Political scandals have important implications in democracies, especially if they involve corruption. When politicians are accused of, or caught, enriching themselves at the expense of the public good, cynicism is fostered and trust reduced (Bowler and Karp 2004). If undeterred, repeated corruption scandals threaten to undermine the legitimacy of elections (Caillier Reference Caillier2010; Stockemer, LaMontagne, and Scruggs 2011; Villoria, Van Ryzin, and Lavena 2012). Surprisingly, the evidence from past studies is mixed as to whether politicians are punished more severely for corruption than other types of scandals (Doherty, Dowling, and Miller 2011). To be sure, corruption scandals are on average damaging to politicians (Basinger Reference Basinger2013; Hirano and Snyder Reference Hirano and Snyder2012). Yet, at the congressional level, where the bulk of research occurs, moral scandals tend to matter as much if not more than those about corruption (Brown Reference Brown2006; Peters and Welch Reference Peters and Welch1980; Welch and Hibbing Reference Welch and Hibbing1997).Footnote 1 Whatever the reasons, the data do not unequivocally support the belief that politicians caught in corruption scandals are doomed to lose their seats.Footnote 2
The inconsistent and tempered effects of corruption scandals pose a challenge to one interpretation about how Democrats regained control of Congress after the 2006 midterm elections. Recall that prior to that election multiple Republicans were involved in corruption, including members implicated in the Jack Abramoff lobbying scandal. Members of both parties responded by calling for reforms, suggesting Republican Party leaders were concerned about the possible ramifications (Dancey Reference Dancey2014). Although few Republicans caught up directly in these matters were on the ballot, Democrats campaigned broadly against Republicans’ “culture of corruption.” Media pundits (Bacon, Cox, and Tumulty 2006; Smith 2006) and political strategists afterwards cited this strategy as a significant reason for Democratic success. As reported by Robert Novak, Karl Rove told congressional Republicans that 2006 was the result of corruption (Novak July 28, 2007). Likewise, Best, Ladewig, and Wong (2013) argue that Democrats successfully parleyed the Abramoff scandal into vote shares. In short, voters are thought to have deliberately retaliated against other Republican candidates for the corrupt activities, purported or actual, of a few of its members.
Given the mixed findings in the literature about the importance of corruption to voters, that claim is questionable. It is also deserving of additional scrutiny because few studies suggest collective partisan accountability can arise from individual scandals (but see, Slomczynski and Shabad Reference Slomczynski and Shabad2011), particularly in a two-party system where attachments to party within the public are relatively weak. Therefore, we reexamine this assumption by analyzing exit-polling data from the 2006 elections. These data allow us to compare voter reactions to a moral scandal that we believe had greater potential to influence the outcome. In late September, Mark Foley (R-FL) was reported to have engaged in salacious behavior toward young male pages. As this scandal unfolded Republican Party leaders were accused of failing to intervene despite knowing about Foley’s actions, possibly as a result of electoral considerations (Hulse and Hernandez Reference Hulse and Hernandez2006; Weisman Reference Weisman2006). This last characteristic is the abuse of office, a mediating variable found to increase the relevance of moral scandals (Doherty, Dowling, and Miller 2011).
THE 2006 CONGRESSIONAL MIDTERMS
Unlike past malfeasance that is sporadic and routinely crosses party lines, the Republican Party in 2006 had been disproportionately rocked by multiple scandals. Many of these scandals involved corruption, including the Abramoff lobbying episode.Footnote 3 Notably, several of these resulted not just in criminal investigations, but also in indictments and lengthy jail sentences. Democrats responded by attempting to turn corruption scandals into an electoral liability for all Republicans. In the parlance of the literature on issue ownership, Democrats tried to make Republicans “own” the issue of corruption (Egan Reference Egan2013; Petrocik Reference Petrocik1996; Sides Reference Sides2007). As we know now, Democrats gained 30 House and six Senate seats in the 2006 midterms, capturing the majority in both chambers for the first time in more than a decade.Footnote 4 Media verdicts were swift and nearly unanimous: “it was the dishonesty, stupid” (Bacon, Cox, and Tumulty 2006).Footnote 5
Yet multiple pieces of evidence challenge that interpretation. First, the public seemed more interested in the Foley scandal. No election issue was covered more extensively after Labor Day (Johnson Reference Johnson2006). Conversely, the Abramoff scandal received less media coverage than other scandals between 2000 and 2010 (Romano Reference Romano2014), and the public was relatively uninterested in it. According to a Pew study (Pew 2006), only 38% of the public reported following the news about Abramoff, whereas 80% were following the death of 12 coal miners in West Virginia, and 51% reported following the flooding in California. Second, Republican leaders were apparently alerted to Foley’s behavior several years prior but failed to intervene. Accurate or not, the impression was given that several Republican leaders abused the power of the office in protecting a Republican seat rather than children from a sexual predator (Washington Times 2006).
DATA, DESIGN, AND METHODS
We test the claim that Democrats were successful at making Republicans in general responsible for corruption scandals of individual partisans by analyzing the 2006 National Election Pool General Election Exit Poll. With these data, we also examine how much the Foley scandal mattered to voters.
The exit poll included more than 13,000 completed surveys, although for comparability purposes we eliminated approximately 1,500 preelection phone survey respondents.Footnote 6 One limitation to our data is that the poll used a split ballot design, halving the sample size for our subsequent analyses. More important, the poll asked about corruption only in one version of the survey and about the Foley scandal in the other. Thus, we cannot directly compare the effects of the two types of scandals, nor can we determine whether the Foley scandal contributed to voters’ perceptions about corruption. Despite these issues, approximately 5,000 complete responses are available to analyze in each version of the poll.
We test the claim that Democrats were successful at making Republicans in general responsible for corruption scandals of individual partisans by analyzing the 2006 National Election Pool General Election Exit Poll. With these data, we also examine how much the Foley scandal mattered to voters.
Variables
We analyze respondents’ vote choice for the House race as our dependent variable.Footnote 7 Vote choice is a generic measure of support for the Democratic or Republican candidate in the respondents’ district rather than a specifically named candidate. A vote for the Democrat candidate is coded as a “1”; a vote for the Republican is coded as a “0.”
The theoretically central independent variables measure voters’ attitudes about two types of scandals, corruption and morality (precise question wording and variable coding are provided in the appendix). Unfortunately, reactions to each scandal were measured with a single question, and each type was measured on a different version of the poll. In the first form (version A), voters’ opinion about corruption was measured by asking them how important “corruption and scandals in government” were to their House vote.Footnote 8 On the second form (version B), voters were asked if they approved or disapproved, and how strongly, of the way “Republican leaders in Congress handled Mark Foley and the congressional page scandal.”
To account for the influence of other variables known to affect vote choice, we include measures of respondents’ feelings about the president and their opinions about Iraq, terrorism, the economy, and illegal immigration. Some opinions were measured with different question wording across the two forms, and a few items were not measured altogether in one or the other version. The poll, for example, included two measures of opinions about Iraq and President Bush on each form of the survey, but it only measured opinions about terrorism and immigration on version A, and it only asked for voters’ assessment of the direction of the country and whether they thought local or national conditions were more important to their vote in version B. Finally, we also control for voters’ partisanship and several standard demographic characteristics: sex, race, age, income, education, and religion. While the demographic variables were measured identically across both forms, religion was measured only in version A while education was measured only in version B.
RESULTS
Before more rigorously analyzing respondents’ voting decisions, we start by reviewing the factors they identified as being important to their vote. These data are valuable to reexamine because postelection media analysis highlighted the percentage of voters saying corruption was important to their vote, largely ignoring the fact that the poll asked about six other issues (Lester Reference Lester2006). We present the percentage of voters saying an issue was “very” or “extremely” important to their vote in table 1.Footnote 9
Table 1 Issue Importance and the House Vote
Note: Entries are percentages. All N’s > 5,000.
First, and contrary to the narrative that corruption drove the election, a majority of voters identified every issue except for the Saddam Hussein verdict as being important to their vote. Second, there is substantial variance in the levels of issue importance, ranging from 58% (values) to 83% (economy). Third, while the media correctly reported that a slightly higher percentage of voters said corruption was more important to their vote than the war in Iraq, this fact is misleading because voters were just as likely to say terrorism was as important to their vote as corruption, and the economy was cited more often than any other issue. Altogether, it appears the role of corruption was likely inflated because other issues were at least on par with it in terms of self-assessed importance.
House Vote: Corruption
We now examine the effect of corruption on the House vote, controlling for opinions about other issues and voter demographics.Footnote 10 In addition to constructing a core model that includes (1) all voters, we also present reduced models for (2) Democrats, (3) independents, (4) Republicans, and (5) voters in the three states where corruption scandals were most prominent (California, Ohio, and Pennsylvania) plus Florida, in the chances that the distinction was blurred between corruption and the Foley scandal.Footnote 11 Given past findings about partisans turning a blind eye to their own party’s scandals (Anduiza, Gallego, and Muñoz Reference Anduiza, Gallego and Muñoz2013), it is possible that some voters were more receptive than others to Democrats’ message. The results are presented in table 2.
Table 2 Probit Regression of the Importance of Corruption on the 2006 House Vote (1=Vote for Democrat)
Note: Entries are probit coefficients; **p<.01; *p<.05, two-tailed tests. Models control for sex, age, race, party identification, religion and income. Four key states are California, Florida, Ohio, and Pennsylvania.
The results for model 1 both confirm and challenge prior interpretations of the midterm elections. On the one hand, the negative signs for coefficients “Bush Approval” and “Iraq1” indicate that voters were more likely to choose a Republican House candidate as approval of Bush and the war in Iraq increased. Meanwhile, the positive coefficient for terrorism indicates that voters who thought the Democratic Party would do a better job at protecting the nation from terrorism were more likely to pick a Democratic candidate, and Democrats also benefited from Iraq being more important. This set of results reaffirms the notion that the election was a classic referendum on the incumbent party (Jacobson Reference Jacobson2007). On the other hand, there is no sign that corruption as an issue helped Democrats.
When we examine the results separately for partisans and independents (models 2–4), corruption again fails to have influence on voting decisions. Overall, the results are remarkably similar to those reported for all voters. The same issues were important to Democrats, Republicans, and independents, except that Democrats were also influenced by the economy and immigration, and only independents were affected by the importance they placed on Iraq. It is possible that corruption failed to nationalize the electorate, but that it affected voters in key states, so we reduce the sample to voters from the four states where Republican corruption scandals originated (model 5). Even in these pivotal states, however, the importance of corruption did not affect voters’ choices. Instead, the results again mirror the findings for voters nationwide.
House Vote: Foley Scandal
In this section, we replicate the regression models for investigating the role of corruption, but instead examine voters’ reactions to the Foley scandal. These results are reported in table 3. Unlike corruption, the Foley scandal significantly helped Democrats running for the House. Starting first with all voters (model 1), feelings about the president and opinions about Iraq were important, as expected, but this model provides the first evidence that a scandal directly affected voting behavior. In addition to disapproval over Republicans’ handling of the page scandal, Democrats were also aided by anger toward Bush and those who wanted to cast their congressional vote against the president. We also find the feeling that the war in Iraq had improved the security of the United States (Iraq3) resulted in greater support for Republicans.
Table 3 Probit Regression of the Foley Page Scandal’s Effect on the 2006 House Vote (1=Vote for Democrat)
Note: Entries are probit coefficients; **p<.01; *p<.05, two-tailed tests. Probit models control for sex, age, race, party identification, education, and income.
The analyses are replicated for partisans and independents (models 2–4) and, with one noteworthy exception, generate similar results. The key difference is that Democrats’ vote choice was not affected by the Foley scandal. Disapproval over the Republicans’ handling of it led to significantly greater support for the Democratic candidate among independents and Republicans, however. The substantive effects of independent variables are difficult to interpret from probit coefficients, so we generate changes in the predicted probabilities for the key explanatory variables, which we report in table 4. For example, a respondents’ probability of voting for a Democrat in the House race increases 53% if they said they were casting their ballot as a vote against Bush. Most relevant, respondents who expressed the maximum disapproval over Republicans handling of the Foley scandal were 22% more likely to vote for a Democrat compared to someone who expressed the minimal amount of disapproval. The magnitude of the impact is on par with believing the war in Iraq improves US security, and just slightly less than the variation in anger toward Bush (from the minimum to the maximum).
The key difference is that Democrats’ vote choice was not affected by the Foley scandal. Disapproval over the Republicans’ handling of it led to significantly greater support for the Democratic candidate among independents and Republicans, however.
Table 4 Changes in the Probability of Voting for a Democrat due to the Foley Scandal and Other Issues
Note: **p<.01, *p<.05. Predicted probabilities derived from probit regressions, all other variables held at their means.
CONCLUSIONS
We do not find evidence to support the contention that corruption was central to voting decisions in 2006. Citizens might have said corruption was important to their vote, but this attitude had no independent effect on vote choice after considering the influence of other variables. Instead our analysis indicates that Mark Foley’s immoral behavior and its ensnaring of Republican leadership was important to voters’ preferences, helping Democrats to recapture the House. This finding is important not just for debates about the effects of different kinds of scandals, but also for the causal mechanisms behind incumbent losses. The primary reason scandals are thought to affect elections is through the emergence of stronger challengers, but here we show that voters can make a conscious decision to “throw the bums out.”
It is possible that corruption failed to affect the vote because of measurement issues. Although we cannot dismiss the possibility, there are theoretical reasons to doubt it would alter the conclusion had opinions been measured differently. Besides previously identified reasons, we would add that if corruption is complicated (and further uninteresting) to voters, as the Abramoff scandal appeared to be, it is less likely to affect voters’ behavior than matters that are easier to understand. Furthermore, perceptions of personal responsibility are often absent for corruption scandals (Alford et al. 1994). While individual members of Congress might get caught committing specific acts of corruption, the public views the problem as endemic to the political system. The Pew poll we cited earlier, for example, found that 80% of respondents believed it was common for lobbyists to bribe members of Congress. If corruption is perceived as institutional, then neither party can easily lay claim to more ethical behavior.
Our positive results for the Foley scandal are novel, given the limited research into collective sanctions for partisan scandals in the United States. Normally a scandal like this is contained to the individual district in which it occurred. The difference here, we believe, is that evidence surfaced to implicate a substantial number of party members of sheltering Mark Foley for political gain. When others facilitate an unambiguously immoral behavior for political reasons, and all those involved are from the same party, a matter that previously seemed quite personal and easily contained can spread with significant negative electoral repercussions.
As Doherty, Dowling, and Miller (2011) argue, when moral scandals are coupled with abuses of office, they can become powerful events. Previous studies, however, had not demonstrated that this combination could generate collective sanctions for other members of the same party. Indeed, the assumption seemed to be that only corruption scandals were capable of doing this. Best, Ladewig, and Wong (2013), for example, argue that the Abramoff corruption scandal had the same effect we attribute to the Foley scandal. We believe methodological differences help explain our incongruent conclusions. Among them, in our study, we use attitudinal measures for voters’ reactions to the different types of scandals. Best, Ladewig, and Wong (2013) use indirect measures for corruption, the amount of news stories covering the Abramoff scandal, and the amount of campaign contributions legislators had received from Abramoff. They do not estimate Foley’s impact. While both studies find parties might be vulnerable to valence issues involving scandals, future work is required to clarify the necessary conditions and the kinds of scandals that are capable of generating this effect.
APPENDIX: Variable Coding
INDEPENDENT VARIABLES
Bush Approval: “Do you approve or disapprove of the way George Bush is handling his job as president?” (1–4; strongly approve = 4).
Against Bush: “Was one reason for your vote for Congress today to express support for George Bush, George Bush was not a factor, to express opposition to George Bush?” (1–3; opposition = 3)
Bush Angry: “Which comes closest to your feelings about the Bush administration: enthusiastic, satisfied, but not enthusiastic, dissatisfied, but not angry, or angry?” (1–4; angry = 4)
Economy1: “Do you think the condition of the nation’s economy is excellent, good, not so good, poor?” (1–4; poor = 4).
Economy2: “Compared to two years ago, is your family’s financial situation better, about the same, worse?” (1-3; worse = 3).
Economy3: “Which best describes your family’s financial situation? You feel as if you are getting ahead financially, have just enough money to maintain your standard of living, or are falling behind lately?” (1–3; falling behind = 3)
Iraq1: “How do you feel about the U.S. war in Iraq?” (1-4; strongly approve = 4)
Iraq2: “What should the U.S. do in Iraq now? Send more troops, maintain the same number of troops, withdraw some troops, withdraw all of the troops?” (1–4; withdraw all of the troops = 4)
Iraq3: “Do you think the war in Iraq has improved the long term security of the United States? (1–2; yes = 2)
Immigration: “Should most illegal immigrants working in the United States be offered a chance to apply for legal status or deported to the country they come from?” (1–2; deported = 2).
Terrorism: “Which party do you think would make the country safer from terror? Only the Democratic Party, only the Republican Party, both would, neither would.” (1–3; Republican Party =1, Both/Neither = 2, Democratic Party = 3).
Wrong Track: “Do you think things in this country today are generally going in the right direction or seriously off on the wrong track?” (1–2; wrong track = 2)
Local/National: “In your vote for U.S. House, which mattered more to you: local issues or national issues?” (1–2; national = 2)
Importance/Corruption: “In your vote for U.S. House, how important was [the war in Iraq/terrorism/the economy/illegal immigration/values such as same-sex marriages or abortion/corruption and scandals in government]?” (1–4; extremely important = 4).
Foley: How do you feel about the way Republican leaders in Congress handled Mark Foley and the congressional page scandal? (1–4; strongly disapprove = 4)
CONTROL VARIABLES
Sex: 0-1; female = 1.
Age: categorical (1-9; 9 = 75 or over).
Race: 0-1; white = 1.
Income: categorical (1-8; $200,000 or more).
Religion: 0-1; born-again = 1.
Education: categorical (0-5; postgraduate study = 5).
Party Identification: (1-3; Republican = 3).
