«A Dissertation Submitted to the Graduate Faculty of the Louisiana State University and Agricultural and Mechanical College in partial fulfillment of ...»
Ironically, the results presented in this chapter suggest that citizens tend to exact stiff punishments on officials accused of financial fraud while forgiving officials of extramarital affairs. Citizens treat scandal severity in exactly the opposite manner than do journalists.
Whereas the press obsesses over the private sexual liaisons of public officials, citizens appear more cautious about officials who take their constituents tax dollars. This pattern of citizen judgment suggests that democracy may benefit from increased media attention to financial misconduct. Assuming that citizens receive news about financial malfeasance, it appears that citizens will use such information to punish officials accused of severe breaches of democratic, legal and ethical norms. Yet financial scandals tend to receive little media scrutiny, leaving citizens unaware, or marginally aware of their representatives’ financial misconduct.
Political scandals have routinely plagued the United States Congress (Farquhar, 2003).
During the past decade, intense public and media scrutiny has resulted in both major political parties losing Congressional seats in the aftermath of a scandal (Basinger, 2013; Marion, 2010;
Praino et al., 2013). To date, several studies have examined factors that affect an official’s survivability after scandal accusations are made, focusing primarily on the impact of publicity on post-scandal electoral survivability. For instance, Herrick (2000) found that the frequency of Washington Post stories about a House scandal related negatively to House member electoral security. Similarly, Shea (1999) found that frequency of negative local news stories adversely impacted a congressman’s electoral “survivability,” or the ability to avoid resignation and win reelection. In a more recent analysis, Ulbig and Miller (2012) found that news coverage of gubernatorial scandal related to declines in public support for the accused. Examining presidential scandals, Entman (2012) found a “cascading” effect of scandal publicity involving a reinforcing cycle of opposition party claims, press coverage, and public response. During election periods, the decision for the opposition to pursue scandal may have particularly damaging effects on the incumbent official’s electoral prospects (Fogarty, 2013).
Previous research has also examined the impact of scandal media coverage and other, personal and institutional factors on electoral outcomes in subsequent election cycles. For instance, examining the House banking scandal of 1992, Jacobson and Dimock (1994) found that the members implicated suffered at the ballot box. As the authors put it, “the House banking scandal was the major reason for the unusually high turnover of House seats in 1992. It contributed significantly to exit from the House by all routes: retirement, defeat in the primary
conclusions about the House banking scandal and other types of financial and sexual misconduct (Ahuja, Beavers, Berreau, Dodson, Hibbing, Hourigan, Showalter, & Walz, 1994; Basinger, 2013; Groseclose & Krehbiel, 1994; Stewart, 1994; Welch & Hibbing, 1997). Both financial and sexual scandals have real electoral consequences, affecting public opinion, the outcome of elections and vote share in subsequent election cycles (Brown, 2006; Peters & Welch, 1980).
There is reason to believe, however, that the public does not always get to decide the electoral fortunes of the accused. The anecdote related above regarding former Representative Anthony Weiner provides us with evidence that during sex scandals intense media scrutiny compels party leaders to shed the member in question, forcing them to resign their seats (Hernandez, 2011). In other cases, for instance, in Mark Foley’s congressional page sex messaging scandal, Republican leadership also moved quickly, pressuring Foley to resign so that media attention would die away and regular legislative business could resume uninterrupted (Halloran, 2011). Since for-profit media norms and routines may oblige outlets to pay more attention to sexual scandals than to financial scandals, there is also an expectation that sex scandals will result in electoral “defeat” at higher rates than will financial scandals – sex scandals receive such intense coverage, members often have to step aside rather than draw attention from their party’s political agenda. This chapter investigates this possibility using aggregated data, modeling the interactive effects of scandal type and frequency of media coverage on electoral outcomes in the U.S. House of Representatives over a 16-year period.
Data. To analyze the above hypotheses, an aggregate dataset was constructed that captured variables associated with members of the U.S. House of Representatives who were
accused of a scandal were identified using a variety of sources, including the Congressional Quarterly Almanac, investigative reports listed on the House Ethics Committee website (http://ethics.house.gov), and reference texts discussing House members accused of scandal (Herrick, 2003; Long, 2007; Marion, 2010; Roberts, 2001; Thompson, 1995). One problem with relying on these sources is that they tend to exclude non-criminal scandal or scandals that the opposition party decides not to pursue. For example, in the case of Anthony Weiner, an ethics committee investigation was called for by House Republicans, but was never pursued after Weiner resigned from office. Since more recent cases like Weiner’s may not be listed in any of these sources, each source was cross-referenced with Wikipedia’s ongoing, crowd-sourced “list of federal political sex scandals in the United States.” Combined, these sources resulted in an inclusive list of members accused of scandal from 1996 to 2012 (Appendix C).
Members were only included in the analyses if they met two criteria: (1) the member was investigated by the House Ethics Committee or violated a criminal law, or (2) received media coverage regarding an act of financial or sexual misconduct. Both criteria were necessary. For a scandal to be considered as such, the press must reveal the scandal while the representative is still in office (Marion, 2010, p. 11; Thompson, 2000). For instance, it is known that President John F. Kennedy had extramarital affairs while in office. Yet the press chose not to cover these events while Kennedy was president, eliminating the possibility that Kennedy’s power would be delegitimized by scandal while he retained his position (Marion, 2010, p. 11). Therefore, while we may think of the Kennedy affairs as “scandals,” they would not meet the criteria for inclusion here since Kennedy’s power was not compromised by press coverage and public reaction. Based on these criteria, n = 55 U.S. House scandals were used in the analysis.
(2000) and Basinger’s (2013) analyses of congressional scandals were used as frameworks. First, this study collected and aggregated demographic and behavioral data on each U.S. House member to use as controls in the subsequent procedures (Table 4). First, data were collected on the representative’s number of years in Congress, or tenure, at the time the scandal occurred.
Tenure factors into the likelihood a House member will resign or retire his post (Hibbing, 1982).
Members who have represented their districts for long periods of time may be better able to weather scandal accusations that new members who have fewer resources in their home districts.
Descriptive Statistics for Electoral Analysis
Note: n = 51.
Two representatives ran unopposed, garnering a vote share of 100.00% in the pre-scandal election. Following Herrick (2000), the study used Washington Post stories as a proxy for national media attention. Media coverage was logged to account for significant skew (p.001) that resulted from many scandals garnering very few stories and some scandals engendering high amounts of coverage. Subsequent modeling was checked using the unlogged media variable.
Data were also collected on whether the member held a leadership position in the House at the time the scandal was revealed. A leader was defined as a House member who held the rank
Members in these positions of leadership, who possess institutional power, might be more insulated from the effects of scandal. Leaders might also receive more media attention because they hold more power in the chamber. Since party control of the House changes from time to time, it is also important to consider a member’s party affiliation. The Republican variable therefore accounts, in part, for potential differences in formal sanctions for members of in-group and rival parties. The analysis also controlled for vote share in the election prior to the scandal accusations. The vote share variable was continuous, and used a measure of electoral security and competition within the member’s district. House members who had a high vote share in the election prior to a scandal accusation are assumed to have a higher degree of electoral security.
Independent Variables. Media coverage was measured by the frequency of Washington Post stories that mentioned a scandal. This measure acts as a proxy for national media attention and is based on Herrick’s (2000) study of media effects on post-scandal electoral outcomes.
Importantly, story mentions were only measured from the day a scandal was first reported, until the day the member resigned, or until the day of the next election. This procedure was used to eliminate any irrelevant stories prior to the scandal event, and to eliminate scandal coverage that occurred after resignation or Election Day and had no effect on the member’s electoral fortunes.
The LexisNexis Academic database was used to collect data. Media stories were found by searching key terms associated with a scandal (e.g., “Anthony Weiner” “sex” “scandal”). For a news story to be included in the analysis, the story had to mention the representative by name, and make a direct reference to the scandal. In each case, key terms related to the scandal varied and when scandals overlapped, coding for inclusion became problematic. Therefore each article was analyzed to make ensure it contained a mention of both the scandal and the member’s name.
in context, but can be delineated as primarily consisting of financial or sexual misconduct (Thompson, 2000). Each scandal included in the analysis was coded as either “financial” (e.g., bribery) or “sexual” (e.g., extramarital affair) in nature and represented in the analysis using a dichotomous dummy variable. Since this study’s hypotheses make specific predictions about sex scandals, sex scandals were coded high and financial scandals were used as the reference group.
By interacting the scandal type dummy variable with the media coverage variable, it was possible to assess the frequency of media stories devoted to financial and sexual scandals.
Dependent Variable. A trichotomous dependent variable assessed each representative’s electoral outcome in the wake of scandal. This variable measured whether the House member won reelection after being accused of a scandal, whether they ran for reelection and lost, or whether they resigned after the accusations were made public. Winning reelection was coded low and used as the reference group. Thus, the analysis compares the effects of media coverage and scandal type on the probability of losing reelection or resigning to winning reelection.
Importantly, scandals may influence whether a congressman decides to strategically retire or resign from office (Groseclose & Krehbiel, 1994; Wilkins, 2012) and this analysis accounts for this possibility by making resignation a unique outcome on the dependent variable.
Media Coverage of U.S. House Scandals. The study first used a series of one-sample ttests to explore the impact of scandal type – financial or sexual – on frequency of scandal news coverage. To show changes in scandal coverage over time, scandals were divided into three periods for this analysis, corresponding to the presidential administrations of Bill Clinton (1996George W. Bush (2001-2008), and Barack Obama (2009-2012) (see Figure 9).
Average Number of Washington Post Stories By Scandal Type and Time Period Note: n = 51.The Y-axis represents the average number of Washington Post news stories covering U.S. House members accused of financial and sexual scandals from 1996 to 2000 (n = 6), 2001 to 2008 (n = 21), and 2009 to 2012 (n = 23). These values were logged to reduce skew.
mentioning a House sex scandal X! was greater than the average number of stories mentioning House financial scandals X!. Using a one-tailed test (X! − X! ), these differences were significant during the Clinton period t(6) = 2.24, p.05, the Bush period t(21) = 1.58, p.10, and the Obama period from 2009 to 2012 t(23) = 1.89, p.05. Notably, financial scandals involving former lobbyist Jack Abramoff occurred during the Bush period, temporarily increasing the average number of stories about House financial scandals. Yet this appears to be an exception to
providing additional support for H1, which predicted sex scandals will receive more coverage.
Media Coverage and Electoral Outcomes. Considering the unordered, trichotomous nature of the dependent variable, the predictive analysis used a multinomial logistic regression procedure (Long & Freese, 2006). This type of multivariate regression can measure the effect of each independent variable on the probability of being coded in one of the categories of the dependent variable, in this case, as winning reelection, losing reelection or resigning.