The Impact of Campaign Attacks on Campaign Interest and Participation, Papers of Social Work

Download The Impact of Campaign Attacks on Campaign Interest and Participation and more Papers Social Work in PDF only on Docsity! HOW DOES IT ALL 'TURNOUT'? EXPOSURE TO ATTACK ADVERTISING, CAMPAIGN INTEREST, AND PARTICIPATION IN AMERICAN PRESIDENTIAL ELECTIONS (Chapter 3 in Campaign Reform: Insights and Evidence, Larry M. Bartels and Lynn Vavreck, eds., Univ. of Michigan Press) Lynn Vavreck Department of Government Dartmouth College ABSTRACT Do attack advertisements cause people to lose interest in campaigns and stay home on Election Day? This is the question addressed in this chapter through the use of an endogenous ordered probit model, which allows attention to the campaign to affect interest in the campaign and turnout, but also allows interest and likelihood of voting to influence attention. I test the hypotheses, popular among reformers, that paying attention to increasingly attack based presidential campaigns has negative effects on campaign interest and electoral participation. I do not find this to be the case. While it is true that campaigns use more attack now than ever before, this does not translate directly into a decline in voter interest in campaigns or abstention on election day. In fact, in some years, attending to attack-driven campaigns increases interest and turnout. If reformers want to outlaw the practices of attack campaigning they must come up with reasons other than low interest and participation levels. I thank Doug Rivers for helping me to conceptualize this problem and Jeffrey B. Lewis for writing the program that generated the endogenous ordered probit FIML estimates. I am grateful to Larry Bartels, upon whose grant this work was supported, and who provided a productive and provocative place for me to think about campaign reform. I am thankful to the Pew Charitable Trusts for their support of this project and many others like it that aim to restore honor to politics. 2 Introduction The keystone of American democracy is the political campaign – a critical link through which potential governors communicate with citizens about problems, solutions, and basic ideologies. Candidates sell their ideas, their histories, and even themselves while voters listen, evaluate, and eventually cast ballots. It is a system in symbiotic balance. Voters need information, but not too much (Popkin 1990); and candidates want to provide information, but only specific kinds. The balance, however, may be shifting. It appears that citizens have lost faith in government and their participation in elections continues to decline. Voters seem to want something more from candidates – but what? In this chapter I explore whether and why Americans are “tuning out” presidential campaigns and more importantly, if the campaigns themselves have anything to do with it. Throughout our political history citizens have lauded the clever wit of candidates. Who can resist smiling at Blaine’s slogan highlighting Cleveland’s illegitimate child during the 1884 campaign, “Ma, Ma, where’s my Pa?” To which Cleveland replied once elected, “Gone to the White House, ha ha ha!” Invective and vituperative strategy routinely amuse voters – and may even draw them into the process. The intemperance and revelry of campaigns was so burdensome that in 1860, perhaps the most important election in American history, Abraham Lincoln actually said very little about his ideas and policies, for fear his words would be severely distorted by his opposition. Instead, he and the Republican Party organized “Wide Awakes” – torchlight parades in which young men six-feet four inches and taller marched alongside rail-splitters’ battalions in support of Lincoln (Boller 1996). Today, many people speculate that these types of campaigns 5 Preserving the Public Good Policy reforms are aimed at preserving or creating a public good. In the case of campaign reform, the public good that is being preserved can be thought of as an informed and engaged electorate. This electorate is by no means “perfect” – the public good as it exists, or existed in the early parts of this century, is simply an electorate in which many but not all of the eligible voters take the time to learn about candidates and vote in elections. Specifically, those who do participate are those who care about whether their preferred policies are enacted into legislation. Those who do not participate do so for any number of various reasons. Perhaps they have four kids and no spare time to learn about politics, or they recently moved to a new home and did not register to vote. Whatever their reasons for being disinterested in campaigns and not participating in elections their absence from the system is either an irregularity (they usually vote but didn’t this time due to some circumstance independent of the campaign itself) or it is habitual (they never vote). The informed and engaged electorate survives despite these types of non-participants. What should worry scholars of campaigns and political reformers is when citizens who are usually interested in politics and vote in elections stop doing so because the system itself has spurned them. This changes the modestly sized informed and engaged electorate into a disgruntled and pessimistic electorate, in which few citizens enjoy civic duties and only a small percentage actually participate. When citizens who care about seeing their preferences on policies passed into law don’t participate because the system discourages them from doing so a “public bad” is created. 6 Some scholars believe the preponderance of attack-based campaigning in contemporary campaigns is the catalyst that has started the shift toward a public bad (Ansolabehere and Iyengar 1995; Patterson 1993; Campaign Discourse Mapping Project 1996; Lau et al. 1995). The story, however, may not be that simple. While it appears true that public satisfaction with campaigns and elections has been declining steadily over four decades, it is also true that people know more now about candidates’ issue positions than they ever have (Bartels 2000). Some scholars claim that attack or negative campaign ads have actually increased the amount of issue information available to the electorate, have disseminated that information clearly to voters, and have generated concern about election outcomes (Mayer 1996; Geer and Finkel 1997; Geer 2000). Do campaign attacks destroy interest and participation or provide clear and easy information to voters about candidates? Do attack antics deter from the public good, leave it unaffected, or enrich it? To get a handle on answering this question one must decide on what criteria the public good is evaluated. Bartels (2000) argues that we can evaluate campaigns on two fronts – how they affect public attitudes towards the electoral process, and how they affect voter learning about candidates during campaigns. Attack advertising may negatively affect people’s attitudes about the electoral process (making it bad for voters since it gets in the way of their instrumental policy goals), but it may actually increase the level of voter learning about candidates’ issue positions (making it good for candidates who want to efficiently and effectively distinguish themselves from their opponents). I turn now to sorting out which, if any, of these scenarios is at work. Does Attack Disengage and Demobilize? 7 In what is perhaps the most elaborate experimental investigation of this relationship, Ansolabehere and Iyengar (1995) find that attack advertising demobilizes the electorate, particularly disarming those who have weak partisan affiliations. As in any experimental setting, these scholars have the ability to control what the subjects see and how much attention the subjects pay to what is being shown (in essence, they “force” the subject to pay attention). This experimental set-up has strong internal validity, giving many insights into the kinds of psychological effects attack advertisements may have, given the controlled level of attention to the advertisement. But, this design cannot shed any light on how attack advertising might affect the levels of attention people pay to campaigns (attack ads may actually make people pay less attention or more). Nor can this set up assess the same dynamic and endogenous relationship among attack advertising and participation. If we care about reforming attack-based electioneering, we must discover whether it creates interest in campaigns or destroys it; and whether it encourages people to pay attention to campaigns or makes them tune-out. In order to ascertain this, both variables must be allowed to change: attention to attack advertisements and interest (or participation) must vary simultaneously – since the two are related to one another in a system, not unlike supply and demand. In a subsequent survey-based study Finkel and Geer (1998) found counter results about the relationship of campaign negativity to participation. Their results indicate no relationship between declining turnout and campaign tone. Additionally, they find that although campaign tone is associated with lower levels of efficacy, negative campaign tone is also associated with greater concern about the outcome of the election. Again, however, Finkel and Geer have investigated the same problem that Ansolabehere and his 10 year indicates exposure to these attack strategies. But just how much attack-based campaigning goes on in American presidential elections? /Figure 1 about here./ In Figure 1, I present the percentages of negative campaign advertisements over six presidential election years (Geer 1998) and the percentage of negative news coverage over the same period (Patterson 1993). As is easily observed, American presidential campaigns have become more negative over the course of this period, culminating in the 1996 election, in which almost half of all campaign appeals were negative in tone.4 Additionally, the tone of news coverage of presidential campaigns has also declined steadily over this period.5 The first link in reformers’ chain has been substantiated – campaigns are more attack oriented today than they were 20 years ago. Second Element: Have Levels of Interest and Participation Changed? In Figure 2, I present data that bear on the questions of campaign interest and voter participation. The survey data were collected by the National Election Study. The turnout data is the aggregate recorded vote level for each election. /Figure 2 about here./ In Figure 2, I present survey results originally offered in Bartels (2000). These data are the average levels of campaign interest for a typical survey respondent (a 45- year-old white female with a high school diploma and median family income). It appears from these data that interest in campaigns has declined gradually over time, with the exception of the 1960 and 1992 campaigns. 6 In 1996, campaign interest sank to its all- time low, just under 40 percent. Simultaneously, participation dropped to an all-time low. 11 As interest deteriorates, so does participation. The correlation of these two time trends is .86. The second element of the reformers’ story is confirmed. Campaign interest and electoral participation are declining over time, at the same time that campaign negativity is increasing. It is easy to understand why scholars and reformers are anxious to blame the behavior of candidates in campaigns for the disengagement and demobilization of the electorate. But a more rigorous test of this claim can be done. Specifying A Systems Model As I said earlier, a unique feature of the relationship between campaign interest and exposure to attack-based campaigning is that both variables affect one another. Essentially, both are dependent variables, even though I want to use each of them as an independent variable to explain the other. Estimating endogenous relationships such as the one between exposure to attacks and campaign interest (or exposure and participation) requires special methodological attention. In simultaneous equation modeling there is more than one equation to estimate in each system; there is one equation for each endogenous variable. In system 1, I estimate one equation for campaign interest and another equation for campaign attention (which is serving as a measure of exposure to campaign attacks). Then for system 2, I estimate one equation for campaign attention and one equation for participation. In this type of work, one cannot estimate the parameters of a single model without taking into account the information provided by the other equations in the system. To ignore the simultaneity by simply using Ordinary Least Squares would result in biased and inconsistent estimates; 12 that is, as the sample size increased indefinitely the estimates would not converge to their true population value. Therefore, some other method should be used. As a solution to this problem I use a full information maximum likelihood (FIML) model.7 This approach allows me to estimate all of the equations in the system at the same time, taking full account of the endogeneity.8 Although in this case FIML is the best possible choice among estimators, these models are not without drawbacks.9 Data, Variables, Models The survey data for these analyses is provided by the American National Election Study (NES). The NES surveys a random sample of the American population before and after every presidential election. I draw here on their efforts from 1976 to 1996.10 I begin with System 1 – the simultaneous relationship between campaign attention, (exposure to campaign attacks) and campaign interest. Where “Traits” include categorical variables for gender, race, education, and partisan strength, along with an ordinal variable constructed from the log of income. These variables are believed to influence campaign interest and attention in typical sociological and psychological ways (Campbell et al 1960; Lazarsfeld et al 1944; Rosenstone and Wolfinger 1980). The variables that uniquely identify the attention equation are “Turnoutt-1”, which is a measure of whether the respondent voted in the last presidential election, and “Reside”, which is the number of years a respondent has lived 1431211 ATTENTION ResideTurnoutTraitsInterest εββββα +++++= −t 1543211 INTERESTUrban TelevisionNewspaperTraitsAttention υγγγγγα ++++++= System 1 15 Particularly interesting is 1988 in which the percentage of negative appeals in advertisements rose to almost 50 percent. In this highly negative campaign year paying attention to the campaign in the media increases a voter’s likelihood of being interested in the campaign by the highest level on record. There does not appear to be a negative relationship between exposure to campaign attacks and campaign interest. In fact, when there is any relationship at all between these variables it is a positive relationship. Campaign attacks seem to be increasing the level of voter interest in campaigns or leaving it unchanged. One thing can be said with considerable certainty, there is no evidence that attacks are causing voters to lose interest in presidential campaigns. Does Being Interested in Campaigns cause People to Pay More Attention? Now I present data on the second half of System 1 – the effect of being interested in the campaign on the level of attention one pays to the campaign in the media. In the previous section I showed that paying attention leads to either no change in interest or to an increase in interest. In those years when interest is increased what then happens to the level of attention? In the second half of Table 1 I present these data. Once again, controls and instrumental variables behave mostly as anticipated. Being older, male, educated, white, and a high wage earner increases the probability of paying attention to campaigns in the media. The number of days a respondent watches television news is highly and positively related to the amount of attention he or she pays to the campaign in the media. The number of days a respondent reads a daily newspaper also shows a positive, albeit more modest, relationship to campaign attention. In only one year (1996) did the size of the respondent’s community affect his or her level of attention. 16 In terms of campaign interest – once again, interest and attention are positively related. In four of the six cases in the analysis as people became more interested in the campaign they paid more attention to the campaign in the media. In the two remaining years no significant relationship was found in either direction. In Figure 4, I graphically present these results. /Figure 4 about here./ In 1980 through 1992 interested citizens paid more attention to the campaign. These years show significant increases in negativity, and as I mentioned earlier, 1988 and 1992 exhibit particularly large increases in attack. Despite this growth in distasteful campaigning, people who were interested in campaigns increased their attention levels. This is another significant finding: attack based campaigning does not seem to have a negative relationship with campaign attention. That is, campaign attacks are not causing citizens to ignore campaigns. In fact, it appears that the opposite is true. Here is where evaluating the relationship, as a system, becomes important. This analysis, which allows for endogeneity between campaign interest and attention, suggests that as people pay attention to the campaign, even and especially in highly attack oriented campaign years, they become more interested in the campaign. As people become more interested in the campaign, they pay more attention to it in the media. This is represented by the positive and significant coefficients on campaign interest and campaign attention in the equations from System 1. These variables are related to one another positively, not negatively as many reformers believe. Results: System 2 17 Does Attack Campaigning Lead to Lower Participation on Election Day? It is possible that even as negative campaigns attract people’s interest they discourage them from participating in elections. A person might be curious about the campaign, attracted to the colorful antics, and still be left with a feeling of lowered political efficacy – a feeling that their vote cannot affect the system or the outcome in any way. Negative campaigns may be interesting to voters, but not inspiring – in short, they may, despite their interest to voters, cause voters to stay away from the polls. In the first half of Table 2, I present data on this possibility. /Table 2 about here./ The results in Table 2 once again confirm that being older, white, female, educated, and a high wage earner positively affect the probability of voting in a presidential election. Furthermore, having voted in the previous presidential election makes it more likely that a person will vote in the current election; as does having lived at an address for a long time. Even being a strong partisan makes turnout more likely. Paying attention to the campaign increases the likelihood of voting in four of the six years under analysis. Even in 1996, when almost half of all campaign appeals were negative, people who paid attention to these appeals were more likely to vote than not. In Figure 5 I present this relationship graphically. /Figure 5 about here./ Much like the relationship between campaign attention and interest, the relationship between campaign attention and turnout is positive. Particularly interesting is 1992 – a year when negative appeals in campaigns rose by almost 10 percent. In this year, paying attention to the campaign made people more likely to vote in the November 20 the electoral process. There may, however, be other possible criteria for evaluating campaigns that would better showcase the poor effects of negativity and attack. One possible criterion for the evaluation of campaigns is how they affect system legitimacy or support (Buchanan 1998). Broader than just attitudes toward the campaigns and the current election, this guideline illuminates people's confidence in their entire system of governance. As political scholars and reformers we may care about these types of beliefs even though they do not appear to have any behavioral manifestations at the current time. We should care because one day the level of support for the system might sink to such a profound low that people begin abstaining from elections on these grounds. If campaign negativity were driving feelings of loss of legitimacy it would also be to blame for this eventual demobilization. A second possible standard for judging campaigns might be the effect they have on politicians' credibility once elected. In a recent survey conducted by the Project on Campaign Conduct at the Institute for Global Ethics, 84 percent of respondents report they believe at least some candidates use outright lies during a campaign; 46 percent believe most candidates do. Furthermore, in a separate survey released by the Pew Research Center, 84 percent of political consultants said they thought it was perfectly ethical to use factually true statements out of context to get a candidate elected. People are losing respect for politicians and this attitude is likely a direct result of the behavior of candidates (and consultants) in their campaigns. In the Campaign Conduct survey, 78 percent of respondents suggest that negative, attack-oriented campaigning produces leaders who are less ethical and less trustworthy. Even more of the sample said that they lose faith in an elected official when he or she 21 lies. Citizens may not be voicing this disrespect with their feet (abstaining from elections or punishing candidates who behave poorly) but the fact remains that politicians are not held in high regard. The larger problem associated with this lack of respect for politicians is one of candidate recruitment. As respect for elected officials goes down the number of quality challengers for political positions also declines. In the survey of campaign consultants funded by the Pew Chritable Trusts, 42 percent of consultants claim that the quality of candidates for congressional races is "falling," another 48 percent rate the overall quality of congressional candidates as only fair or poor. This is certainly a public bad about which scholars and reformers should care. What can be done to curb the distasteful and misleading practices of some campaigners and at the same time restore respect to political leaders? Some of the reforms proposed to combat negative campaigning because it was believed to disengage and demobilize may be appropriate reforms toward this end. The ad watch is a reform that was used extensively in 1992 and less in 1996 to discourage candidates from using misleading images and statements in their ads. Essentially, an independent organization (usually a media outlet such as CNN or The Washington Post) fact-checks each frame of a political advertisement for its “truth.” When done appropriately (Jamieson 1997) these ad watches can discourage candidates from running questionable ads because they do not want to bear the cost of being ad watched and found to be misleading voters. The ad watch works because it affects the incentives of candidates. Candidates behave better because of the ad watch. 22 Another reform that may go a long way toward restoring the respect of elected leaders is free television advertising time with format restrictions. If the government pays for a certain amount of television time for candidates it could legally apply restrictions on what candidates who accept the free time do during their spots. One idea is that the candidates must appear in the ads themselves, with no voiceovers or music – just the candidate talking to the camera. This talking-head format was used in 1996 on several networks that provided free TV time to candidates. Although not many voters recall seeing an ad like this there is no evidence from network ratings that people changed the channel or turned off the television when these talking-head spots came on (Taylor 1996; Free Air Time and Campaign Reform 1997). Further, there is evidence that indicates when candidates have to speak about their campaigns themselves they do not do so in negative or attack oriented ways, but instead use positive promotional styles and substantiate claims with evidence (Adasiewicz 1997). Finally, the most promising of potential reforms being put forth in this election cycle is the Campaign Code of Conduct effort promoted by the Institute for Global Ethics and funded by the Pew Charitable Trusts. In two states during the fall 1998 elections, Ohio and Washington, candidates in each race were asked to come together to draft and sign a set of pledges that structured the acceptable behavior in their particular election. The Institute provided a framework for the activity of drafting the Code of Campaign Conduct, but candidates in the races wrote the actual document. The code includes things like an agreement to at least one public debate, a promise to refrain from demeaning references and demeaning visual images of each other, a pledge to disclose campaign finances on the Internet, and an agreement not to use or allow to be used personal attacks, 25 A Causal Analysis” American Journal of Political Science 27: 284-293. Campbell, J.E. 1992. “Forecasting the Presidential Elections in the States,” American Journal of Political Science 36: 386-407. Campbell, J.E. and K.A. Wink. 1990. “Trial-Heat Forecasts of the Presidential Vote.” American Politics Quarterly 18:251-269. Carter, Jimmy. 1978. The Presidential Campaign 1976. Washington D.C.: The United States Printing Office. Cohen, D. Undefeated: The Life of Hubert H. Humphrey. Minneapolis. Downs, A. 1957. An Economic Theory of Democracy. New York: Harper-Row. Enelow, J.M. and M.J. Hinich. 1984. The Spatial Theory of Voting. New York: Cambridge University Press. Fair, R.C. 1978. “The Effect of Economic Events on Votes for President.” Review of Economics and Statistics 60:322-325. Finkel, S.E. 1993. “Re-Examining the ‘Minimal Effects’ Model in Recent Presidential Campaigns.” Journal of Politics 55(1):1-21. Fiorina, M.P. 1981. Retrospective Voting in American National Elections. New Haven: Yale University Press. Franklin, C.H. 1991. “Eschewing Obfuscation? Campaigns and the Perceptions of U.S. Senate Incumbents” American Political Science Review 85: 1193-1214. Geer, J.G. 1990. Advertising Videotape aired in POS 332, Arizona State University. Geer, J.G. 1994. “Campaign Advertisements in Presidential Elections: A Look at some Evidence,” unpublished manuscript. 26 Gelman, A. and Gary King. 1993. “Why are American Presidential Elections so Predictable when Polls are so Variable?” British Journal of Political Science 23:409-51. Graber, D.A. 1988. Processing the News. White Plains, New York: Longman, Inc. Hibbs, D.A. 1982. “President Reagan’s Mandate from the 1980 Elections: A Shift to the Right?” American Politics Quarterly 10: 387-420. Heckman, J.J. 1978. “Dummy Endogenous Variables in a Simultaneous Equation System” Econometrica 46: 931-960. Hetherington, M.J. 1996. “The Media’s Role in Forming Voters’ National Economic Evaluations in 1992,” American Journal of Political Science 40:372. Holbrook, Thomas M. 1994. "Campaigns, National Conditions, and U.S. Presidential Elections" American Journal of Political Science 38 (4): 25-46. Holbrook, Thomas M. 1996. Do Campaigns Matter? Thousand Oaks: Sage. Iyengar S. and D. Kinder. 1987. News That Matters: Television and American Opinion. Chicago: University of Chicago Press. Key, V.O., Jr. 1966. The Responsible Electorate. New York: Vintage Books. Lazarsfeld, P.F., B.R. Berelson, and H. Gaudet. 1948. The People’s Choice: How the Voter Makes up his Mind in a Presidential Campaign, second edition. New York: Columbia University Press. Lewis-Beck, M.S. and T. Rice. 1992. Foercasting Elections. Washington DC: CQ Press. Lorant, S. 1968. The Glorious Burden: The Amerian Presidency. New York: Knopf. 27 Kiewiet, Roderick R. and Douglas Rivers. 1984. "A Retrospective on Retrospective Voting". Political Behavior 6 (4): 369-393. Kinder, D., G. Adams, et al. 1989. “Economics and Politics in the 1984 Presidential Election,” American Political Science Review 33: 491-515. Kramer, Gerald. 1971. "Short-term Fluctuations in U.S. Voting Behavior, 1896-1964". American Political Science Review 65 (1): 131-143. Kreps, D.M. 1990. A Course in Microeconomic Theory. Princeton: Princeton University Press. Markus, Gregory. 1988. "The Impact of Personal and National Economic Conditions on the Presidential Vote: A Pooled Cross-Sectional Analysis". American Journal of Political Science 32 (1): 137-154. Page, B.I. 1976. “The Theory of Political Ambiguity,” American Political Science Review 70:742. Page, B.I. 1978. Choices and Echoes in Presidential Elections. Chicago: University of Chicago Press. Page, B.I. and R.A. Brody. 1972. “Policy Voting and the Electoral Process: The Vietnam War Issuem” American Political Science Review 66:979. Parmet, H.S. 1972. Eisenhower and the American Crusade. New York: MacMillan. Patterson, T.E. 1993. Out of Order. New York: Knopf. Patterson, Thomas E. 1980. The Mass Media Election: How Americans Choose their President. New York: Praeger. Patterson, Thomas E. and Robert D. McClure. 1976. The Unseeing Eye. New York: Putnam. 30 6 These data on campaign interest suggest that other institutional factors such as the number of candidates, the amount of news coverage of the campaign, the types of issues debated, the experience-level of the candidates, or the quality of the campaigners might affect people’s interest levels in campaigns. For example, the 1960 and 1992 elections are years in which the “challenging” or “insurgent” party candidates ran highly effective campaigns (Vavreck 1997). It is not surprising that the clever campaigns of John F. Kennedy and Bill Clinton generated interest among voters (as compared to say the not so effective insurgent party campaigns of Michael Dukakis and George McGovern). 7 Since the systems I wish to estimate involve ordered categorical dependent variables FIML was the only solution that could itself generate consistent and efficient results. Limited information techniques, such as a two-stage ordered probit model, would not have corrected for errors in the variance-covariance matrix, leaving the resulting estimates inefficient. Full details regarding the assumptions and restrictions used to generate the FIML estimates and Monte Carlo results are available from the author. The Endogenous Ordered Probit FIML Estimation Program is available upon request and was written by Jeffrey B. Lewis, Department of Politics, Princeton University. 8 For a detailed technical discussion of FIML models see Greene (1993, p. 612). For a simple discussion of the method see Christ (1966, p. 395). 9 Mainly, the disadvantages are computational and presentational. The burden of estimating a FIML model is slightly greater than that of a limited information model because the investigator must typically program the estimation routine. Also, because these models generate solutions that are highly non-linear in the parameters, the coefficients are not easily interpreted. To best display the substance of the results, I rely 31 on graphical presentations of the effects, although I present the complete results in tabular form as well. The non-linearity in the parameters also makes it difficult to interpret compound variables such as interactions. So, while I agree with one reviewer that it would be interesting to pool these data and do the analysis on all years combined, computationally this would be complex. 10 Prior to 1976 the NES did not ask respondents about the levels of attention they paid to the campaign in the media. 11 For a detailed yet intuitive explanation of the tradeoffs and difficulties in selecting instrumental variables see Bartels (1991).