGNALING Table 4. OLS Regression Estimates of the Preferred Model Predicting the of the Public Who Believed In sharp contrast to notions of threat and Percentage Vietnam War Was a Mistake, 1965 to disruption, Lohmann (and others) empha- 1973 size persuasion as the hallmark of success- ful movements. As discussed above, one IndependentVariable Coefficient (S.E.) mechanism of persuasion is Lohmann's -7.363 (5.116) (1993) notion of signaling. To operation- Intercept alize the concept, we speculate, in Hypoth- Protest Events esis 3, that large protests (e.g., ones that in- All events -.230 (.109) volve 10,000 or more demonstrators)consti- Violence by demonstrators 1.429 (.720) tute credible signals of majority support for policy change, and thus should encourage Violence by police .533 (.372) increased and more favorable government Other Covariates action on war-relatedmeasures. Once again, U.S. militarydeaths in Vietnam .621** (.174) we see that the hypothesis is born out with respect to the pace of House and Senate ac- Call-ups in Selective Service .016 (.010) mod- tion, but not its direction. The relevant Conscientiousobjectors .018 (.116) els are Models 5 and 6 in both tables. Large demonstrations bear a rela- Publishedarticles related -1.606 (1.957) strong, positive to Vietnam tionship to the monthly counts of roll calls, but in fact have an equally strong negative Note: Number of observations= 22; F (7, 14) = = = effect on the likelihood of pro-peace out- 6.56; R2 .767; adjusted R2 .650. comes from those votes. So our results offer *p < .05 **p< .01 (two-tailed tests) only partial support for the idea that pro- tests, as moderate expressions of a prefer- ence for change, may shape Congressional shaping favorable outcomes on those votes. action. In contrast, the most extreme, violent, and But this is only half of Lohmann's argu- generally threatening of the protests do ap- ment regarding signaling. The other half pear to increase the likelihood of pro-peace concerns the opposed policy positions that voting, but at the same time they depress forms of extremist action are likely to en- the pace of House and Senate roll calls. courage among elected officials. In contrast to we in Hypothesis 1, speculate Hypothesis PUBLIC OPINION MOBILIZATION 4 that protests featuring either violence by demonstrators or property damage will de- What about our third and final mechanism? crease both the pace of Congressional ac- We sought, in our initial discussion of public tion and the likelihood of pro-peace voting. opinion mobilization, to delineate a mecha- We alreadyknow the results of these analy- nism very different from disruptive protest ses: The more violent or extreme the events, or signaling. Instead of directly shaping the the fewer the roll calls. There is, however, no pace of Congressional action through the use evidence that these same extremist signals of disruptive tactics, or a moderate signal of negatively impact the direction of Congres- the public's preference for policy change, we sional action. Indeed, in Table 3 both of these wondered whether antiwar protestors might proxies are positively related to the likeli- exert an indirect influence on House and Sen- hood of pro-peace outcomes. ate voting through the mediating mechanism Besides contradicting the hypothesized of public opinion mobilization. The hypoth- effect of extremist signals on Congressional esized mechanism involves two linked but action, this latter finding sharpens the inter- clearly sequenced dynamics. In the first, pro- esting contrast between threat and persua- test activity helps to mobilize increased pub- sion that appears to run through the data. lic opposition to the War.In turn, this shift in Persuasive mechanisms (e.g., signaling) ap- public opinion contributes to an increase in pear to have been effective in compelling either Congressional attention or opposition increased Congressional voting, but not in to the war. THE POLITICAL IMPACT OF ANTIWAR PROTESTS 71 5

We begin with the link between antiwar light of Hypothesis 6-is the negative rela- protests and public opinion change. We of- tionship (marginally significant at p < .053) fer, in Hypotheses 5 and 6, two very differ- between lagged protest events and shifts in ent characterizations of this relationship. public opinion. Clearly the growing antiwar Speculating that the perceived "victimiza- sentiment among the general public was not tion" of demonstratorswould generate sym- responsive to the overall pace of movement pathy and support for the movement, we activity. posit in Hypothesis 5 that events featuring There is, however, one category of protest injuries to demonstrators or violence by the event that may be related to public opinion police would be related to growing antiwar trends on the issue. Protests featuring vio- sentiment among the general public. In con- lence by demonstratorsmay be related (mar- trast, we assume, in Hypothesis 6, that it is ginally significant at p < .067) to growing the simple pace of protest events that shapes opposition to the war.8 This finding is con- public opinion on the war. In Table 4 we test sistent with the association reported in Table these two possibilities. 3 between demonstrator violence and the Before we turn to these results, however, outcome of House and Senate voting. How we again call attention to a significant meth- should we interpret these results? The sug- odological lacunae: We are only working gestion is that protests per se did not affect with 22 public opinion data points in this public opinion, but that especially violent phase of the analysis. The best public opin- ones did, perhaps by attracting more public ion time-series concerning the war involves attention and dramatizing the growing soci- a single question (". .. do you think the U.S. etal rift occasioned by the war. We should made a mistake in sending troops to fight in not, however, take this growing public op- Vietnam?") that was asked of the American position to the war as any kind of endorse- public 22 times during our study period. As ment of the movement. As Schuman's a result of the small number of public opin- (1972) research shows, escalating antiwar ion observations we do not have the statisti- activity was related to both an increase in cal leverage to introduce many independent opposition to the war and a public opinion variables into the analysis. In general, the backlash against the movement. limited number of observations recommends So much for the impact of antiwarprotests considerable caution in interpreting the fol- on public opinion. But what of the hypoth- lowing results. esized effect of public opinion on Congres- Now to the data at hand. In Table 4 we re- sional action? Is the growing opposition to gress changes in the percentage of the pub- the war positively related to either the pace lic pronouncing Vietnam to have been a or direction of House and Senate roll calls? "mistake" on a handful of significant inde- The answer, reflected in the results in Tables pendent variables. Not surprisingly,growing 2 and 3, is, once again, mixed. While grow- public opposition to the War appears to have ing public opposition to the war is highly been powerfully shaped by the pace of the predictive of the pace of Congressional ac- war itself-the number of deaths in Vietnam. tion, it is not related (or only weakly so) to So one of our strain measures bears a strong the direction of voting. predictive relationship to shifts in public on the war. In most of our opinion contrast, DISCUSSION event proxies appear to be unrelated, or only weakly related, to these same public opinion In closing, we underscore what we see as the trends. (Failing to attain significance in our most significant implications of the results process of stepwise elimination, they are ex- cluded from the model reportedhere.) Some- 8 The between "eventswith vio- what surprisingly, given the "victimization" relationship this includes our measures of events lence by demonstrators"and "publicopinion" is thesis, at the < .10 level. with either violence or to significant only p But, given police injuries how difficult it is to achieve statisticalsignifi- is demonstrators. There no significant pub- cance with only 22 dataobservations, we are in- lic opinion shift associated with these kinds clined to regardthis result as real and theoreti- of protest events. Even more surprising-in cally meaningful. 7 16 AMERICAN SOCIOLOGICAL REVIEW reported above. To do so, we return to the SIGNALING specific issues with which we led the article. But Lohmann's (1993) provocative concep- tual work on the function of mass REFINING OUR UNDERSTANDING OF signaling movements a more direct link be- MOVEMENT OUTCOMES suggests tween state action and what might be With respect to the general issue of move- thought of as a mobilized, action-oriented ment outcomes, our research supports a proxy for public opinion. Is there evidence marked departurefrom the normal approach of this second persuasive mechanism in our to the topic. We now have a substantial body data?The answer would appearto be no. Not of systematic empirical studies that convinc- only is our operational measure of moderate ingly link movement activity to outcomes of movement signals (e.g., large demonstra- various kinds. What we do not yet have is tions) negatively related to the direction of anything approachinga theory or theories of House and Senate voting, but our proxies for what factors or dynamic mechanisms help extremist action (e.g., events featuring vio- account for these effects. Our focus on dis- lence by demonstrators and property dam- ruptive protest, signaling, and public opin- age) bear a strong positive association with ion mobilization should not be mistaken for pro-peace outcomes. such a theory. We do, however, think that the related to these mechanisms are findings DISRUPTIVE PROTEST suggestive and encouraging of the more gen- eral mechanism-based approach we are ad- In sharp contrast to the two preceding per- vocating here. So much so that we want to suasive mechanisms, some analysts have underscorethe potential significance of each long seen threat/disruption as the sine qua of them. non of successful movements. Seeing merit in this general argument, we nonetheless to between two dif- PUBLIC OPINION MOBILIZATION- sought distinguish very ferent mechanisms that are often to- A NEGLECTED COMPONENT OF lumped under the of MOVEMENT DYNAMICS gether general heading "disrup- tion." The first emphasizes the strategic ad- Given the prominence assigned by political vantage gained by movements through the scientists to public opinion shift as a crucial use of extreme and/or violent forms of pro- mechanism of policy change (and the volu- test. The second view stresses the functional minous empirical evidence consistent with benefits of public protest that sparks vio- this view), it is incumbent on social move- lence against movement activists. ment analysts to pay far more attention to Granting qualified support to the second, the impact (or lack thereof) of movement or "functional victimization" view, large activity on public opinion change. If shifts protests and those featuring police violence in public opinion generally presage changes are highly predictive of increases in war-re- in state policy (Page and Shapiro 1983; lated Congressional roll calls, but surpris- Stimson et al. 1995), then social movements ingly they are just as strongly negatively re- may be able to indirectly shape governmen- lated to pro-peace outcomes. The reverse is tal action by stimulating changes in public true for the most extreme forms of move- opinion. We wondered whether this two-step ment-initiated protest. That is, those demon- mechanism of public opinion mobilization strations that feature violence by demonstra- was operating during the Vietnam War era. tors or property damage depress the pace of Interestingly, only one of the two steps is voting, but substantially increase the likeli- clearly verified in our findings. While grow- hood of pro-peace votes. ing public opposition to the war is related to Whether these results are peculiar to this the pace and valence of Congressional vot- movement or revealing of more general links ing, antiwar protests (with the exception of between these dimensions of disruption and those featuring violence by demonstrators) various outcomes remains to be seen. The do not appear to have been the catalyst for point is, by consistently attending to these the public opinion change. dimensions of protest events, we may be THE POLITICAL IMPACT OF ANTIWAR PROTESTS 717 able to fine-tune our understanding of the generic mechanisms on which these dilem- kinds of mechanisms that shape outcomes mas may turn. across a good many contentious episodes. TOWARD A MORE SYSTEMATIC PERSUASION VERSUS THREAT IN UNDERSTANDING OF THE "WAR AT MOVEMENT DYNAMICS HOME" While interesting in themselves, the results Besides the general theoretical and method- per the specific mechanisms appear to con- ological goals we established at the outset of form to a more general-and somewhat this article, we are also concerned with shed- counter intuitive-pattern. Our measures of ding light on a contentious episode-the extreme threat/disruption (e.g., violence by Vietnam era antiwar movement-that has demonstrators and property damage) are, in been oddly neglected by social-movement fact, positively related to pro-peace voting, scholars. From this perspective, we think our but also to depressed rates of monthly roll findings provide an interesting, systematic calls. In contrast, our two persuasive mecha- confirmation of what many people experi- nisms-signaling and public opinion shift- enced during the peak period of domestic predict increases in roll-call votes, but little contention over the Vietnam War. We con- or no (or even a negative) effect on vote out- clude by relating two of our findings to the comes. Even the results concerning the lived experience of those years. "functional victimization" hypothesis can be First, reviled though it was by many, the interpreted in light of this general pattern. antiwar movement nonetheless exerted a Courting violence by one's opponents can powerful agenda-setting force on the federal be seen as an extreme form of persuasion- government and U.S. society more generally. an effort to curry favor by mobilizing sym- As our results show, antiwar protests-espe- pathy on behalf of the movement. Inter- cially those of a large or injurious nature- preted in this way, these particular results compelled public and Congressional atten- again fit the general pattern. As with our tion. That said, our data also clearly under- other persuasive mechanisms, violence by score the limits of what the movement was police and/or injuries to demonstrators in- able to achieve in relation to U.S. policy in crease the pace of roll-call voting, without Vietnam. Although extreme forms of public improving the chances of pro-peace out- protest may have helped shift public opinion comes. against the war, the protests themselves had This general pattern underscores the contradictory effects on the direction of daunting strategic dilemmas that social House and Senate voting. Certain dimen- movements confront in their efforts to shape sions or forms of protest (e.g., large demon- outcomes and the need of movement ana- strations, violence by police, etc.) appear to lysts to attend much more closely to the dy- decrease the likelihood of pro-peace out- namic relationships among tactics, targets, comes. Finally, in their earlier ground- and the ways in which certain outcomes breaking work, Burstein and Freudenburg (e.g., the successful courting of media atten- (1978) showed that whatever effect antiwar tion through victimization) may preclude protests had was confined to the period prior others (policymaker support for policy to 1970. change). In the case of the U.S. antiwar These results provide statistical confirma- movement, it appears that certain forms of tion of what analysts and activists experi- protest were effective in compelling Con- enced during the period. Drawing on the per- gressional action (e.g., large demonstrations, ceived lessons of the civil rights struggle, those involving police violence, etc.) but not antiwar activists were motivated by an im- in shaping the outcome of that action, while plicit understandingof, and faith in, conven- others (violence by demonstrators) had ex- tional democratic theory. Educate the public actly opposite effects. Only by attending to about the evils of the war and mobilize and the variable patterning of such findings can demonstrate that growing opposition, and we begin to understand the general dilem- eventually policymakers would be per- mas confronting social movements and the suaded to modify their actions and bring the 71 8 AMERICAN SOCIOLOGICAL REVIEW conflict to a close. Or so antiwar activists Developmentof Black Insurgency,1930-1970 believed. As our results confirm, this ex- (rev. ed, University of Chicago Press, 1999); pected responsiveness to popular protest and Freedom Summer (Oxford University Press, which was awarded the 1990 C. public opinion was only partly realized. The 1988), Wright kind of extreme tactics that attracted media Mills Award as well as being a finalist for the American Association's best book attention, and influ- Sociological shaped public opinion, prize for 1991; and of Contention enced action also Dynamics Congressional depressed (Cambridge University Press, 2001) with Sid the overall rate of House and Senate voting. Tarrow and Charles Tilly. Threat/disruption may have shaped certain vote outcomes, but it did so at the cost of a Yang Su is Assistant Professor of Sociology at and generalized backlash in Congress. University of California, Irvine, is also an didn't the of de- affiliate faculty member of the Center for the Why "politics protest" there. His main research in- liver the achieved the civil Study of Democracy gains by, say, terests include collective action and social move- rights movement, whose tactical repertoire ments in both China and the United States. He is also mixed threat and persuasion? Perhaps working on publishing the results from his dis- because the antiwar movement never mobi- sertation project on collective violence during lized anything like the general public sup- the Cultural Revolution in China. He is also in- port and sympathy that the early civil rights volved in a new project (with Doug McAdam) ex- struggle achieved. In turn, we view this gen- ploring the dynamics of church burnings in the eral antipathy to the movement as a by- United States in the late 1990s. product of its perceived lack of commitment to democratic practices and the general poli- REFERENCES tics of persuasion. Indeed, when in the late Amenta, Edwin, Bruce G. Carruthers, and 1960s, black activists abandoned-or many YvonneZylan. 1992. "A Hero for the Aged? appeared to abandon-nonviolence, policy- The TownsendMovement, the Political Me- makers were less consistently responsive to diation Model, and U.S. Old-Age Policy. the movement than they had been in the ear- 1934-1950." American Journal of Sociology lier civil rights phase of the struggle. 98:308-39. This observation motivates us to close by Andrews,Kenneth T. 1997. "TheImpact of So- speculating on the paradoxical nature of cial Movementson the PoliticalProcess: The Civil politics in the United States and the peculiar Rights Movement and Black Election Politics in American strategic challenge it poses to movements. Mississippi." Sociologi- To be movements must cal Review 62:800-19. maximally effective, . 2001. "Social Movements and be while nonetheless Policy disruptive/threatening, Implementation:The MississippiCivil Rights to conform to a democratic appearing poli- Movement and the War on Poverty, 1965- tics of persuasion. Democratic theory not- 1971." American Sociological Review 66:71- withstanding, threat and disruption (and 95. even violence) have been effective means of Arnold, R. Douglas. 1990. The Logic of Congres- mobilizing power in the United States, but sional Action.New Haven,CT: Yale Univer- typically not when practiced by groups per- sity Press. Kenneth J. 1963. Social Choice and Indi- ceived as antidemocratic. The early civil Arrow, movement mastered this vidual Values. New York: John Wiley and rights strategic Sons. of hand. The antiwar movement sleight Astin, Alexander,Helen Astin, Alan Bayer,and never the odd mix of did, producing agenda Ann Bisconti. 1975. The Power of Protest. San setting impact and minimal policy respon- Francisco,CA: Jossey Bass. siveness reflected in our results. Banaszak, Lee A. 1996. WhyMovements Succeed or Fail: Opportunity, Culture, and the Doug McAdam is Professor of Sociology at Struggle for Woman Suffrage. Princeton, NJ: StanfordUniversity and Director of the Center PrincetonUniversity Press. for AdvancedStudy in the BehavioralSciences. Bartels,Larry M. 1991. "ConstituencyOpinion He is the authoror co-authorof eight booksand andCongressional Policy Making: The Reagan morethan 50 articles in the area of political so- Defense Buildup." American Political Science ciology, with a special emphasison the studyof Review 85:457-74. social movementsand revolutions.Among his Berkowitz, William R. 1974. "Socioeconomic best-known works are Political Process and the IndicatorChanges in GhettoRiot Tracts."Ur- THE POLITICAL IMPACT OF ANTIWAR PROTESTS 719

ban Affairs Quarterly 10:69-94. Erikson, Robert S., Gerald Wright, and John P. Boettcher, Thomas D. 1985. Vietnam: The Valor Mclver. 1993. Statehouse Democracy: Public and the Sorrow. Boston, MA and Toronto, Opinion and Policy in the American States. Canada: Little, Brown and Company. Cambridge, MA: Cambridge University Press. Burstein, Paul. 1979. "Public Opinion, Demon- Feagin, Joe R. and Harlan Hahn. 1973. Ghetto stration, and the Passage of Antidiscrimination Riots: The Politics of Violence in American Legislation." Public Opinion Quarterly 43: Cities. New York: Macmillan. 157-72. Finkel, Steven E., Edward N. Muller, and Karl- . 1993. "Explaining State Action and the Dieter Opp. 1989. "Personal Influence, Collec- Expansion of Civil Rights: The Civil Rights tive Rationality, and Mass Political Action." Act of 1964." Research in Political Sociology American Political Science Review 83:885- 6:117-37. 905. . 1998. "Bringing the Public Back In: Fording, Richard C. 1997. "The Conditional Ef- Should Sociologists Consider the Impact of fect of Violence as a Political Tactic: Mass In- Public Opinion on Public Policy?" Social surgency, Welfare Generosity, and Electoral Forces 77:27-62. Context in the American States." American Burstein, Paul and William Freudenburg. 1978. Journal of Political Science 41:1-29. "Changing Public Policy: The Impact of Pub- Franzosi, Roberto. 1987. "The Press as a Source lic Opinion, Antiwar Demonstrations, and War of Socio-Historical Data: Issues in the Meth- Costs on Senate Voting on Vietnam War Mo- odology of Data Collection from Newspapers." tions." American Journal of Sociology 84:99- Historical Methods 20:5-15. 122. Gallup International, Inc. 1965-1973. Gallup Button, James W. 1978. Black Violence. Opinion Index. Princeton, NJ: Gallup Interna- Princeton, NJ: Princeton University Press. tional, Inc. . 1989. Blacks and Social Change. Gamson, William. [1975] 1990. The Strategy of Princeton, NJ: Princeton University Press. Social Protest. 2d rev. ed. Reprint, Belmont, Colby, David C. 1982. "A Test of the Relative CA: Wadsworth. Efficacy of Political Tactics." American Jour- Giugni, Marco. 1998. "Was It Worth the Effort? nal of Political Science 26:741-53. The Outcomes and Consequences of Social Cress, Daniel M. and David A. Snow. 2000. "The Movements." Annual Review of Sociology Outcomes of Homeless Mobilization: The In- 24:371-93 fluence of Organization, Disruption, Political Greene, William H. 1997. Econometric Analysis. Mediation, and Framing."American Journal of 3d ed. Upper Saddle River, NJ: Prentice-Hall. Sociology 105:1063-1104. Jenkins, J. Craig. 1985. The Politics of Insur- Dahl, Robert A. 1956. A Preface to Democratic gency: The Farm Worker Movement in the Theory. Chicago, IL: University of Chicago 1960s. New York: Columbia University Press. Press. Jenkins, J. Craig and Charles Perrow. 1977. "The Danzger, M. Herbert. 1975. "Validating Conflict Insurgency of the Powerless: Farm Workers' Data." American Sociological Review 40:570- Movements (1946-1972)." American Socio- 84. logical Review 42:249-68. Davenport, Christian. 2001. "The Rashomon Ef- Jennings, Edward T. 1979. "Civil Turmoil and fect in the Social Sciences." Department of the Growth of Welfare Rolls: A Comparative Political Science. Unpublished manuscript. State Policy Analysis." Policy Studies Journal University of Maryland. College Park, MD. 7:739-45. DeBenedetti, Charles. 1990. An American Or- Katzenstein, Mary F. 1990. "Feminism with deal: The Antiwar Movement of the Vietnam American Institutions: Unobtrusive Mobiliza- Era. Syracuse, NY: Syracuse University Press. tion the 1980s." Signs 16:27-54. Downs, Anthony. 1957. An Economic Theory of . 1998. Faithful and Fearless: Moving Democracy. New York: Harper and Row. Feminism into the Church and Military. Duyvendak, Jan Willem. 1995. The Power of Princeton, NJ: Princeton University Press. Politics: New Social Movements in France. Kelly, William R. and David Snyder. 1980. "Ra- Boulder, CO: Westview. cial Violence and Socioeconomic Changes Earl, Jennifer. 2000. "Methods, Movements, and among Blacks in the United States." Social Outcomes: Methodological Difficulties in the Forces 58:739-60. Study of Extra-Movement Outcomes." Re- Koopmans, Ruud. 1993. "The Dynamics of Pro- search in Social Movements, Conflicts and test Waves: West Germany, 1965 to 1989." Change 22:3-25. American Sociological Review 58:637-58. Eisinger, Peter K. 1973. "The Conditions of Pro- 1995. Democracy from Below: New So- test Behavior in American Cities." American cial Movements and the Political System in Political Science Review 67:11-28. West Germany. Boulder, CO: Westview. 720 AMERICAN SOCIOLOGICAL REVIEW

Kriesi, Hanspeter, Ruud Koopmans, Jan Willem test Group Success: The Impact of Group Duyvendak, and Marco G. Giugni. 1995. New Characteristics, Social Control, and Context." Social Movements in WesternEurope: A Com- Sociological Focus 14:177-92. parative Analysis. Minneapolis, MN: Univer- Monroe, Alan D. 1998. "Public Opinion and Pub- sity of Minnesota Press. lic Policy." Public Opinion Quarterly 62:6-28. Lieberson, Stanley and Arnold R. Silverman. Mueller, Carol. 1997. "InternationalPress Cov- 1965. "The Precipitants and Underlying Con- erage of East German Protest Events, 1989." ditions of Race Riots." American Sociological American Sociological Review 62:820-32. Review 30:887-98. Nie, Norman H. and Sidney Verba. 1975. "Po- Lipsky, Michael. 1968. "Protest as a Political litical Participation."Pp. 1-74 in Handbook of Resource." American Political Science Review Political Science, vol. 4, Nongovernmental 62:1144-58. Politics, edited by F. I. Greenstein and N. Lohmann, Suzanne. 1993. "A Signaling Model of Polsby. Reading, PA: Addison-Wesley. Informative and Manipulative Political Ac- Oliver, Pamela E. and Daniel J. Myers. 1999. tion." American Political Science Review 87: "How Events Enter the Public Sphere: Con- 319-33. flict, Location, and Sponsorship in Local Long, J. Scott. 1997. Regression Models for Cat- Newspaper Coverage of Public Events." egorical and Limited Dependent Variables. American Journal of Sociology 105:38-87. Thousand Oaks, CA: Sage. Olson, Mancur. 1965. The Logic of Collective McAdam, Doug. 1982. Political Process and the Action. Cambridge, MA: Harvard University Development of Black Insurgency, 1930-1970. Press. Chicago, IL: University of Chicago Press. Olzak, Susan. 1987. "Causes of Ethnic Protest . 1983. "Tactical Innovation and the Pace and Conflict in Urban America, 1877-1889." of Insurgency."American Sociological Review Social Science Research 16:185-210. 48:735-54. . 1989. "Labor Unrest, Immigration, and . 1989. "The Biographical Consequences Ethnic Conflict." American Journal of Sociol- of Activism." American Sociological Review ogy 94:1303-33. 54: 744-60. . 1992. The Dynamics of Ethnic Competi- . 1998. "The Future of Social Move- tion and Conflict. Stanford, CA: Stanford Uni- ments." Pp. 229-45, in From Contention to versity Press. Democracy, edited by M. Giugni, D. McAdam, Page, Benjamin I. and Robert Y. Shapiro. 1983. and C. Tilly. Lanham, MD: Rowman and "Effects of Public Opinion on Policy." Ameri- Littlefield. can Political Science Review 77:175-90. McAdam, Doug, Sidney Tarrow, and Charles Piven, Frances F. and Richard Cloward. 1993. Tilly. 2001. Dynamics of Contention. New Regulating the Poor: The Functions of Public York and London, England: Cambridge Uni- Welfare. New York: Vintage. versity Press. Poole, Keith T. and Howard L. Rosenthal. 1997. McCammon, Holly J., Karen E. Campbell, Ellen Congress: A Political-Economic History of M. Granberg, and Christine Mowery. 2001. Roll Call Voting. New York: Oxford Univer- "How Movements Win: Gendered Opportunity sity Press. Structures and U.S. Women's Suffrage Move- Readers' Guide to Periodical Literature. 1965- ments, 1866 to 1919." American Sociological 1974. New York: H. W. Wilson Co. Review 66:49-70. Rochon, Thomas. 1998. Culture Moves. McCann, Michael W. 1994. Rights at Work.Chi- Princeton, NJ. Princeton University Press. cago, IL: University of Chicago Press. Rosenthal, Howard L. and Keith T. Poole. 1991. McCarthy, John D. and Clark McPhail. 1997. United States Congressional Roll Call Voting "The Institutionalization of Protest." Pp. 83- Records, 1789-1987: Reformatted Data 110 in A Movement Society? Contentious Poli- (MRDF). Pittsburgh, PA: Howard L. tics for a New Century,edited by D. Meyer and Rosenthal and Keith T. Poole, Carnegie S. Tarrow. Boulder, CO: Rowman and Mellon University, Graduate School of Indus- Littlefield. trial Administration [producer]. Ann Arbor, McCarthy, John D., Clark McPhail, and Jackie MI: Inter-university Consortium for Political Smith. 1996. "Images of Protest: Dimensions and Social Research [distributor]. of Selection Bias in Media Coverage of Wash- Schumaker, Paul D. 1975. "Policy Responsive- ington Demonstrations, 1982 and 1991." ness to Protest-Group Demands." Journal of American Sociological Review 61:478-99. Politics 37:488-521. Meyer, David and Sidney Tarrow. 1997. A Move- Schuman, Howard. 1972. "Two Sources of Anti- ment Society? Contentious Politics for a New war Sentiment in America." American Journal Century.Boulder, CO: Rowman and Littlefield. of Sociology 78:513-36. Mirowsky, John and Catherine Ross. 1981. "Pro- Sears, David 0. and John B. McConahay. 1973. THE POLITICAL IMPACT OF ANTIWAR PROTESTS 721

The Politics of Violence: The New Urban Outcomes." Pp. 281-395 in Violence in Blacks and the Watts Riot. Boston, MA: America: Historical and Comparative Per- Houghton Mifflin. spectives, edited by H. D. Graham and T. R. Sen, Amartya K. 1970. Collective Choice and So- Gurr. New York: Bantam. cial Welfare. San Francisco, CA: Holden-Day. Tarrow, Sidney. 1989. Democracy and Disorder: Shorter, Edward and Charles Tilly. 1974. Strikes Protest and Politics in Italy, 1965-1975. Ox- in France, 1830-1968. Cambridge, MA: Cam- ford, England: Oxford University Press. bridge University Press. .1993. "Social Protest and Policy Reform: Snyder, David and William R. Kelly. 1976. "In- May 1968 and the Loi d'Orientation in dustrial Violence in Italy, 1878-1903." Ameri- France." Comparative Political Studies 25: can Journal of Sociology 82:131-62. 579-607. . 1977. "Conflict Intensity, Media Sensi- . 1998. Power in Movement. New York tivity and the Validity of Newspaper Data." and London, England: Cambridge University American Sociological Review 42:105-23. Press. Soule, Sarah, Doug McAdam, John McCarthy, Tilly, Charles, Louise Tilly, and Richard Tilly. and Yang Su. 1999. "Protest Events: Causes or 1975. The Rebellious Century, 1830-1930. Consequences of State Action? The U.S. Cambridge, MA: HarvardUniversity Press. Women's Movement and Federal Congres- Uhrig, Noah and Nella Van Dyke (with Kyra sional Activity." Mobilization 4:239-56. Greene). 1996. Dynamics of Collective Protest Spilerman, Seymour. 1970. "The Causes of Ra- in the U.S., 1960-1980. Manual for microfilm cial Disturbances: A Comparison of Alterna- copying and event coding (coding manual). tive Explanations."American Sociological Re- Department of Sociology, University of view 35:627-49. Stanford, Stanford, CA. Unpublished. .1976. "StructuralCharacteristics of Cit- U.S. Government Printing Office. 1967-1973. ies and the Severity of Racial Disorders." Semi-Annual Report of the Director of Selec- American Sociological Review 41:771-93. tive Service. Selected volumes from July 1967 Steedly, Homer R. and John W. Foley. 1979. through December 1973. Washington, DC: "The Success of Protest Groups: Multivariate U.S. Government Printing Office. Analysis." Social Science Research 8:1-15. Welch, Susan. 1975. "The Impact of Urban Riots Stimson, James A., Michael B. MacKuen, and on Urban Expenditures."American Journal of Robert S. Erikson. 1995. "Dynamic Represen- Political Science 29:741-60. tation." American Political Science Review. Wonnacott, Thomas H. and Ronald J. Wonnacott. 89:543-64. 1990. Introductory Statistics for Business and Taft, Philip and Philip Ross. 1969. "American Economics. (4th ed.). New York: John Wiley Labor Violence: Its Causes, Character, and and Sons.