Hostname: page-component-7b7567c8c-tnzcf Total loading time: 0 Render date: 2026-11-13T15:29:53.876Z Has data issue: false hasContentIssue false
  • English
  • Français

Surviving Scandal: The Institutional and Political Dynamics of National and State Executive Scandals

Published online by Cambridge University Press:  29 December 2013

Brandon Rottinghaus
Brandon Rottinghaus*
Affiliation:
University of Houston
Rights & Permissions [Opens in a new window]

Abstract

Which factors shorten or lengthen the survival of a scandal involving a chief executive? Using new data tracking scandals involving presidents and governors from 1972 to 2011, I chart the duration of each political, personal, and financial scandal faced by an elected official, their staff, or nominees. I specifically examine institutional, political, and economic factors to investigate what factors quicken a “negative” end to a scandal. National chief executives and their staff are more likely to survive a scandal when they have more partisans in the legislature but are less likely when there is greater political opposition, however there is no comparative effect at the state level. Positive economic growth and public approval have no effect on survival of a scandal at either the national or state levels. These findings clarify how the political environment shapes the duration of executive scandal.

Information

Type
Features
Information
PS: Political Science & Politics , Volume 47 , Issue 1 , January 2014, pp. 131 - 140
Copyright
Copyright © American Political Science Association 2014 

If reports in the media are any indication, political scandals resulting from corruption or personal indiscretions are ubiquitous (Entman Reference Entman2012; Sabato, Stencel, and Lichter Reference Sabato, Stencel and Lichter2000). Governors fly to international locations for extra-marital trysts or attempt to sell Senate seats. Presidents and their senior staff conspire to cover up political crimes. Political nominees are accused of financial misdeeds related to federal income tax returns. Governors are impeached and removed from office for serious crimes. Other observers and scholars concur with these grim assessments (Dunn Reference Dunn1999; Marion Reference Marion2010; Sabato Reference Sabato1993). Scandals are argued to be on the rise because the media is more invasive, communications technology is more pervasive, laws are stricter, and political opponents thrive in using these tactics as political weapons (Garment Reference Garment1992; Ginsberg and Shefter Reference Ginsberg and Shefter1999; Nyhan Reference Nyhan2009).

Clearly, scandals are a common part of political life for most chief executives, but which factors shorten or lengthen the duration of a scandal? The emergence of scandals presents strategic choices that relate to the political environment, most prominently the amount of political strength a politician possesses (Kurtz Reference Kurtz1991). For instance, one factor often referenced in the survival of a scandal is the amount of political support for a politician in the legislature, where greater numbers within the politician's political party provide for “pillars of support” (Shear Reference Shear2011). The state of the economy is also often pointed to as a factor in surviving scandal. For instance, President Clinton was suggested to have survived the Monica Lewinsky scandal because he presided over a robust economy (Zaller Reference Zaller1999). Public approval may also play a role where greater political strength for an executive flows from higher poll numbers (Andolina and Wilcox Reference Andolina, Wilcox, Rozell and Wilcox2000). Of course, the “entertainment” value of a scandal, often related to salacious charges of infidelity or indiscretions, often factor into a quick end for politicians embroiled in scandal (Garment Reference Garment1992). These political disruptions, both short and long, may have consequences for the durability of elected officials.

The survival of scandal, then, has serious implications for governance at the state and national level. In this article, I examine the life cycle of executive scandals to determine which factors hasten an ignominious end to individuals facing scandal. If scandal is “politics by other means” (Ginsberg and Shefter Reference Ginsberg and Shefter1999; Kurtz Reference Kurtz1991), then the fate of politicians' political scandals should be tied to the environmental factors governors and presidents face. I argue that political strength is the key to surviving scandal: the stronger politicians are the more likely they are to weather the effects of (or “survive”) a scandal. Using two new data sets of political scandals involving presidents at the national level and governors at the state level from 1972 to 2011, I chart the duration of each scandal faced by an elected official, their staff, or nominees. Then, the relevant political and economic factors to this time span are connected to investigate what factors quicken the “end” of the scandal, defined as when the scandal ends in a negative outcome for the individual or the administration. These results help clarify the conventional wisdom of how chief executives manage crises and the role of the political environment in affecting the scope of accusations of scandal.

POLITICAL STRENGTH AND SURVIVING SCANDAL

The crux of the argument is simple: the more political strengths chief executives have, the more likely they are to weather a scandal in which they or their political associates find themselves. As political strategies shifted from “electoral” to “institutional” combat, politicians have been forced to confront their political opponents in alternative institutional venues, often through accusations of wrongdoing leading to scandal (Ginsberg and Shefter Reference Ginsberg and Shefter1999). This means, among other things, politicians must use their institutional political power to quell opposition within the system. This “public integrity war” has made it impossible for the public and the media to distinguish between legitimate character issues and those motivated by ideology (Roberts and Doss Reference Roberts and Doss1997). An unsupportive (or unsympathetic) political environment often creates a negative effect for a president in the wake of scandal (Brody and Shapiro Reference Brody and Shapiro1989). Of course, the definition of political strength is potentially extensive and can be operated in many ways. This section identifies three key institutional and political elements that encompass political strength: partisan legislative strength, public approval, and economic success.

The first factor in predicting survival of scandal is partisan legislative support. Specifically, surviving scandal is intimately linked to the ability of the chief executive to continue to govern. One element of this is related to the likelihood of removal from office for the crimes associated with the scandal (Baumgartner and Kada Reference Baumgartner and Kada2003).Footnote 1 The ability to stay in office is directly related to the amount of support a chief executive might have, when more partisans in the legislature is likely to mean a greater chance of political survival (see Dimock and Jacobson Reference Dimock and Jacobson1995). Put another way, more partisans create a “legislative” shield for the chief executive (Hinojosa and Perez-Linan Reference Hinojosa and Perez-Linan2006). A chief executive's fellow partisans are less likely to vote to impeach or remove a chief executive from office and can therefore be counted on as a measure of political strength during a scandal. If chief executives are unable to govern in the aftermath of a scandal, in the form of blocked initiatives, delayed votes, or lengthy and distracting investigatory hearings (Busby Reference Busby1999; Quirk Reference Quirk1998), they may be more likely to step aside after a scandal breaks. Indeed, Meinke and Anderson (Reference Meinke and Anderson2001) found that scandals have significant negative effects on presidential support on key legislation, suggesting a negative relationship between effective governing and scandal-ridden presidents. Nyhan (Reference Nyhan2009) similarly found that there were more verified scandals when government was divided, suggesting institutional friction leads to more accusations of scandal. The ability to govern (and govern effectively) through partisan support should be related to the duration of a scandal.

Second, public approval is also argued to play an important role in enhancing the political strength of a chief executive. If the public backs a politician (by virtue of having a successful political career) who is linked to a scandal, the politician should have a reservoir of support to preclude removing that person from their position. An impeachment move would be unpopular for legislators, and it would be unwise for a chief executive to step down considering strong popular backing. Presidents under fire from scandal may maintain positive public support by being linked to popular initiatives (Quirk Reference Quirk1998) or by successful public relations (Sonner and Wilcox Reference Sonner and Wilcox1999). Approval and legislative strength may be linked, as Hinojosa and Perez-Linan (Reference Hinojosa and Perez-Linan2006) note for presidents, where “popular presidents are more capable of enduring accusations, while declining presidential approval typically provides a strong signal for legislators to defect from the president's camp” (655).

Third, the state of the economy is often pointed to as a key factor for politicians, especially presidents, in weathering charges of scandal. By some standard, a strong economy is the result of effective governing, and voters and legislators ought to be less likely to remove a chief executive who has governed over economic success while in office. Chief executives are inoculated by strong economic growth during scandal because the public, separating the job of the chief executive from their private lives even during scandals, rewards success in governing (Miller Reference Miller1999). For instance, retrospective and prospective appraisals were advantageous for President Clinton who faced damaging details about a personal affair. Andolina and Wilcox (Reference Andolina, Wilcox, Rozell and Wilcox2000) note that the public recalled when economic times were worse in past administration and “with most trends tilting in a positive direction, they were loath to rock the boat” (183). A strong economy indirectly helped President Clinton maintain high approval ratings “despite the media frenzy surrounding the Monica Lewinsky scandal” (Ginsberg and Shefter Reference Ginsberg and Shefter1999, 173).

METHODS AND DATA

Defining Scandal

Several definitions of political scandals exist. For instance, Marion (Reference Marion2010) requires that a public figure has been “accused of unethical or immoral behavior” defined as offending behavior or an event “that is disgraceful, shameful or discredits someone” or that transgresses “societal norms, moral codes or values” (11). Thompson (Reference Thompson2000) offers a detailed definition that requires that actions “transgress or contravene certain values, norms or moral codes” (13) and that the actions' disclosure might damage responsible individuals' reputations, so that they attempt to conceal the action. My definition requires that a scandal must involve allegations of illegal, unethical, or immoral wrongdoing. The definition includes adultery because of the unique place of inappropriate sexual relationships in the panacea of political scandals (Rosen Reference Rosen2009). Including sex scandals ensures that we include “transgressions” of conventional morality, but we wish to exclude gossip, innuendo, and unsubstantiated rumors of private behavior.Footnote 2 For scandals of nonsexual nature, an alleged violation of law or a code of ethics is the threshold question. This definition intentionally excludes executive branch incompetence, unpopular policy decisions, or negative press.

This definition of executive scandal requires that the misbehavior identified here must involve the president, a governor, a senior administration official, or a federal nominee. At the federal level, scandals involving the vice president, cabinet secretaries, officials with cabinet level rank (the White House Chief of Staff, or the Director of the Office of Management and Budget), agency heads at the federal level, high-level political appointees (down to the level of deputy, assistant or under secretary), ambassadors and envoys, the First Lady, and senior campaign staff are included. Nominees for national executive- or judicial-branch positions during the period when a nomination is under Senate consideration are also considered. At the state level, I include scandals involving the governor, senior staff (including agency heads and high-level political appointees), the spouse of the governor, and senior campaign staff. I also included lieutenant governors when they were elected as part of a “ticket” with the governor. Other independent state-level executive offices (attorney general, comptroller, auditor, treasurer) are excluded because these individuals are most often elected independent of the governor. The scandal had to take place during the individual's time in office (not, for instance, revelations after the principal or staff member left office). To summarize, only scandals involving executive office elected officials (presidents and governors), affiliated persons, and nominees revealed before an administration ends are considered.Footnote 3

Using this definition, I identified 87 presidential level scandals (involving 126 individuals) that occurred between 1972 and 2009 (ending in January of 2009 with the end of the George W. Bush term) and 39 gubernatorial level scandals (involving 54 individuals) between 1972 and 2011.Footnote 4 Using secondary texts follows the lead of other scholars who have searched for a universe of political scandals (Kim and Bahry Reference Kim and Bahry2008; Puglisi and Snyder Reference Puglisi and Snyder2011). If the principle focus of the scandal was the chief executive, I recorded this as well and included it in the model to control for the fact that the duration of the scandal will be different for the elected chief executive than others, as discussed. This strategy also allows comparable evaluation of chief executives, at both the state and national level, who share similar institutional settings with respect to legislatures and heading an executive branch. The online appendix uploaded to the PS website https://journals.cambridge.org/action/displayJournal?jid=PSC lists all the scandals and the individuals associated with each scandal. Although some scandals are more important than others, alternative models in the appendix demonstrate that, even using multiple measures of the significance of a scandal, that there is no independent effect on surviving scandal.Footnote 5

Type of Scandal

Individual scandals are classified into three types: financial, political corruption, and personal. Financial scandals are scandals when individuals personally financially profit from their actions, such as embezzlement of funds, accepting bribes or political payoffs, and nonpayment of state or federal taxes (Thompson Reference Thompson2000). Political corruption scandals involve abuse of the authority of an individual's office (which is not financial in nature or when an individual does not profit financially from his or her involvement), a breach of public trust, or violation of an individual's oath of office (Markovits and Silverstein Reference Markovits, Silverstein, Markovits and Meier1988). These scandals involve violation of separation of powers (Watergate), violations of the Constitution, the (illegal) injection of politics into nonpartisan decisions, campaign violations, or rules violations (Governor Blagojevich “selling” a Senate seat). Personal scandals involve the immoral or unethical personal behavior of an individual, especially adulterous in nature (Lewinsky affair during the Clinton administration) but also includes use of illegal drugs (Supreme Court nominee Douglas Ginsberg), illegal household staff (Secretary of Labor nominee Linda Chavez), or theft (see Kagay Reference Kagay1999). For purposes of analysis, I only use two categories: financial scandals (Financial Scandal) and personal scandal (Personal Scandal).

Institutional Variables

To examine the effect of institutional variables, I review the relationship between the executive and legislative branches through four variables. First, I use a variable for the size of the opposition (Opponent Average in Congress) by averaging the size (in total percent of seats) of the opposition party in both chambers for each legislative session. Second, I include a dichotomous measure of divided government, where any amount of divided government (in either or both chambers at the federal or state level) is treated as divided government (Divided). Third, variables for the number of copartisans in both the upper (President's or Governors' Partisans Senate) and lower (President's or Governor's Partisans House) chamber in the same party as the president or governor are given.Footnote 6 National-level data were taken from Vital Statistics on American Politics (Reference Stanley and Niemi2011). State data are taken from Klarner (Reference Klarner2003), available on the State Politics and Policy website.Footnote 7

Approval

To measure executive approval at both the national and state level, I recorded the approval measure of the president or governor taken as close as possible to the date of the break of the scandal story (but not after the story broke) (Approval). The question asked “Do you approve or disapprove of the way [President or Governor (name)] is handling his job as [(president) or (governor)]?” Responses are collapsed into positive and negative values, excluding those who responded “don't know.” National-level data were taken from iPoll's report of Gallup Poll data for the appropriate time period. The state-level data was taken from the US Officials' Job Approval Ratings (JARs), a Cooperative Project of the University of Rochester, the University of North Carolina at Chapel Hill, and George Washington University.Footnote 8 When more accurate state-level data could be located at Survey USA, those data were used.Footnote 9

Economic Variables

To measure the state of the economy, we use the change in Gross Domestic Product (GDP) to represent national-level economic conditions and Gross State Product (GSP) at the state level to represent state-level economic conditions. First, for measuring change in GDP at the national level, I calculate the percentage change in gross domestic product from the previous year (in real dollars) as tabulated by the US Census Bureau (GDP Change). Second, I calculated percent change in Gross State Product from the previous year, the sum of incomes earned by labor and capital, including the wages and salaries that workers earn, the incomes earned by individuals and corporations, business taxes, and federal excise taxes (GSP Change).Footnote 10

Control Variables

To examine any temporal aspects of the survival of scandal, I include control variables in the models. First, several states have the provision for a recall, where citizens can remove and replace public officials before the end of their term in office. The variable recall (Recall) was coded dichotomously, where “1” means that the state has the provision for a recall.Footnote 11 Second, because presidents and governors might be more vulnerable in their second terms, I include a variable to measure whether the scandal broke in the respective chief executive's second term (Second Term). This is coded dichotomously where “1” indicates the chief executive is in his or her second term. Third, states that have term limits on their governors' terms in office are coded dichotomously. Of primary concern is whether or not the governor is in his or her last term and is term limited from being elected again. This is coded dichotomously where “1” indicates the chief executives are in their second term of a term-limited time in office (Term Limited Final Term).Footnote 12 Most state constitutions also limit a governor's term to two consecutive or two lifetime terms. In all but one instance in the data (Nevada), the governor is term limited to four years. The president, by virtue of the Twenty-Second Amendment, can only serve two consecutive four-year terms. Fourth, I include a “counter” variable (Scandal Count) that counts the number of scandals starting from the first scandal in an administration that buttresses the independence of the scandals for each administration. An administration official may be more vulnerable to removal if the administration has had more scandals since the administration began reflecting scandal “fatigue” among the media and public (Kumlin and Esaiasson Reference Kumlin and Esaiasson2012; Waisbord Reference Waisbord2004).

Model

Because I am interested in the factors that contribute to a president or governor's “survival” of a scandal, I use a series of duration models.Footnote 13 A Cox proportional hazard model is a semiparametric model that allows the hazard function to be unestimated—this is especially useful when the shape of the baseline hazard model is unknown (Cleves et al. Reference Cleves, Gould, Gutierrez and Marchenko2010).Footnote 14 This model also assumes the shape of the hazard is the same for all the subjects, a reasonable assumption considering that the effect of scandal in the modern era (especially post-Watergate) should be theoretically similar. The Cox model asserts that a hazard rate for the jth subject in the data is h(t| xj ) = h 0(t) exp(x j β x ), where the regression coefficients, βx, are estimated from the data (Cox Reference Cox1972). This also helps control for duration dependence (Zorn Reference Zorn2000). The parameterized effects of each variable are measured with the variables using categorical and continuous variables. We also cluster by scandal because the effects of each scandal (including the severity and personnel involved) are likely to be similar within each scandal, providing within-subject correlation (Lin and Wei Reference Lin and Wei1989). This provides a valid representation of the “sample-to-sample variability” of the coefficients (Cleves et al. Reference Cleves, Gould, Gutierrez and Marchenko2010). Although selection effects are problematic for some duration models (Boehmke, Morley, and Shannon Reference Boehmke, Morley and Shannon2006), the nature of the data collection for these data resolves these possible issues.Footnote 15

Scandal Length and “Failure”

In establishing a time span for the models as required for duration models, the scandal begins (t 1x) when the charges are made public for the first time and a scandal is considered ended (t nx) when the accused individual is exonerated, formally leaves his or her position, or the presidential or gubernatorial administration ends. The origin of the scandal can involve a news story in a major daily newspaper (searched in Lexis-Nexis), a report to Congress, an internal investigation, or other secondary sources. This variable spans the life cycle of the scandal from the first break of the story to the end of the scandal whether that end be politically or legally favorable for the individual charged in the scandal. However, to be more specific as required for predicting the end to the span of the sequence, “failure” in the duration models (or the right censored variable) is identified as when the scandal ends in a negative outcome for the individual or the administration. This includes the following scenarios: an individual being fired from their position, resigning (or being forced to resign), being indicted at any level or removed from their position or office in some way. Failure in this case excludes instances where an individual is charged with a crime after they leave office or they are indicted but remain in office until their term runs out. Each scandal has a definite temporal beginning and end, allowing for completion of the duration models. Several examples are identified in the next section.

DURATION OF EXECUTIVE SCANDALS

In general, scandals tend to be short lived events (Markovits and Silverstein Reference Markovits, Silverstein, Markovits and Meier1988, 3). Figure 1a depicts a histogram of governor-related scandals. The average length of a scandal (excluding one outlier) is 255 days.Footnote 16 Most state-level scandals last for less than 100 days, with only a few lasting longer. For instance, governor James McGreevey of New Jersey admitted that he was a homosexual and had an affair with another man the same day he announced he planned to step down as governor (the end of the scandal, or “failure” in the model). Scandals do not always end badly for governors, even if they are short. For instance, governor Bill Sheffield of Alaska, who was accused of misusing his office and lying under oath, was spared from impeachment after just 11 days of hearings after the Senate decided that there was not enough credible evidence to impeach him. In this instance, the scandal did not end in “failure” (as defined earlier). Longer scandals are generally those that require time to investigate. For instance governor Marvin Mandel of Maryland was arrested in 1975 for allegedly taking financial remuneration for assisting his codefendants in acquiring interest in horse racing tracks and increasing the number of days the state would allow racecourses to race horses (Franklin Reference Franklin1977). After two trials on charges of racketeering and mail fraud, he was found guilty and forced to resign his seat by state law.

Figure 1

Histogram of Presidential and Gubernatorial Scandals

Note: Gubernatorial data span 1972 to 2011. Presidential data span 1972 to 2009.

The data on presidential scandals also demonstrate that most national chief executive scandals end quickly. In fact, 19 scandals required less than one month from start to finish. Figure 1b presents a histogram of the duration of time until the scandal ended—like gubernatorial scandals, most of the cases are over within 100 days. Often, these fast-concluding scandals involve nominees, such as Bernard Kerik (Secretary of Homeland Security nominee), Douglas Ginsburg (Supreme Court nominee) and three “Nannygate” nominees (Kimba Wood, Zoe Baird, and Linda Chavez). Among office holders, Eli Segal, the first CEO of Americorps, resigned abruptly when it was alleged that he had conducted improper business relationships. Peter Bourne, President Carter's Drug Czar, resigned when he was caught writing a prescription for a fictitious individual (under the guise of giving the prescription to a White House colleague). In each case, because the scandal ended in resignation or firing of the official, each was classified as a “failure.” The median scandal lasts six months, but some scandals last much longer, particularly when an independent counsel is brought into play. For instance, the scandal involving President Clinton harassing Paula Jones lasted 53 months.

WHAT INCREASES THE RISK OF A SCANDAL ENDING BADLY?

The analysis in tables 1 (state-level executives) and 2 (national-level executives) are the results of the models run with a Cox proportional hazard model as specified previously.Footnote 17 Parameterized models (included in the online appendix https://journals.cambridge.org/action/displayJournal?jid=PSC) demonstrate similar findings, suggesting robustness to the findings. In general, the data fit the model well. Cox-Snell (Cox and Snell Reference Cox and Snell1968) residuals plotted against Nelson-Aalen cumulative hazard function demonstrates substantial similarity, suggesting that the model fits the data. In addition, both tables 1 and 2 list the Harrell's C concordance statistic and the Somers' D statistic, measures of the agreement of predictions with observed failure (Harrell et al. Reference Harrell, Califf, Prior, Lee and Rosati1982; Harrell, Lee, and Mark Reference Harrell, Lee and Mark1996)—these demonstrate moderate to high levels of correctly identified order of the survival times.Footnote 18 The coefficients in each table are exponentiated coefficients of the hazard ratio, meaning that these coefficients have an interpretation of the ratio of the hazard for a 1-unit change in the corresponding covariate (Cleves et al. Reference Cleves, Gould, Gutierrez and Marchenko2010). Hazard ratios less than 1.0 indicate an increase in duration to “failure,” whereas hazard ratios greater than 1.0 indicate a decrease in duration. In essence, the models substantively explain the amount of time to “failure” based on the conditions outlined earlier—coefficients greater than 1.0 show a quicker negative end to a scandal, where coefficients less than 1.0 show a longer negative end to a scandal (akin to “surviving” scandal).

Table 1

Gubernatorial Scandals Duration Models

Note:

‡ Truncated from 1993 to 2009 because of availability of approval data. Coefficients are exponentiated coefficients of the hazard ratio. Dependent variable: total time (in days) from when the story broke to when the scandal ended. Failure means the resolution to the scandal ended in resignation, firing or removal from office. The standard errors are robust standard errors, clustered by each scandal.

*** indicates statistical significance at p < .01.

** p < .05.

* p < .10.

Robust standard errors in parentheses.

Table 2

Presidential Scandals Duration Models

Note: Coefficients are exponentiated coefficients of the hazard ratio. Dependent variable: total time (in days) from when the story broke to when the scandal ended. Failure means the resolution to the scandal ended in resignation, firing or removal from office. The standard errors are robust standard errors, clustered by each scandal.

*** indicates statistical significance at p < .01.

** p < .05.

* p < .10. Robust standard errors in parentheses.

Table 1 identifies scandals involving state-level chief executives or staff.Footnote 19 The most consistent predictor of “failure” (or non-“failure”) in the duration models in each specification is whetherthe target was the governor. In each model specification, being the governor was associated with a 50% to 70% decline in the hazard of having the scandal result in removal from office, resignation, indictment, or conviction (our defined “negative” end). Being a powerful state official means that these individuals can more easily weather scandal because they are more protected by vested political interests and seek to keep a tight grip on state power (see Brown Reference Brown2006, Reference Brown2007). Putting it another way, figure 2a uses a kernel smoother to graph the hazard function for the “survival” of each political actor at the state level. The plotted lines show that the likelihood of a governor having a negative end to a scandal is significantly less than for others. The change in the probability of “failure” for governor and staff over time parallel each other, suggesting a degree of shared fate among state-level executive officials.

Figure 2

Cox Proportional Hazard Smoothed Functions: Chief Executive versus Staff

Note: Cumulative hazard functions of chief executives versus staff. Kernel smoother (Gaussian) employed which averages values over a moving window of data.

Personal scandals hasten a quick negative end to scandal at the state level (in models 1 and 2), where individuals facing personal scandals were three times more likely to have their careers end in removal from office, resignation, indictment, or conviction than scandals not of a personal nature. This finding fits a pattern found previously when personal scandals are more likely to be shorter than financial ones (Basinger and Rottinghaus Reference Basinger and Rottinghaus2012a) because alleged financial wrongdoing may require time to investigate and involve criminal wrongdoing, which tends to prolong their duration relative to personal scandals that rarely trigger criminal liability for the individuals involved. Indeed, such financial scandals are treated differently than other types of scandals (Andolina and Wilcox Reference Andolina, Wilcox, Rozell and Wilcox2000). The media also focus more attention on scandals that are salacious (Kiousis Reference Kiousis2003; Maurer Reference Maurer1999). Individuals also seek to limit personal embarrassment to themselves or their family when revelations of a personal scandal become public so they often resign more quickly when faced with a personal scandal.

Importantly, neither the institutional, economic, nor the public approval variables are statistically significant for cases involving state-level scandals in any of the modeles in Table 1.Footnote 20 First, for the institutional variables, this includes measures of political support such as the number of fellow partisans in the legislature (in either chamber) or measures of opposition such as divided government, or the average size of the opposition in the legislature. This suggests that governors' and their staffs' survival during scandal is not related to their institutional support or opposition.Footnote 21 Second, gubernatorial approval when the scandal breaks is also not statistically significant in model 1 (the time is truncated to 1993 to 2011 because of the availability of accurate polling data). The nature of the scandal seems to be of greater importance than the ability of the governors or their staff to count on loyal partisans for survival assistance. Because most of the cases of state-level executive scandals involve the governor and governors are less likely to have scandals end their political careers, institutional and political issues may not play a major role in affecting their chances for survival. Third, the change in GDP, signaling an improving economy, had no effect on the survival of governors. This finding is contrary to what other scholars have asserted, as discussed earlier, where a positive economy was thought to be a boon to surviving scandal.

Turning to national chief executive scandals, table 2 identifies the results of several model specifications involving the president, the cabinet, the White House staff, or executive nominees. Even more so than for governors, presidents are substantially less likely to end a scandal with removal from office, resignation, indictment, or conviction than their staff or executive nominees. Clearly the White House seeks to protect the president, even at the expense of public accountability. The formal structure of the White House encourages individuals who are damaging the administration to step aside. These staff can be thought of as political “lightning rods,” which are erected to draw the heat of blame away from more highly ranked administrators (Ellis Reference Ellis1994). Indeed, embattled political advisers historically have been the first to depart from office in the aftermath of a scandal (Cannon Reference Cannon2005). Figure 2b uses a kernel smoother to graph the hazard function for the “survival” of both types of political actors at the federal level. The plotted lines again show that the likelihood of a president having a negative end to a scandal is significantly less than for others, even more than the difference between the governor and gubernatorial staff in figure 2a.

Mirroring the modest effect found for governors, presidential scandals involving personal matters or financial matters tend to result in removal from office, resignation, indictment, or conviction much more quickly than financial scandals. The results show that the hazard of the scandal ending in “failure” is between one-and-a-half and two-and-a-half times greater for personal scandals than financial scandals. Both personal and financial scandals quicken an end to a political career, but personal scandals tend to do so more quickly. Although personal scandals (especially scandals involving personal indiscretions) may not be a “career killer” as it once was, the results from table 2 demonstrate that scandals involving personal indiscretions do hasten a more quick end to a political career (Baker Reference Baker2009). The public does distinguish between types of scandals (Doherty, Dowling, and Miller Reference Doherty, Dowling and Miller2011; Kagay Reference Kagay1999), and while both financial and personal scandals have different negative effects, when charting whether each scandal causes a more rapid end to a political career, personal scandals seem more damaging.

Considering the political and institutional mechanisms for surviving scandal, polled approval has no effect, but institutional factors have a large effect.Footnote 22 Like that of governors in table 1, the approval of the president before the scandal breaks has no effect on the duration of a scandal. On one hand, in terms of institutional support, when presidents have more support from copartisan members of the House or Senate, they are 9% and 10% (respectively) less likely to have a scandal end in removal from office, resignation, indictment, or conviction. On the other hand, putative opposition to the president has a positive effect on scandals ending in a politically challenging manner. When presidents have a greater percentage of opponents in Congress, scandals are 10% more likely to end in removal from office, resignation, indictment, or conviction, whereas when government is divided, presidents are 20% more likely to have scandals end in their removal from office, resignation, indictment, or conviction. In short, political strength in the form of more partisan supporters leads to a greater probability of survival whereas more opposition leads to a quick and negative end, consistent with expectations. This finding mirrors what scholars find in other presidential democracies (Kim and Bahry Reference Kim and Bahry2008).

Similar to the gubernatorial scandal findings, national economic success has an effect on scandal survival. Similar to table 1 explaining gubernatorial scandals, in table 2 the survival of presidential scandals are not affected by a change in Gross Domestic Product from the prior year. Although a good economy may serve as ballast against surviving some scandals, in general, there is no substantive effect of economic success on the likelihood of an executive scandal ending badly. The institutional support or opposition for the president is a more significant predictor of the survival of a scandal.

CONCLUSION

Scandals can result in untold damage to chief executives (Bowler and Karp Reference Bowler and Karp2004; Genovese Reference Genovese, Genovese and Farrar-Myers2012; Quirk Reference Quirk1998). Among other outcomes, scandals can decrease trust in government (Chanley, Rudolph, and Rahn Reference Chanley, Rudolph and Rahn2000), generate negative policy implications (Szasz Reference Szasz1986) and create extensive and often far reaching ramifications for American politics (Farrar-Myers Reference Farrar-Myers, Genovese and Morgan2012; Genovese and Morgan Reference Genovese and Morgan2012). Witcher (Reference Witcher, Jeffrey and Maxwell-Long2004) reminds us that three of the last eight presidents (Richard Nixon, Ronald Reagan, and Bill Clinton) have undergone a major investigation into their behavior and that of their advisers or associates. Journalists also lament that news reports consistently reveal “headlines from Washington to New York and beyond filled with word of scandal or allegations of wrongdoing” (Zeleny Reference Zeleny2010). Given this ubiquity and the importance of scandal and the toll that such events have on cooperation and governing, we need to better understand the dynamics of what shapes the duration of a scandal and the way scandals conclude.

Conventional wisdom holds that politicians must be cunning strategists, effective leaders, and popular to survive a political scandal (Kurtz Reference Kurtz1991; Shear Reference Shear2011). Although there is some general truth to this, this article shows some consistent patterns affecting national and state chief executives. If surviving scandal is the politician's (or staff 's) goal, it is better to be the elected representative and have greater support in the legislature. If a scandal is personal, financial, or when opposition within the legislature is great, scandal is much more difficult to survive. A booming economy does not necessarily help executive officials survive a scandal. And, although more popular politicians are thought to find it easier to weather a scandal (Sonner and Wilcox Reference Sonner and Wilcox1999), these data demonstrate approval of either presidents or governors has no effect on the duration of scandal. Greater political support from a president's (but not a governor's) partisans is generally helpful in surviving scandal. The ability to survive scandal, then, is connected to the executive's ability to govern and maintain a viable political coalition rather than the amount of adulation received by the public, affirming the partisan institutional importance of governing in crisis.

Footnotes

1 For impeachment at the federal level, a majority of the House members vote to impeach the president on specific articles of impeachment and a two thirds majority of the Senate is required for conviction and removal from office. Impeachment at the state level varies by state but in most states a majority (sometimes a two thirds majority) is required for impeachment in the House (lower chamber) and two thirds majority is required in the Senate (upper chamber) for removal (Book of the States 2010). In Alaska, the process is reversed, where the Senate impeaches the governor and the House sits as jury to decide guilt.

2 By using newspapers and official sources to determine when scandals broke (and hence whether or not they existed), this minimized the presence of false or malicious rumors from the data.

3 “Scandal” is rarely applied to individuals who are longtime civil service employees who may engage in tomfoolery or financial corruption—their cases are handled at lower levels and the implications for this wrongdoing are generally minimal and certainly not associated with the president or governor.

4 We generated the list of scandals using a several step process. As a first step, two research assistants generated lists of “events” that fit one or more elements of our definition of scandal. The list was drawn from books that claim to be “encyclopedias” and “almanacs” of scandals, supplemented by analytical books and other commentaries on White House or gubernatorial scandals. The main sources that our research assistants relied on were: Etzioni (Reference Etzioni1995), Garment (Reference Garment1992), Greenberg (Reference Greenberg2000), Grossman (Reference Grossman2003), Long (Reference Long2007), Marion (Reference Marion2010), Ross (Reference Ross1988), Schultz (Reference Schultz1999) and Genovese and Farrar-Myers (Reference Genovese and Farrar-Myers2012). As a third step, the list of events plus all materials were given to a third research assistant, along with a rubric for deciding whether each event was a scandal.

5 Appendix table 5 lists results from models with two variables measuring the importance of significance of a scandal: first the number of stories in the New York Times for the duration of the scandal and, second, a measure of whether or not the crime in question was a felony infraction (as opposed to an immoral act or a misdemeanor). In both cases for both presidential and gubernatorial scandals, there is no independent effect of either variable.

6 Some states only have one chamber, but since there were no recorded scandals in these states for this period, this was not an issue.

7 State Politics and Policy website (http://www.indstate.edu/polisci/klarnerpolitics.htm), accessed July 5, 2010.

8 U.S. Officials' Job Approval Ratings (JARs) (http://www.unc.edu/~beyle/jars.html), accessed June 23, 2012.

9 Survey USA (http://www.surveyusa.com/50statetracking.html), accessed June 23, 2012.

10 In the report for revised estimates for GSP in 1997, the report noted (with respect to the differences between GDP and GSP): “Total GSP for the nation, which is derived as the sum of the state estimates, differs from GDP because GSP excludes the statistical discrepancy, the compensation of federal civilian and military personnel stationed abroad, and government consumption of fixed capital for military structures located abroad and for military equipment, except office equipment.”

11 National Conference of State Legislatures booklet on Recall of State Officials. Nineteen states have recall and the provisions for who can be recalled and the signature requirements are all state specific. In each case, the governor is eligible to be recalled.

12 These data were taken from The Book of the States from the Council of State Governments (www.thegreenpapers.com/Hx/Lengthoftermgovernor.phtml), accessed July 10, 2012.

13 Several elements of normality are violated that precede the use of OLS regression, most prominently the presences of data censoring.

14 In alternative models, when the hazard is parameterized using either an exponential or Weibull function, the results are substantially similar.

15 The problem emerges in a non-random sample if unobserved factors affect the duration of an event and whether the event is observed at all. In this case, the media may be a factor that influences both the reporting of a scandal and the length. However, because the start and end of scandals are endogenously determined by the behavior of an individual, the role of the media is often secondary. Further, the media may only report what official sources are reporting rather than investigating and uncovering scandals on their own, which happen rarely (Entman Reference Entman2012; Lang and Lang Reference Lang and Lang1983). However, as a precaution, we examined whether or not there was a selection effect using an estimator developed by Boehmke (Reference Boehmke2005). The results, available in the appendix, are substantially similar to those in the text, suggesting no significant selection effect.

16 Republican Governor of Mississippi Kirk Fordice was accused of having an affair with his high school sweetheart while in office (Ayers Reference Ayers1999). Governor Fordice did not admit the affair until later but a considerable amount of time was spent speculating about the Governor's travels and financial purchases until he disclosed the details (Harrison Reference Harrison1996).

17 Alternative models that identify a competing risk of the scandal ending in firing, prosecution or dismissal and the scandal ending because of a lack of indictment or the executive's term in office ending reveal substantially similar results.

18 The Harrell's C ranges from 0 to 1 and the Somers' D ranges from −1 to 1. For instance in the Harrell's C in tables 1 and 2, the models correctly identify the order of survival times for pairs upward of 60%.

19 Some of the cases were dropped due to missing data for some of the observations (especially related to the ending of the scandal).

20 An interaction between the size of partisan support and public approval was also not statistically significant.

21 In alternative models a dummy variable was used to indicate whether or not the requisite number of votes were required for impeachment or removal from office (The Book of the States 2010).

22 An interaction between the size of partisan support and public approval was also not statistically significant.

References

REFERENCES

Andolina, Molly W., and Wilcox, Clyde. 2000. “Public Opinion: The Paradoxes of Clinton's Popularity.” In The Clinton Scandals and the Future of American Government, eds. Rozell, Mark J. and Wilcox, Clyde. Washington, DC: Georgetown University Press.Google Scholar
Ayers, B. Drummond Jr. 1999. “Political Briefing; Political Soap Opera Rivets Mississippi.” New York Times, June 11. Google Scholar
Baker, Gerard. 2009. “Sex Americana.” The Wall Street Journal, July 3. Google Scholar
Basinger, Scott, and Rottinghaus, Brandon. 2012a. “Skeletons in the White House Closets: An Empirical Investigation into Modern Presidential Scandals.” Political Science Quarterly 127 (2): 213–39.CrossRefGoogle Scholar
Basinger, Scott, and Rottinghaus, Brandon. 2012b. “Stonewalling: Explaining Presidential Behavior During Scandal.” Political Research Quarterly 65 (2): 290–302.Google Scholar
Baumgartner, Jody C., and Kada, Naoko. 2003. Checking Executive Power: Presidential Impeachment in Comparative Perspective. Westport, CT: Praeger.Google Scholar
Boehmke, Frederick J. 2005. DURSEL: A Program for Duration Models with Sample Selection (Stata version). Version 2.0. Iowa City, IA: University of Iowa. Google Scholar
Boehmke, Frederick J., Morley, Daniel S., and Shannon, Megan. 2006. “Selection Bias and Continuous-Time Duration Models: Consequences and a Proposed Solution.” American Journal of Political Science 50 (1): 192–207.Google Scholar
Book of the States 2010. The Council of State Governments. Google Scholar
Bowler, Shaun, and Karp, Jeffrey A.. 2004. “Politicians, Scandals and Trust in Government.” Political Behavior 26 (3): 271–87.Google Scholar
Brody, Richard A., and Shapiro, Catherine R.. 1989. “Policy Failure and Public Support: The Iran-Contra Affair and Public Assessment of President Reagan.” Political Behavior 11: 353–69.Google Scholar
Brown, Lara M. 2006. “Revisiting the Character of Congress: Scandals in the U.S. House of Representatives, 1966–2002.” The Journal of Political Marketing 5 (2): 149–72.Google Scholar
Brown, Lara M. 2007. “It's Good to Be an Incumbent: Scandals, Corruption, and the 2006 Midterm Election.” Extension of Remarks, APSA Legislative Studies' Section Newsletter, Vol. 30, No. 1. Google Scholar
Busby, Robert. 1999. Reagan and the Iran-Contra Affair: The Politics of Presidential Recovery. New York: St. Martin's Press.Google Scholar
Cannon, Lou. 2005. “Lessons of Scandals Past.” The Washington Post, October 31. Google Scholar
Chanley, Virginia A., Rudolph, Thomas J., and Rahn, Wendy. 2000. “The Origins and Consequences of Public Trust in Government. A Time Series Analysis.” Public Opinion Quarterly 64 (3): 239–56.Google Scholar
Cleves, Mario, Gould, William, Gutierrez, R. Obertto G., and Marchenko, Yulia V.. 2010. An Introduction to Survival Analysis Using Stata. College Station, TX: StataCorp.Google Scholar
Cox, D. R. 1972. “Regression Models and Life-Tables.” Journal of the Royal Statistical Society, Series B 34: 187–220.Google Scholar
Cox, D. R., and Snell, E. J.. 1968. “A General Definition of Residuals.” Journal of the Royal Statistical Society, Series B 30 (2): 248–75.Google Scholar
Dimock, Michael A., and Jacobson, Gary C.. 1995. “Checks and Choices: The House Bank Scandal's Impact on Voters in 1992.” Journal of Politics 57 (4): 1143–159.Google Scholar
Doherty, David, Dowling, Conor M., and Miller, Michael G.. 2011. “Are Financial or Moral Scandals Worse? It Depends.” PS 44: 749–57.Google Scholar
Dunn, Charles W. 1999. The Scarlet Thread of Scandal. New York: Rowman and Littlefield.Google Scholar
Ellis, Richard J. 1994. Lightning Rods: The Politics of Blame Avoidance. Lawrence: University of Kansas Press.Google Scholar
Entman, Robert M. 2012. Scandal and Silence: Media Responses to Presidential Misconduct. Malden, MA: Polity.Google Scholar
Etzioni, Amitai. 1995. Capital Corruption: The New Attack on American Democracy. 2nd ed. New Brunswick, NJ: Transaction Publishers.Google Scholar
Farrar-Myers, Victoria A. 2012. “The Ripple Effect of Scandal and Reform.” In Watergate Remembered, eds. Genovese, Michael A. and Morgan, Iwan W.. New York: Palgrave.Google Scholar
Franklin, Ben A. 1977. “Governor Mandel Guilty on Charges of Fraud and Racketeering.” New York Times, August 24. Google Scholar
Garment, Suzanne. 1992. Scandal: The Culture of Mistrust in American Politics. New York: Anchor Books.Google Scholar
Genovese, Michael A. 2012. “Presidential Corruption: A Longitudinal Analysis.” In Corruption and American Politics, eds. Genovese, Michael A. and Farrar-Myers, Victoria A.. Amherst, NY: Cambria Press.Google Scholar
Genovese, Michael A., and Farrar-Myers, Victoria A.. 2012. Corruption and American Politics. Amherst, NY: Cambria Press.Google Scholar
Genovese, Michael A., and Morgan, Iwan W.. 2012. Watergate Remembered: The Legacy for American Politics. New York: Palgrave McMillan.Google Scholar
Ginsberg, Benjamin, and Shefter, Martin. 1999. Politics by Other Means: Politicians, Prosecutors and the Press from Watergate to Whitewater. New York: W.W. Norton.Google Scholar
Greenberg, Gerald. 2000. Historical Encyclopedia of U.S. Independent Counsel Investigations. Westport, CT: Greenwood.Google Scholar
Grossman, Mark. 2003. Political Corruption in America: An Encyclopedia of Scandals, Power, and Greed. 2nd edition. Westport, CT: ABC-Clio; Google Scholar
Harrell, F. E. Jr., Califf, R. M., Prior, D. B., Lee, K. L., and Rosati, R. A.. 1982. “Evaluating the Yield of Medical Tests.” Journal of the American Medical Association 247: 2543–546.Google Scholar
Harrell, F. E. Jr., Lee, K. L., and Mark, D. B.. 1996. “Multivariate Prognostic Models: Issues in Developing Models, Evaluating Assumptions and Adequacy, and Measuring and Reducing Errors.” Statistics in Medicine 15: 361–87.Google Scholar
Harrison, Eric. 1996. “Mystery Shrouds Crash of Mississippi Governor.” Los Angeles Times, December 18. Google Scholar
Hinojosa, Victor J., and Perez-Linan, Anibal S.. 2006. “Presidential Survival and the Impeachment Process: The United States and Colombia.” Political Science Quarterly 121 (4): 653–75.Google Scholar
Kagay, Michael R. 1999. “Public Opinion and Polling during Scandal and Impeachment.” Public Opinion Quarterly 63 (2): 449–63.Google Scholar
Kim, Young Hun, and Bahry, Donna. 2008. “Interrupted Presidencies in the Third Wave Democracies.” Journal of Politics 70 (3): 807–22.Google Scholar
Kiousis, Spiro. 2003. “Job Approval and Favorability: The Impact of Media Attention to the Lewinsky Scandal on Public Opinion of Bill Clinton.” Mass Communication & Society 6: 435–51Google Scholar
Klarner, Carl E. 2003. “Measurement of the Partisan Balance of State Government.” State Politics and Policy Quarterly 3 (Fall): 309–19.Google Scholar
Kumlin, Staffan, and Esaiasson, Peter. 2012. “Scandal Fatigue? Scandal Elections and the Satisfaction with Democracy in Western Europe, 1977–2007.” British Journal of Political Science 42 (2): 263–82.CrossRefGoogle Scholar
Kurtz, Howard. 1991. “Scandal.” The Washington Post, September 22. Google Scholar
Lang, Gladys Engel, and Lang, Kurt. 1983. The Battle for Public Opinion: The President, the Press, and the Polls during Watergate. New York: Columbia University Press.Google Scholar
Lin, D. Y., and Wei, L. J.. 1989. “The Robust Inference for the Cox Proportional Hazard Model.” Journal of the American Statistical Association 84 (4): 1074–78.Google Scholar
Long, Kim. 2007. The Almanac of Political Corruption, Scandals, and Dirty Politics. New York: Delacorte Press.Google Scholar
Marion, Nancy E. 2010. The Politics of Disgrace: The Role of Scandal in American Politics. Durham, NC: Carolina Academic Press.Google Scholar
Markovits, Andrei S., and Silverstein, Mark. 1988. “Power and Process in Liberal Democracies.” In The Politics of Scandal: Power and Process in Liberal Democracies, eds. Markovits, Andrei S. and Meier, Mark Silverstein New York: Holmes and.Google Scholar
Maurer, Paul J. 1999. “Media Feeding Frenzies: Press Behavior during Two Clinton Scandals.” Presidential Studies Quarterly 29 (1): 65–79.Google Scholar
Meinke, Scott R., and Anderson, William D.. 2001. “Influencing from Impaired Administrations: Presidents, White House Scandals, and Legislative Leadership.” Legislative Studies Quarterly 26 (4): 639–59.Google Scholar
Miller, Arthur H. 1999. “Sex, Politics, and Public Opinion: What Political Scientists Really Learned from the Clinton-Lewinsky Scandal.” PS: Political Science 32 (4): 721–29.Google Scholar
Nyhan, Brendan. 2009. Strategic Outrage: The Politics of Presidential Scandal. PhD Dissertation, Duke University. Google Scholar
Puglisi, Riccardo, and Snyder, James M. Jr. 2011. “Newspaper Coverage of Political Scandals.” Journal of Politics 73 (30): 931–50.Google Scholar
Quirk, Paul J. 1998. “Coping with the Politics of Scandal.” Presidential Studies Quarterly 28 (4): 898–902.Google Scholar
Roberts, Robert N., and Doss, Marion T. Jr. 1997. From Watergate to Whitewater: The Public Integrity War. Westport, CT: Praeger.Google Scholar
Rosen, David. 2009. Sex Scandal America: Politics and the Ritual of Public Shaming. Toronto: Key Publishing House.Google Scholar
Ross, Shelley. 1988. Fall From Grace: Sex, Scandal, and Corruption in American Politics from 1702 to Present. New York: Ballantine Books.Google Scholar
Rottinghaus, Brandon, and Bereznikova, Zlata. 2006. “Exorcising Scandal in the White House: Presidential Polling in Times of Crisis.” Presidential Studies Quarterly 36: 493–505.Google Scholar
Rozell, Mark J. 2002. Executive Privilege: Presidential Power, Secrecy, and Accountability. Lawrence: University Press of Kansas.Google Scholar
Rozell, Mark J., and Wilcox, Clyde. 1999. “Scandal and Government: Current and Future Implications of the Clinton Presidency.” PS: Political Science and Politics 32 (3): 554–57.Google Scholar
Sabato, Larry J. 1993. Feeding Frenzy: How Attack Journalism Has Transformed American Politics. New York: Free Press.Google Scholar
Sabato, Larry J., Stencel, M., and Lichter, S. R.. 2000. Peepshow: Media and Politics in the Age of Scandal. Lanham, MD: Rowman and Littlefield.Google Scholar
Schultz, Jeffrey. 1999. Presidential Scandals. Washington, DC: CQ Press.Google Scholar
Shear, Michael D. 2011. “No Guide to Survive a Scandal.” The New York Times, June 12. Google Scholar
Sonner, Molly W., and Wilcox, Clyde. 1999. “Forgiving and Forgetting: Public Support for Bill Clinton During the Lewinsky Scandal.” PS: Political Science and Politics 32 (3): 554–57.Google Scholar
Stanley, Harold W., and Niemi, Richard G.. 2011. Vital Statistics on American Politics, 2011–2012. Washington, DC: CQ Press.Google Scholar
Szasz, Andrew. 1986. “The Process and Significance of Political Scandals: A Comparison of Watergate and the ‘Sewergate’ Episode at the Environmental Protection Agency.” Social Problems 33 (3): 202–17.Google Scholar
Thompson, John B. 2000. Political Scandal: Power and Visibility in the Media Age. New York: Wiley.Google Scholar
Waisbord, Silvio. 2004. “Scandals, Media and Citizenship in Contemporary Argentina.” American Behavioral Scientist 47 (8): 1072–98.Google Scholar
Witcher, Russ. 2004. “The Media.” In Watergate and the Resignation of Richard Nixon: Impact of a Constitutional Crisis, eds. Jeffrey, Harry P. and Maxwell-Long, Thomas. Washington, DC: CQ Press.Google Scholar
Woodward, C. Vann. 1974. Responses of the Presidents to Charges of Misconduct. New York: Delacorte Press.Google Scholar
Zaller, John R. 1999. “Monica Lewinsky's Contribution to Political Science.” PS: Political Science and Politics 31 (2): 182–89.Google Scholar
Zeleny, Jeff. 2010. “Now It's Democrats Who Feel Sting of Scandal.” The New York Times, March 5. Google Scholar
Zorn, Christopher J. W. 2000. “Modeling Duration Dependence.” Political Analysis 8 (3): 367–80.Google Scholar
Figure 0

Figure 1Histogram of Presidential and Gubernatorial ScandalsNote: Gubernatorial data span 1972 to 2011. Presidential data span 1972 to 2009.

Figure 1

Table 1Gubernatorial Scandals Duration Models

Figure 2

Table 2Presidential Scandals Duration Models

Figure 3

Figure 2Cox Proportional Hazard Smoothed Functions: Chief Executive versus StaffNote: Cumulative hazard functions of chief executives versus staff. Kernel smoother (Gaussian) employed which averages values over a moving window of data.

Supplementary material: PDF

Rottinghaus supplementary material

Appendix tables 1 and 2

Download Rottinghaus supplementary material(PDF)
PDF 477.9 KB