Exceptional proff only

Agency Design, the Mass Media, and the Blame for Agency Scandals

ALEX I. RUDER Princeton University

I investigate whether formal presidential authority over agencies moderates attribu- tions of responsibility for agency activity. Using a survey experiment that manipulates news coverage of a major agency scandal, I find that providing details about formal presi- dential control in news coverage significantly increases attributions of responsibility to the president. I then explore broader questions of accountability by assessing mass media atten- tion to agencies. I use a unique data set of over 150,000 newspaper articles, matched to agency design and regulatory activity data sets, and find considerable variation in total coverage and coverage that mentions the president along with the agency. Combined, the results suggest that media coverage of agency design affects how voters allocate blame for agency actions and that political accountability is hindered by minimal media attention to a large number of agencies that wield significant regulatory powers.

Government agencies are responsible for the regulation and enforcement of laws that affect nearly every aspect of society. Their responsibilities range from the implemen- tation of major health care and financial reform to the criminal enforcement of antitrust laws and the collection of taxes. Yet democratic institutions rarely provide a direct elec- toral check on bureaucrats. Political accountability of agencies thus requires that voters credit or blame elected officials with formal authority over those agencies.

The process by which voters learn which elected officials are responsible for an agency is complicated by institutional structures that feature multiple political princi- pals, layers of bureaucracy, and complicated decision-making procedures often hidden from public view. A large body of research has thus focused on institutional design fea- tures that enhance the clarity of responsibility for agency actions (e.g., Kagan 2001). However, even with this research, we know little about how these features moderate voter attribution of blame. Are responsibility attributions for agency actions influenced by the institutional clarity of responsibility?

Alex I. Ruder is an assistant professor of public policy in the Department of Political Science at the University of South Carolina. His research interests include government agency structure and performance, public policy evaluation, and education and workforce policy implementation. Research for this article was conducted at Princeton University.

Presidential Studies Quarterly 45, no. 3 (September) VC 2015 Center for the Study of the Presidency

514

Proper allocation of blame and credit is necessary for democratic governance. Voters learn who is responsible for an action, form judgments about their performance in office, and then vote according to those judgments (Fiorina 1981; Key 1966). Concerns about the proper allocation of blame motivates many scholars to advocate for more presidential control over agencies. However, in order for increased presidential formal authority to enhance political accountability, voters must actually know that the president wields that authority.

How do voters learn about the complex institutional details concerning govern- ment agencies? As scholars have argued in the context of legislative institutions, voters learn about the institutional context through the mass media (Rudolph 2003). Unlike lawmaking, which has tallied roll-call votes and often a formal signing statement given by the president, most agency policy making occurs in a black box (Kagan 2001). Citizens, without media coverage, are unlikely to know which elected officials or bureaucrats are responsible for agency policy making or performance.

The informational role of the media has been shown to enable accountability by increasing voter knowledge of elected officials, which, in turn, creates incentives for those officials to be more responsive to their constituents (Prat and Str€omberg 2011; Snyder and Str€omberg 2010). In American politics, mass media provides voters with much informa- tion about political institutions and officials such as congressional actions (Arnold 2004) and the president’s public agenda (Barabas 2008). The media also serves Congress and vot- ers as a crucial watchdog over the actions of government agencies (Rosen 1998).

One challenge facing observational studies seeking to identify the influence of agency institutional context on responsibility attributions is controlling and measuring the information that voters encounter in the news. In this article, I use a survey experi- ment to identify the effect of agency design information on attributions of responsibility for an agency scandal. The survey experiment allows me to manipulate the information that respondents receive about agency design, while holding constant other potentially confounding details about the agency.

In the experiment, respondents read a real newspaper article about an agency scandal. Scandals provide an opportunity to study one of mass media’s most important functions— revealing government malfeasance to citizens. Officials themselves are unlikely to provide information to the public about misdoings in government. Absent these disclosures of wrongdoing, the mass media is the public’s only source of information concerning officials who are not representing their interests (Snyder and Puglisi 2011). Thus, out of all types of media coverage, coverage of agency scandals is arguably the most important place to convey information about political control of the bureaucracy. In addition, using news coverage allows me to manipulate an information source that is readily available to the public and is a primary source to voters about agency actions and institutional context.

In one experimental condition, I include agency-design information—the institu- tional context—that explains how the agency is structured to allow greater presidential influence over its operations. I provide respondents with agency-design details widely promoted by scholars as enhancing presidential control: presidential appointments and removal of agency leadership, the agency’s location in the president’s cabinet, and central- ized regulatory review by the White House Office of Management and Budget (OMB).

Ruder / BLAME FOR AGENCY SCANDALS | 515

The second experimental condition presents the same newspaper article but with no information on agency design.

I find that informing respondents about agency-design features that enhance presi- dential control significantly increases attributions of responsibility to the president. This result is consistent with presidential control theorists who argue that increasing presiden- tial formal authority over agencies enhances clarity of responsibility for voters.

Coverage of scandals, however, is only one part of the media’s role informing voters about public affairs. Simple media attention to the day-to-day activities of regulatory agen- cies enhances accountability by providing voters with crucial information about the numerous ways that agency activity influences them and society. To assess mass media attention to federal agencies, I investigate how much opportunity voters have to learn about agencies through the news. I collect over 150,000 newspaper articles about a large sample of U.S. federal agencies and examine how coverage varies by agency, agency-design features, and regulatory activity. I match news coverage to data about agency-design char- acteristics from Lewis’s work (2003, 2008) and regulatory review activity from the OMB. The analysis presents a mixed assessment of how well the news enables accountability of federal agencies. I demonstrate a high degree of heterogeneity across agencies in both the attention they receive in the news and the frequency with which journalists mention the president along with the agency. Some agencies with vast regulatory powers—such as the Food and Drug Administration (FDA)—receive relatively little coverage that mentions the president, despite the president’s significant powers over that agency.

The article is organized into the following sections. The first section reviews the institutional context of government agencies, paying particular attention to design fea- tures that increase the president’s formal authority over agencies. The second section describes the theoretical motivation and hypothesis. The third section presents the exper- imental design and results. The fourth section places the experimental results in a broader context of press coverage of government agencies. The fifth section concludes.

The Institutional Context: Agency Design

When creating agencies, elected officials balance political control with agency per- formance (Bawn 1995; Huber and McCarty 2004). The fundamental trade-off is that more political influence and control—through appointees rather than experts—reduce the autonomy of agencies to pursue what expert bureaucrats consider good public policy.

A second consideration is how much authority the president wields over an agency relative to Congress and the agency bureaucrats (Moe and Wilson 1994; Volden 2002). Theorists suggest that Congress resists presidential control of agencies when the presi- dent is ideologically opposed to Congress. Other works suggest that congressional con- trol obscures responsibility for agency actions (e.g., Lowi 1979) and that greater presidential control enhances clarity of responsibility and greater political accountability (Kagan 2001).

Battles over agency structure can lead to a variety of complex agency structures, but scholars simplify the empirical analysis of agency design by focusing on four broad

516 | PRESIDENTIAL STUDIES QUARTERLY / September 2015

categories of institutional design. Each category is defined by a set of institutional fea- tures that enhances or limits presidential control over the agency.1 In Table 1, I list the principal agency types identified by Lewis (2003, 2008), along with their design charac- teristics and examples of agencies.

These broad categories, of course, mask significant within-type heterogeneity. However, given their prominence in the literature, they provide a useful starting point to analyze differences in coverage. Political battles over agency design often focus on these broad categories. For example, congressional Republicans have continued to challenge the powerful Consumer Financial Protection Bureau and call for a change in leadership structure from a single agency director to a bipartisan regulatory commission (McCoy 2013). Another example involves the Federal Emergency Management Agency (FEMA), where critics, angered by perceived politicization of the agency, called for FEMA to be removed from the president’s cabinet and made into an independent agency (Cilluffo et al. 2009). Agencies located within the Executive Office of the President and the presi- dent’s cabinet feature the highest levels of presidential control. For cabinet departments such as Justice and Homeland Security, the president has significant control over budg- ets, personnel, and policy through regulatory review. Power over cabinet departments extends to agencies that belong within the cabinet departments. This subcabinet level includes agencies such as the FDA and FEMA.

The president also has these powers over executive agencies such as the Environ- mental Protection Agency (EPA). In terms of political insulation, executive agencies are not substantially different than cabinet-level agencies. The design feature that

TABLE 1 Insulation Characteristics of Different Agency Types

Design Description Example

Executive Office of the President

Located within the White House, under the direct control of the president’s Chief of Staff

Council on Environmental Quality, Office of Man- agement and Budget.

Cabinet/Subcabinet Officials serve at president’s will; agen- cies within cabinet face additional budgetary pressure from presidentially appointed secretaries.

Homeland Security, Food and Drug Administration

Executive Agency Located outside the cabinet structure. Budget line independent of cabinet departments. President still can remove officials at will.

Environmental Protection Agency

Regulatory Commission Led by multimember, bipartisan com- mission with staggered terms. For- cause removal of officials only.

Securities and Exchange Commission

Note: Types are listed in increasing order of insulation, with regulatory commissions being the most insulated from presidential control. Classification of agency types adapted from Lewis (2003).

1. I exclude government corporations because they are technically not regulatory or enforcement agencies, but rather independent corporations subsidized by the government.

Ruder / BLAME FOR AGENCY SCANDALS | 517

distinguishes these agencies is that their budget line in the federal budget is independent of any particular cabinet department budget.

Regulatory commissions are the agencies most insulated from presidential control through several mechanisms (Barkow 2010). First, commission members are removable only for cause and not simply at the president’s will. Second, a commission is not required to submit significant regulations to the OMB for cost–benefit analysis. Third, a commis- sion cannot be composed of commissioners from a single party; moreover, members serve staggered, fixed-length terms. All these factors reduce the president’s ex ante and ex post ability to control a commission’s policy and enforcement agenda.

Theoretical Expectations

Arguments for greater political control often focus on the president and features of the executive branch that increase the clarity of responsibility for voters (Kagan 2001; Lessig and Sunstein 1994). Even the judicial branch, especially since the Supreme Court’s Chevron decision, justifies judicial deference to executive agencies due to their being under the direction of an electorally accountable executive (Chevron U.S.A., Inc. v. Natural Resources Defense Council, 1984). However, more presidential authority would have little effect on electoral accountability if voters were unaware of this power.

The implicit assumption in this literature is that voter knowledge of agency design moderates attributions of blame. In other words, voters adjust their attributions of blame to be consistent with the formal degree of control a president holds over an agency. This behavior is consistent with arguments that voters are sophisticated in allocating blame, meaning voters consider factual details in order to form their opinions regarding responsibility.

The literature of blame attribution suggests several reasons why institutional con- text moderates blame attribution. One prominent theme in this literature focuses on the clarity of responsibility provided by features of political systems. Responsible party gov- ernment focuses on the ease with which voters can hold a single party responsible for gov- ernment actions (Schattschneider 1942). Economic voting studies find that features of political institutions, such as a greater share of control by governing parties, periods of unified government, budgetary power, or proposal power, increase clarity of responsibil- ity for economic outcomes (Alt and Lowry 1994; Duch and Stevenson 2008, 2013; Powell and Whitten 1993; Rudolph 2003). In comparative politics, the debate over pres- idential and parliamentary constitutions includes discussions of the clarity of responsibil- ity provided by presidential systems, which feature a single elected official responsible for governmental outcomes (Cheibub 2007; Mainwaring and Shugart 1997).

As Anderson (2000) shows, formal features of a political system moderate the strength of the relationship between economic conditions and blame or reward. This explanation of institutional complexity underlies more recent studies of blame attribu- tion in the wake of disaster, where scholars have focused on the complexity of responsibil- ity attributions in a federalist system (Arceneaux and Stein 2006; Gomez and Wilson 2008; Maestas et al. 2008; Malhotra and Kuo 2008).

518 | PRESIDENTIAL STUDIES QUARTERLY / September 2015

Many features of agency design are directly intended to increase or decrease the president’s responsibility over a given agency. Ruder (2014), for example, shows how bipartisan commission structures insulate the president from attributions of responsibil- ity. An open question, however, is how features that enhance presidential control provide greater clarity of responsibility to voters seeking to allocate blame in the wake of an agency scandal. Based on the discussions of agency design in the first section and institu- tional moderators of blame attribution in this section, I state the following hypothesis:

Centralized Control Hypothesis: Voters are more likely to blame the president when news includes information about formal institutional structures that give the president more control over an agency, compared to when news does not include information about formal institutional structures that give the president more control over an agency.

The first section suggests several features of agency design suitable to test the hypothesis. Commissions, for example, feature a bipartisan structure and for-cause removal of officials, both of which reduce presidential control. An alternative approach is to focus on agency-design features that increase presidential control. The advantage of this approach is that it addresses a key argument in favor of increasing presidential con- trol over agencies: the enhanced accountability derived through clarity of responsibility.2

Executive and cabinet agencies both feature several agency-design features that increase presidential control. In the analysis below, I focus on two features in particular: the appointment power and centralized control of presidential power through regulatory review. The appointment power is an Article II power granted to the president and remains one of the most important ways that the president manages the executive branch (Lewis 2008).

White House review of agency rulemaking is a feature of institutional design that has received considerable attention from presidential and public-law research (Acs and Cameron 2013; Croley 2003; West 2006). The OMB, serving the president, has been called a gatekeeper for agency regulations and annually reviews hundreds of significant agency regulations (Government Accountability Office 2003). The review of agency reg- ulations allows the White House, through the OMB, to ensure that agency regulations “reflect Presidential priorities and . . . ensure that economic and other impacts are assessed as part of regulatory decision-making" (OMB 2013).

Experimental Design and Results

To test the hypothesis regarding agency design and blame attribution, I conduct a survey experiment about the Minerals Management Service (MMS) oversight of the oil and gas industry, leading to the Deepwater Horizon explosion and oil spill in the Gulf of Mexico in April 2010. I choose this particular agency scandal for several reasons. Most importantly, the scandal represents a regulatory failure, with congressional hearings,

2. In related work, I show that commission features reduce attributions of responsibility in the con- text of antitrust cases (Ruder 2014).

Ruder / BLAME FOR AGENCY SCANDALS | 519

independent investigative reports, and media citing the agency’s lax regulatory approach and regulatory capture by the oil and gas industry. Scandals over regulatory failures address a core agency function with costly consequences for public health and safety. Fur- thermore, agency-design features, such as regulatory review, are specifically structured to provide the president with more or less power over regulation, allowing for a more plausi- ble examination of how these features moderate blame attribution.

The second reason I choose this incident concerns the widespread news coverage and public attention to the disaster. Less than a month after the Deepwater Horizon exploded, nearly 85% of U.S. adults knew a lot or some about the oil spill (60 Minutes/ Vanity Fair 2010). Americans were watching the news to learn about the oil spill. News coverage of the spill’s aftermath also frequently included President Barack Obama. In other words, the spill scandal is a situation in which the public was widely exposed to the scandal, the agency, and the president. This allows me to assess whether agency-design information can make any difference in attributions, given a media environment satu- rated with coverage of the president, agency officials, and state and local leaders trading blame for the disaster.3

The salience of the disaster also presents a potential weakness of this approach. Con- sidering that so many individuals have knowledge of the disaster, preformed opinions regarding blame attribution may not respond to new information about agency design. Another potential weakness concerns the elapsed time since the disaster and its after- math. Blame attribution may be most important to study as a scandal unfolds, not years after the event has occurred.

Even so, agency scandals and their aftermath can unfold over long periods of time. As this process unfolds, voters can attribute responsibility immediately after the disaster or retrospectively as more information about the disaster and the government’s response becomes available. As recently as March 2014, articles are still being published about health consequences and regulatory decisions linked to the 2010 oil spill (Hays 2014). Congressional investigations can drag on for months, and criminal investigations can take years to complete. Sources or newly discovered information can reveal previously unknown details that lead to new inquires into a scandal. These factors contribute to keeping agency scandals in the news long after such large disasters occur, suggesting that press coverage has the potential to continue to influence blame attributions.

Experimental Design

An observational study would require the unrealistic ability to measure exposure of individuals to news coverage with agency-design information and, controlling for other factors, estimate its effect on blame attributions. A more internally valid method is experimental manipulation of news articles, which has a long tradition in media research on framing effects.

3. A different approach is to study attribution in the context of less salient events where the public has little prior knowledge of the agency’s actions. Ruder (2014) takes this approach by focusing on routine agency actions and not scandals.

520 | PRESIDENTIAL STUDIES QUARTERLY / September 2015

I conduct the experiment using Amazon’s Mechanical Turk (MTurk) to recruit 305 respondents in April 2014.4 I take several standard precautions with regard to subject recruitment from MTurk. I warn subjects that they will not be paid for taking the survey more than once; I check respondent Internet protocol (IP) addresses to reduce the chance that anyone took the survey twice. I also warn respondents that they will not be compen- sated if they too quickly read the prompts and answer questions, which suggests guessing and not taking the experiment seriously. I also limit the survey to respondents with over a 97.5% task approval rate, which is an indicator of the reliability and quality of the MTurk respondent. The average survey completion time was five minutes and twenty-eight sec- onds. After respondents completed the survey, they received 60 cents compensation.

To gauge attribution of responsibility, I first ask respondents to read a real news article about the resignation of MMS Director Elizabeth Birnbaum amid criticism of the agency’s lax industry oversight.5 After reading the article, I ask respondents to “Please indicate how much responsibility you think each of the following has for the U.S. Miner- als Management Service’s role in the Deepwater Horizon disaster.” I present respondents with a list of four entities: the Obama administration, the Bush administration, Con- gress, and the agency leadership. Respondents select A Lot, Some, A Little, or None.

I include both the Bush and Obama administrations because the disaster happened only 16 months into the Obama administration; reasonably, individuals could conclude that Bush administration policies led to the disaster. In fact, polling conducted at the time suggests that individuals felt Bush administration policies were a more likely con- tributor to the regulatory failure. Among U.S. adults, 18% blame the policies of the Bush administration for the conditions that led to the spill, compared to only 3% for the Obama administration (CBS News/New York Times Poll 2010). This difference between George W. Bush and Barack Obama is particularly large among Democrats and inde- pendents, but even Republicans were more likely to blame the Bush administration.

I randomly assign respondents to one of three experimental conditions. In the first condition, respondents view no news article and simply attribute responsibility for the disaster. This condition provides baseline estimates of attributions of responsibility in the absence of news coverage. In the second condition, which I call the No Presidential Control condition, respondents read an article that makes no mention of elected officials or agency-design information. It simply discusses the resignation and the agency’s poor enforcement record. In the third condition, the Presidential Control condition, I add the following sentences near the beginning of the article:

MMS is under the direction of Secretary of the Interior Ken Salazar, who has led the cabinet-level department as part of the Obama administration. According to the U.S Office of Management and Budget, since 2009 alone the Obama administration has reviewed and approved several MMS regulations related to deepwater oil exploration.

4. Berinsky, Huber, and Lenz (2012) discuss the benefits and drawbacks of the MTurk subject pool. They show that the MTurk subject pool is comparable to other student and adult convenience samples and is often more representative of the U.S. adult population. However, the samples tend to be younger, more edu- cated, and more liberal than U.S. population averages.

5. Article used based on Kauffman (2010).

Ruder / BLAME FOR AGENCY SCANDALS | 521

In addition, the article includes the following text: “Elizabeth Birnbaum, who was appointed director of the Minerals Management Service in June 2009, submitted her res- ignation to Interior Secretary Ken Salazar.” In the third condition, I add “by President Obama” after “. . . was appointed director of the Minerals Management Service.”

The third condition includes three factors related to the hypothesis stated in the second section: the president appointed the leader, MMS is a cabinet-level agency under the direction of a cabinet secretary, and the OMB reviews and approves the agency’s regu- lations related to offshore drilling. The theoretical expectation is that this added informa- tion increases attributions of responsibility by clarifying the degree of the president’s responsibility over the agency. The articles and survey instrument are available in Appen- dix S1 and Appendix S2, respectively.

Results

The ordinal nature of the data in the experiment suggests two alternative strategies for statistical analysis. I use a simple nonparametric test for ordinal variables, the one- sided Mann-Whitney U test, which tests the null hypothesis of equality of two groups of responses against the alternative hypothesis that one group is greater than the other. The second approach would be to use ordinal logistic regression, but I choose the nonparamet- ric alternative because I include no pretreatment covariates. The hypothesis in the second is strictly about agency-design information; it is not focused on heterogeneous treatment effects across groups, which I leave for future research. The results presented below are robust to different empirical strategies, such as a chi-square test or transforming the data into dichotomous high and low attribution groups and performing an equality of propor- tions test. These additional results are available in Appendix S1.

The results presented in Table 2 show the percent of respondents in each response category across the two experimental conditions. In Panel A, I show results from respond- ent attributions to the Bush administration. The one-sided Mann-Whitney U test cannot reject the null hypothesis that attribution levels are the same in both experimental condi- tions. The information about agency design had no effect on increasing the attributions to the Bush administration. Attributions appear slightly higher in the No Presidential Control condition, though the alternative one-sided and two-sided Mann-Whitney tests cannot reject the null hypothesis of equal response distributions.

The results for the Obama administration provide a stark contrast to the results for the Bush administration. The difference in responses across Presidential Control and No Presidential Control conditions is large and statistically significant. In the No Presidential Control condition, 17% of respondents said the Obama administration had None as a degree of responsibility. In the Presidential Control condition, only 5% of respondents chose None. Moreover, only 11% of respondents chose A Lot in the No Presidential Control condition, while 23% chose A Lot in the Presidential Control condition.

One way to simplify the presentation of these results is to recode the four attribu- tion levels into a single continuous measure that ranges from 0 (no attribution) to 1 (the highest degree of attribution). I recode none to 0.00, a little to 0.33, some to 0.66, and a lot to 1.00. The implied assumption of this recoding is that the response values are

522 | PRESIDENTIAL STUDIES QUARTERLY / September 2015

equidistant from each other. Once recoded, I calculate the average value of this variable and call it mean attribution. The measure is presented in the final column of Table 2. For the Bush administration, the difference in mean attribution between the two experimen- tal conditions is 20.07 (0.51–0.58). In contrast, the difference in mean attribution for the Obama administration is 0.14 (0.62–0.48).

Blame of Other Officials

The survey also asked respondents to attribute responsibility to other officials with formal responsibility over the agency. These other officials are Congress and the agency leadership. These results provide an assessment of how respondents change blame attri- butions to other officials when provided with information about presidential control. For example, learning about presidential control over the agency may reduce the respondents’ propensity to blame Congress or the agency leadership. Alternatively, if respondents do not treat responsibility for agency failures as zero-sum, then respondents may not change their attributions to Congress and the agency leadership when presented with informa- tion about presidential control.

The results for Congress and the agency leadership are in Table 3. For simplicity, I present only the mean attribution scores. The mean attribution scores are consider- ably higher for the agency leadership (column 2) than for Congress (column 3), or for the Bush and Obama administrations, as presented in Table 2. The Presidential Control treatment has a small negative effect reducing attributions of responsibility to these two actors, though these results are only significant at the 0.10 level. This result sug- gests that respondents allocate blame away from these officials and toward the administration.

TABLE 2 Results from Experiment Testing the Centralized Control Hypothesis: Percent Respondents in Each Cell by Experimental Condition

A. Bush Administration

Condition A Lot Some A Little None Mean Attribution

No Presidential Control 17% 48% 29% 7% 0.58 Presidential Control 18% 37% 28% 18% 0.51 Mann-Whitney Test Statistic: U 5 4,550, p 5 0.92

B. Obama Administration

Condition A Lot Some A Little None Mean Attribution

No Presidential Control 11% 40% 33% 17% 0.48 Presidential Control 23% 48% 25% 5% 0.62 Mann-Whitney Test Statistic: U 5 6437, p < 0.01

Note: Panel A shows results from the Bush administration response. Panel B shows results from the Obama administration response. One-sided Mann-Whitney test statistics and p-values shown. n 5 202, with 101 respondents assigned randomly to each experimental condition.

Ruder / BLAME FOR AGENCY SCANDALS | 523

Treatment Comparison to Baseline Attributions

The results presented in the sections above compare the effect of information about presidential control relative to a baseline in which respondents read a newspaper without information on presidential control. This experimental design tests the hypothesis as described in the second section. An alternative strategy is to examine how information on presidential control changes attributions of responsibility relative to a baseline group who receives no experimental intervention. While not a test of the centralized control hypothe- sis, this strategy illuminates how respondents allocate responsibility absent media coverage and how different types of media coverage change those allocations of responsibility.

In Table 4, I present results of both the No Presidential Control and Presidential Con- trol conditions, when compared to a baseline condition where I simply ask respondents to attribute responsibility to the four officials. Panel A shows the mean attribution scores for the No Presidential Control condition and the baseline No Media Baseline condition. When asked to attribute responsibility for each official, respondents in the No Media Baseline condition are more likely to blame the agency leadership than Congress or the presidential administrations.

Respondents allocate similar degrees of attribution to the Bush and Obama adminis- trations (0.52 and 0.54, respectively), which differs from results of polls conducted after the disaster. The results presented here are not strictly comparable as there was extensive media coverage of the disaster and its response in 2010 and several years have passed since the event. The relatively equal levels of attribution could reflect uncertainty over the details of the disaster, the federal agency in charge, or the exact dates involved in the scandal.

Presenting respondents with a news article without presidential control information has the largest change on attributions of responsibility to the agency leadership (0.69 to 0.86). This large change is most likely driven by the content in the news article, which focuses on the resignation of agency leadership. The No Presidential Control condition shows no statistically significant differences in attribution of responsibility for the remaining offi- cials. Respondents were slightly more likely to blame the Bush administration and slightly less likely to blame the Obama administration when compared to the No Media Baseline condition. This pattern is more consistent with the national polling results from 2010, in which the Bush administration received more blame than the Obama administration.

Panel B in Table 4 shows results for the Presidential Control condition and the No Media Baseline condition. As with the results in Panel A, significantly more respondents

TABLE 3 Allocation of Blame to Other Officials

Condition Agency Leadership Congress

No Presidential Control 0.86 0.63 Presidential Control 0.82 0.56 p-value 0.058 0.054

Note: Results show mean attribution scores to both agency leadership and Congress, by experimental con- dition. p-values are for one-sided Mann-Whitney test of the alternative hypothesis that attribution in the Presidential Control condition is less than attribution in the No Presidential Control condition.

524 | PRESIDENTIAL STUDIES QUARTERLY / September 2015

allocate responsibility to the agency leadership after reading the news article when com- pared to respondents in the No Media Baseline condition (from 0.69 to 0.82). The results for the presidential administrations are substantively comparable to the results presented in Table 2: the Presidential Control treatment causes a statistically significant increase in attributions of responsibility toward the Obama administration. There is no significant difference in attributions to the Bush administration.

Summary of Experimental Results

The results in Table 2 provide support for the centralized control hypothesis: infor- mation in the news about the president’s centralized control over the agency increases attributions of blame to the president, relative to news that omits that information. The results support the hypothesis only with regard to the Obama administration.

What explains the absence of an effect with regard to the Bush administration? One likely explanation involves the sophisticated allocation of blame for the conditions that led to the disaster. Without a clear understanding of presidential formal authority over the agency, respondents are more likely to blame eight years of Bush administration regulatory policy and the failures that occurred under that administration (which are mentioned in the experiment’s news article). However, once informed about the influence the Obama administration wields over the agency—through appointment of the agency leadership, removal of the agency leadership, and regulatory review—respondents were more likely to allocate blame to the Obama administration. Respondents, however, may also be acting myopically (Bartels 2008) by irrationally allocating more blame to the cur- rent administration than it actually deserves.

TABLE 4 Results from Experiment Examining Treatments Compared to No Media Baseline Condition

A. No Presidential Control Comparison

Condition Agency

Leadership Congress Bush

Administration Obama

Administration

No Media Baseline 0.69 0.61 0.52 0.54 No Presidential Control 0.86 0.63 0.58 0.48 p-value <0.01 0.74 0.24 0.19

B. Presidential Control Comparison

Condition Agency

Leadership Congress Bush

Administration Obama

Administration

No Media Baseline 0.69 0.61 0.52 0.54 Presidential Control 0.82 0.56 0.51 0.62 p-value <0.01 0.18 0.80 0.04

Note: All cells contain mean attribution to the president. One-sided Mann-Whitney test statistics and p-values shown. 103 respondents assigned to No Media Baseline condition.

Ruder / BLAME FOR AGENCY SCANDALS | 525

The change does not mean that respondents place no blame on the Bush adminis- tration, but that respondents shift toward more blame on the Obama administration. More information about presidential control clarifies for respondents how much responsi- bility the Obama administration had for the agency; before respondents were exposed to the agency-design information, they possibly had a vague or inaccurate notion of presi- dential responsibility for the agency. Knowledge of the institutional context, in other words, enhanced clarity of responsibility.

The News and Federal Agency Accountability

Results in the preceding section show that news coverage that emphasizes the agency institutional context increases attributions of responsibility to the president dur- ing a major agency scandal. Providing details about the institutional context is not the only way that journalists enable the political accountability about agencies. Journalists, for example, can use different frames to describe policy (Bolson, Druckman, and Cook 2014; Ruder 2014), slant coverage toward a particular ideology (Gentzkow and Shapiro 2010; Groseclose and Milyo 2005), or emphasize some issues while ignoring others (Larcinese, Puglisi, and Snyder 2011; Puglisi 2011). Scandals, too, are not the only events that attract journalists to agencies. Agencies issue and enforce major regulations across the public policy domain. It is important to assess whether the public gets information about these day-to-day agency activities and does not learn about agencies only when the press covers relatively infrequent agency scandals.

A minimal requirement for accountability is simply covering agencies so that the voting public receives information about agency policy making. In the congressional con- text, Arnold (2004) shows that media coverage affects voter awareness of representatives and challengers in elections. Similarly, Snyder and Str€omberg (2010) show that voters who live in areas with less press coverage about their congressional representative are less likely to recall their representative’s name or evaluate their performance in office. Accord- ing to their findings, coverage impacts more than voter knowledge; it affects the respon- siveness of elected officials to their constituencies (Snyder and Str€omberg 2010).

To understand the information available to voters about federal agencies, it is first important to know which agencies receive extensive coverage in the press and which agencies do not. Some agencies, such as the MMS, make significant policy-making deci- sions yet infrequently make the news. Other major agencies, such as the FDA, are fea- tured far more frequently in the press. This variation in coverage has implications for whom the public holds accountable for federal agencies, what the public knows about federal agencies, how the public approves or disapproves of an agency’s performance, and how agency officials conduct policy.

Gallup, for example, regularly polls the American public to ask their approval of select federal agencies (Gallup 2013). An important question is how the public obtains the information necessary to make these evaluations. The public has direct experience with some of the agencies Gallup uses in the survey, such as the Internal Revenue Service, allowing the public to use that direct experience in their evaluations. The public is far

526 | PRESIDENTIAL STUDIES QUARTERLY / September 2015

less likely to have direct interaction with many other federal agencies in the survey, such as the Central Intelligence Agency. This makes press coverage of these agencies an impor- tant provider of information about agency performance and official oversight.

For example, between 1987 and 2011, MMS approved over 21,000 applications to drill in the Gulf of Mexico. Between 1987 and 2007, the agency proposed 114 rules to the OMB; three of those rules were economically significant. The public had limited opportunity to learn of this activity through the news. The dark bars in Figure 1 show the number of news articles published annually about the MMS. Overall, the agency receives fairly little coverage. The median number of articles published annually is only three. The maximum number of articles published (86) is in 2010, the year of the oil spill.

6 The second highest annual total of articles, in 2006, was largely driven by another

scandal, which concerned nonpayment of oil royalties. I now present patterns of newspaper coverage of U.S. federal agencies. I show that

considerable heterogeneity exists in the extent to which journalists cover agencies and in the frequency that journalists mention the president along with the agency—a minimum requirement to connect the agency’s policies to the president. I then focus on regulators

FIGURE 1. Annual Articles about the Minerals Management Service. Sources: New York Times Annotated Corpus and ProQuest Newspapers.

6. Additional years for the Times acquired from ProQuest Online Newspapers. See Appendix S3 and Appendix S5 for search terms and results and search parameters for statistics on oil leasing. Oil leasing data from online database of Bureau of Ocean Energy Management. Review data from OMB.

Ruder / BLAME FOR AGENCY SCANDALS | 527

whose regulations are subject to presidential oversight through OMB regulatory review. I find that even the nation’s largest regulators rarely receive coverage that features the president. The goal of this analysis is not to explain the variation in coverage but to pro- vide descriptive patterns of media coverage of federal agencies. The knowledge of these patterns itself is an important step toward understanding accountability of these agencies and formulating questions for future research.

Aggregate Newspaper Coverage of the Bureaucracy

Newspapers are a major source of news for the public. Approximately 25% of U.S. adults use national newspapers for news, while 69% use local newspapers (Harris Interac- tive 2010). Editors and journalists at these local newspapers rely on influential papers such as the New York Times to set the agenda (Gans 1979; Wilhoit and Weaver 1991).

The data source for this analysis is the New York Times Annotated Corpus (Times), which includes nearly all articles published by the Times between 1987 through 2007 (Sandhaus 2008). Unfortunately, the corpus is limited to these years, but its coverage spans four presidential administrations, including the complete Clinton administration and seven years of the Bush administration. This coverage limits the potential for finding results that are idiosyncratic to a single president or biased by period–specific events such as a presidential election.

The Times, given its influence and focus on politics and regulation, is ideally suited to study agencies. Scholars have frequently used the Times to study presidential politics (see Gilens, Vavreck, and Cohen 2007). For decades, the Times has maintained its position as the country’s leading paper. Presidents and national politics receive extensive coverage in the Times (Cohen 2008). I also find that the Times is not unique in its extensive cover- age of regulatory agencies. Other major papers, such as the Wall Street Journal and the Washington Post, devote similar amounts of coverage to federal agencies.

I create a list of nearly every agency in the U.S. federal government using three sources: the U.S. Government Manual, the official Web site of the White House, and Lew- is’s agency-design data sets.7 To identify subcabinet agencies, I use two sources. First, I use the OMB’s database on rulemaking activity. This list identifies every subcabinet agency or office that has issued a rule or regulation. I supplement this list with Lewis’s data set. While this list misses some obscure subcabinet agencies or offices, it still includes over 150 entities. All agencies used in the analysis are listed in Appendix S1.

After creating the list of agencies, I search the Times corpus for news articles that mention any of these agencies. I then organize the articles into five separate document

7. I use two data sets from Lewis’s work. The Politicization Over Time data includes all agencies included in the Office of Personnel Management Central Personnel Data File, which includes nearly all agen- cies in existence. See Lewis’s data set codebook for the few exceptions, available on Professor Lewis’s Web site, https://my.vanderbilt.edu/davidlewis/data/ (accessed May 3, 2015). For additional coverage, I also use Lewis’s Administrative Agency Insulation data, which include all agencies created during 1946-97, as listed in the U.S. Government Manual, http://www.gpo.gov (accessed May 3, 2015). I exclude the military branches, primarily because the Constitution places these agencies under presidential control, which reduces the ambi- guity of control and the public’s reliance on media for knowledge of who is in charge.

528 | PRESIDENTIAL STUDIES QUARTERLY / September 2015

corpora corresponding to institutional design of the agency.8 I organize by the broad cat- egories of agency design, rather than substantive policy area, to highlight how coverage varies across different institutional contexts. The number of articles and agencies col- lected is listed in Table 5.

Table 5 shows that agencies, at this aggregate level, receive extensive coverage in the newspaper. Three categories—cabinet departments, subcabinet agencies, and regula- tory commissions—each contain nearly 50,000 articles published. The table also shows considerable variation in coverage across agency type. The last column reports the first, second, and third quartile of the annual number of articles published about agencies within each agency type. For example, the median article count for cabinet departments is 110. That high value is a stark contrast to the subcabinet, where 50% of annual agency observations have a value of zero. Many of these subcabinet agencies and offices are small (measured by staff and budget) and have a minimal regulatory impact on society (meas- ured by regulations issued), which is one explanation for the limited news coverage. In addition, coverage of the subcabinet agencies often refers only to the parent cabinet department, although the action was taken by a subcabinet agency. Thus, there are fewer opportunities for these agencies to receive press coverage.

Newspaper Coverage of U.S. Regulatory Agencies

Table 5 counts articles from a wide variety of agencies. Some of these agencies have neither regulatory nor enforcement authority, leading the aggregate numbers in Table 5 to mask how coverage varies by an agency’s regulatory impact on society. An alternative approach is to focus on regulatory agencies only, that is, focus on agencies that promote social and economic regulations on American businesses and individuals (see Coglianese and Howard 1998 for analysis of EPA regulatory coverage). Accountability of regulatory agencies is particularly important given the significant benefits and costs associated with agency regulations. In 2013, the president’s regulatory budget eclipsed $58 billion; those funds pay salaries of over 290,000 agency personnel (Dudley and Warren 2012). Also in

TABLE 5 Number of Articles That Mention Department, Agency, or Office by Agency Type

Institution Type Agencies Articles Quartiles of Annual Articles per Agency

Executive Office of the President 7 7,298 3, 15, 79 Cabinet Department 15 55,456 61, 110, 195 Subcabinet Agency or Office 154 56,574 0, 0, 3 Executive Agency 16 20,471 2, 17, 53 Regulatory Commission 27 48,673 0, 12, 60

Note: The number of department, agency, or office within each type is also listed. Last column reports the first, second, and third quartile of the number of articles published annually for all entities within type.

8. The corpora are not mutually exclusive.

Ruder / BLAME FOR AGENCY SCANDALS | 529

2013, the OMB reviewed 418 proposed agency regulations; 104 of those regulations OMB deemed economically significant.9 Since 1987, the OMB has received over 28,000 agency regulations.10

Figure 2 shows boxplots of annual newspaper coverage of a large sample of U.S. reg- ulatory agencies.11 The figure shows high variation in press coverage across and within agencies. The Federal Reserve, the Securities and Exchange Commission, and the FDA receive the most coverage. Other agencies, such as the Consumer Product Safety Commis- sion, the Interstate Commerce Commission (abolished in 1995), and the Transportation Security Administration receive relatively little attention in the news.

FIGURE 2. Number of Annual Articles by Regulatory Agency. Note: Agencies sorted by median annual number of articles. Difference in group means significant at the .01 level according to analysis of variance test: F 5 65.7.

9. According to Executive Order 12866, any rule that meets the following criteria is economically significant: (1) has an annual economic impact of over $100 million, (2) creates a serious inconsistency with other agency policy, (3) alters the budgetary impact of entitlements, and (4) raises novel legal or policy issues.

10. Figures from http://www.reginfo.gov. 11. I include all regulatory agencies used in Lewis (2008). Lewis obtained the list, which I independ-

ently verified, from the Washington University at St. Louis’ Weidenbaum Center on the Economy, Govern- ment, and Public Policy at http://wc.wustl.edu/regulatory_reports.

530 | PRESIDENTIAL STUDIES QUARTERLY / September 2015

When journalists cover agencies, they enable the public to learn about agency actions. An additional step toward political accountability involves journalists connect- ing agencies’ actions to the elected officials who oversee those agencies. Given the large size of the data set, and the research focus on general media attention to the president and the agency, I limit the analysis to a simple measure: the share of articles about an agency that mentions the president and the agency together. This measure captures media atten- tion to the broad range of issues that involve the president and federal agencies: appoint- ments, political controversy, and public discussions of agency policy. The advantages to this approach are ease of replication and low cost in terms of human coding. The principal disadvantages are a lack of context and a risk of counting incidental mentions of the presi- dent’s name that make no connection between the president and the agency.

In Figure 3, I show first, second, and third quartiles of the annual share of articles about each regulatory agency that mentions the president’s name.12 Observations are

FIGURE 3. Annual Share of Published Articles That Mention the President, by Regulatory Agency. Note: Agencies sorted by median share of articles that mention the president. Observations weighted by annual number of articles that mention the agency. Weighted analysis of variance test rejects equality of group means at the .01 level: F 5 30.9.

12. To be able to calculate the share of articles, I remove agencies with less than five observations, eliminating zeros and agencies with very few observations.

Ruder / BLAME FOR AGENCY SCANDALS | 531

weighted by the annual number of articles that feature the agency. Agencies that receive the most coverage do not necessarily receive the most coverage that mentions the presi- dent. The agencies with the greatest share of articles that mention the president are the Federal Election Commission and the EPA. The EPA is prominent in both the total cov- erage it receives and the share of coverage that mentions the president. In contrast, the Securities and Exchange Commission and the National Transportation Safety Board receive relatively little coverage that features the president.

Political accountability requires accurate attributions of responsibility. Figure 3 reveals one possible way that the accurate assignment of blame can be undermined by news coverage of agencies. Specifically, even some agencies that are relatively insulated from presidential control receive a large share of articles that mention the president. The Federal Reserve, for example, is arguably the most politically insulated regulatory com- mission, but news coverage of the Federal Reserve still mentions the president in over 20% of articles per year. Approximately 11% of these articles that mention the Federal Reserve and the president suggest presidential influence to readers by mentioning appointments.13

Coverage of the Federal Reserve can take many forms, each with different implica- tions for voter attribution of responsibility. Many of these articles are about general eco- nomic policy; the president and the Federal Reserve wield significant influence in economic policy and are likely to appear in articles together due to each actor wielding significant influence over economic affairs. However, in some of these articles, by discus- sing the political issues surrounding Federal Reserve policy, journalists can be using a political frame to discuss the Federal Reserve. Political frames, when compared to more policy-focused frames, can reduce support for policy and influence attributions of respon- sibility (Bolson, Druckman, and Cook 2014; Ruder 2014). In each case, if journalists exclude agency-design information from articles about these independent agencies, then they may fail to increase the institutional clarity of responsibility for some readers. Read- ers may then misjudge the president’s authority over the agency and, ultimately, improp- erly allocate blame for the agency’s action.

Newspaper Coverage of Agencies Subject to OMB Regulatory Review

Figure 3 includes agencies whose regulations are not subject to formal OMB review. One of the main justifications for OMB review is that voters, when seeking to assign blame, will know that the president has set an agency’s regulatory agenda. The White House itself seems to accept this view, stating, as quoted in the second section, that OMB review is intended to ensure that regulations reflect presidential priorities. How- ever, for OMB regulatory review to enable accountability, voters must know which regu- lations are subject to OMB review and thus reflect presidential priorities.

How can the news fulfill this function of informing voters about the president’s control over regulatory agencies? One possibility is that every news article about a rule or regulation mentions that the president has review authority. At the other extreme, news

13. See Appendix S4 for searches to obtain these numbers.

532 | PRESIDENTIAL STUDIES QUARTERLY / September 2015

could never mention the president’s centralized control powers and instead focus on the policy details and interest group activity surrounding the rule.

The experiment described in the third section reveals how this information can influence attributions of responsibility. Recall that the MMS is a cabinet-level agency whose leaders are appointed and removed at will by the president. The treatment men- tions the president’s review of agency regulations, which is one component of the agency- design information that increases attributions of responsibility to the president.

To assess the actual coverage patterns, I use OMB regulatory reports to calculate the 10 agencies with the most economically significant regulations reviewed between 1981 and 2007.14 I focus on the top 10 regulation producers in order to simplify the pre- sentation of results and focus on the agencies with the largest regulatory impact on soci- ety. In addition, to more accurately represent coverage that focuses on regulations, I limit the news collection to articles that mention rule, regulation, or rulemaking. I list these agencies and the share of articles that mention the president’s name in Figure 4.

Once again, the figure reveals considerable heterogeneity in coverage across agencies. Coverage patterns match neither of the extreme conditions stated above. For seven of these agencies, less than 50% of news coverage mentions the president. Only three regulators— the Department of Health and Human Services, the Small Business Administration, and the Centers for Medicare and Medicaid Services—receive coverage that mentions the presi- dent more than 50% of the time. Meanwhile, coverage of the powerful FDA mentions the president approximately 25% of the time. Voters are thus often reading about many promi- nent regulators as though they operate independently from the president and his influence.

It is notable that agencies, such as the FDA, receive such a low frequency of cover- age that mentions the president. FDA regulations and decisions on drug approvals have significant consequences for industry and personal health and well-being. The White House maintains authority over the FDA, including the power to overrule approvals of drugs and medical devices. Despite these controls, the media infrequently associate this agency with the president and, as Carpenter (2011) noted in a New York Times op-ed, presidents infrequently interfere with it. Only once in FDA history has a presidential administration overruled an FDA decision. Therefore, the infrequent appearance of the president in FDA agency news may reflect a hesitancy to politicize technocratic agency decision making (see Bawn 1995). That power to overrule the agency, however, is always present and can influence the agency’s actions on some level.

Discussion

This article builds on work that examines the influence that institutional context and clarity of responsibility have on retrospective evaluations of elected officials. Many

14. The first year available in OMB’s .xml reports at time of access is 1981. I collapse two subcabinet agencies within the Agriculture Department to just the Agriculture Department. While the Centers for Medicare and Medicaid Services (established 2001) and the FDA are part of the Department of Health and Human Services, I list them separately because OMB lists those subcabinet agencies issuing the regulations, and they receive prominent press coverage, independent of the parent cabinet department.

Ruder / BLAME FOR AGENCY SCANDALS | 533

works in this literature study the relationship between institutional clarity of responsibil- ity and economic voting (Alt and Lowry 1994; Anderson 2000; Duch and Stevenson 2008; Powell and Whitten 1993; Rudolph 2003); others focus more broadly on blame allocation given the institutional complexity of a multilayered federalist government (Arceneaux and Stein 2006; Gomez and Wilson 2008; Maestas et al. 2008; Malhotra and Kuo 2008).

There is a lack of empirical work that focuses on the comparative analysis of agency-design features that influence how voters perceive the relative influence of various elected officials. This absence is puzzling for several reasons. First, unitary executive theo- rists in public law argue that agency-design features that enhance presidential control over agencies also increase clarity of responsibility (Calabresi 1995; Kagan 2001). Second, political battles over agency design often center on how much power over an agency the president wields relative to Congress (Lewis 2003; Moe and Wilson 1994). Third, a long tradition in bureaucratic politics posits that legislators delegate authority to agencies, whose complex structures obscure responsibility, in order to mask their own involvement

FIGURE 4. Average Annual Share of Published Articles That Mention the President, by Regulatory Agency. Note: Only 10 agencies with the most economically significant rules proposed are shown. Agencies sorted by median share of articles that mention the president. Observations weighted by annual number of articles that mention the agency. Weighted analysis of variance test rejects equality of group means at the .01 level: F 5 7.73.

534 | PRESIDENTIAL STUDIES QUARTERLY / September 2015

and avoid blame (Fiorina 1982; Lowi 1979). Fourth, one of the most significant develop- ments in the expansion of executive power—increased centralized presidential control over agencies through regulatory review—is fundamentally an agency-design feature meant to enable the president to align agency policy with administration priorities.15 In summary, scholars pay great attention to agency-design features and their implications for clarity of responsibility, but we know little about how the public actually learns of or perceives these differences.

In this article, I show that agency-design features that enhance formal presidential authority can also lead voters to increase attributions of responsibility for an agency’s actions to the president. The findings suggest that these agency-design features lead to greater institutional clarity of responsibility. Informing voters of the institutional con- text, or the formal structures that enhance or insulate the agency from the president, allows voters to form more accurate retrospective judgments and assign responsibility correctly.

Forming retrospective evaluations for agency scandals, and agency actions more generally, is particularly challenging. Presidents who are no longer in office still exert control over agency policy through appointments of officials such as commission mem- bers. For agency actions, voters must determine how much responsibility the current administration bears relative to the previous administration. In this article, respondents in the survey experiment are more likely to blame the Obama administration for the conditions that led to the Deepwater Horizon disaster. It is not clear that this is the strictly rational reaction; the Bush administration implemented years of regulatory actions that formed the U.S. energy policy in the Gulf of Mexico. With respect to this agency scandal, respondents may be myopic (Bartels 2008), believing a new administra- tion bears more responsibility than it deserves. At the same time, respondents may be acting rationally, because President Obama appointed the agency leader who was in charge during the disaster and had approved many regulations related to offshore oil drilling.

These results highlight the need for scholars in executive and bureaucratic pol- itics to consider an important point when discussing the increasing trend of central- ized presidential control: voters lack perfect information about the president’s formal authority. Thus, White House approval or disapproval of an agency regula- tion can lead to greater accountability only if voters know that the agency’s regula- tions are subject to review. The news, as a key source of political information for voters, is but one way voters can learn about the institutional context when forming retrospective judgments.

The results add an additional consideration to discussions of blame attribution in the wake of a scandal. Certainly a complex federal system, a multilayered bureaucracy, and elites shifting blame all obscure responsibility. However, the institutional context of agencies themselves can enhance the clarity of responsibility during an agency scandal. Formal authority resolves some of the ambiguity regarding responsibility.

15. West (2005, 78), for example, calls regulatory review “arguably the most significant constitu- tional extension of executive power in decades.”

Ruder / BLAME FOR AGENCY SCANDALS | 535

One line of future research should consider how the president’s internal staff man- ages agency crises in the wake of a scandal. Previous work that studies presidential involvement in scandals and their lasting impact on the president (e.g., Hacker, Giles, and Guerrero 2003; Rottinghaus and Berenznikova 2006) can be expanded to consider how presidents manage blame when agencies are insulated from their influence.

An additional line of future research should focus on how partisan identifications moderate attributions of responsibility in the context of bureaucratic politics. As dis- cussed above, scholars studying legislative institutions have found that partisan rationali- zations moderate the effect of the legislative institutional context on attributions of responsibility for economic outcomes.

There is at least one reason to expect that the moderating effects of partisanship may differ in the context of bureaucratic politics. For example, agencies possess their own policy preferences, which have been scaled onto the liberal/conservative ideological spec- trum (Bertelli and Grose 2011). Voters may consider the ideological distance of an agency from the president when deciding whether to attribute responsibility to the presi- dent if they believe there is a systematic relationship between the degree of political con- trol and the ideological distance between an agency and the president.

References

60 Minutes/Vanity Fair. 2010. “Gulf Oil Spill/Technology.” Poll conducted by CBS News, May 6- 9. http://www.ropercenter.uconn.edu/ipoll-database/ (accessed May 3, 2015).

Acs, Alex, and Charles M. Cameron. 2013. “Does White House Regulatory Review Produce a Chill- ing Effect and ‘OIRA Avoidance’ in the Agencies?” Presidential Studies Quarterly 43 (3): 443-67.

Alt, James E., and Robert C. Lowry. 1994. “Divided Government, Fiscal Institutions, and Budget Deficits: Evidence from the States.” American Political Science Review 88 (4): 811-28.

Anderson, Christopher. 2000. “Economic Voting in a Comparative Perspective.” Electoral Studies 19 (2): 151-70.

Arceneaux, Kevin, and Robert Stein. 2006. “Who Is Held Responsible When Disaster Strikes? The Attribution of Responsibility for a Natural Disaster in an Urban Election.” Journal of Urban Affairs 28 (1): 43-53.

Arnold, R. Douglas. 2004. Congress, the Press, and Political Accountability. Princeton, NJ: Princeton University Press.

Barabas, Jason. 2008. “Presidential Policy Initiatives: How the Public Learns about State of the Union Proposals from the Mass Media.” Presidential Studies Quarterly 38 (2): 195-222.

Barkow, Rachel E. 2010. “Insulating Agencies: Avoiding Capture through Institutional Design.” Texas Law Review 89 (15): 16-78.

Bartels, Larry M. 2008. Unequal Democracy: The Political Economy of the New Gilded Age. Princeton, NJ: Princeton University Press.

Bawn, Kathleen. 1995. “Political Control Versus Expertise: Congressional Choices about Adminis- trative Procedures.” American Political Science Review 89 (1): 62-73.

Berinsky, Adam J., Gregory A. Huber, and Gabriel S. Lenz. 2012. “Evaluating Online Labor Mar- kets for Experimental Research: Amazon.com’s Mechanical Turk.” Political Analysis 20 (3): 351-68.

Bertelli, Anthony M., and Christian R. Grose. 2011. “The Lengthened Shadow of Another Institu- tion? Ideal Point Estimates for the Executive Branch and Congress.” American Journal of Politi- cal Science 55 (4): 767-81.

536 | PRESIDENTIAL STUDIES QUARTERLY / September 2015

Bolson, Toby, James N. Druckman, and Fay Lomax Cook. 2014. “How Frames Can Undermine Support for Scientific Adaptations: Politicization and the Status-Quo Bias.” Public Opinion Quarterly 78 (1): 1-26.

Calabresi, Steven G. 1995. “Some Normative Arguments for the Unitary Executive.” Arkansas Law Review 48 (1): 23-104.

Carpenter, Daniel. 2011. “Free the F.D.A.” New York Times, December 13. CBS News/New York Times Poll. 2010. “Gulf Oil Spill.” June 16, 2010. Sample: U.S. national

adult with oversample of gulf state residents. http://www.ropercenter.uconn.edu/ipoll-database/ (accessed May 3, 2015).

Cheibub, Jose Antonio. 2007. Presidentialism, Parliamentarism, and Democracy. New York: Cam- bridge University Press.

Chevron U.S.A., Inc. v. Natural Resources Defense Council. 477 U.S. 837 (1984). Cilluffo, Frank J., Daniel J. Kaniewski, Jan P. Lane, Gregg C. Lord, and Laura P. Keith. 2009.

Serving America’s Disaster Victims: FEMA, Where Does It Fit? Issue Brief. Washington, DC: Homeland Security Policy Institute, George Washington University. http://cchs.gwu.edu/sites/ cchs.gwu.edu/files/downloads/IssueBrief_3_HSPI.pdf (accessed April 17, 2015).

Coglianese, Cary, and Margaret Howard. 1998. “Getting the Message Out: Regulatory Policy and the Press.” Harvard International Journal of Press Politics 3 (3): 39-50.

Cohen, Jeffery E. 2008. The Presidency in an Era of Twenty-Four Hour News. Princeton, NJ: Prince- ton University Press.

Croley, Steven. 2003. “White House Review of Agency Rulemaking: An Empirical Investigation.” University of Chicago Law Review 70 (3): 821-85.

Duch, Raymond M., and Randolph T. Stevenson. 2008. The Economic Vote. New York: Cambridge University Press.

———. 2013. “Voter Perceptions of Agenda Power and Attribution of Responsibility for Eco- nomic Performance.” Electoral Studies 32 (3): 512-16.

Dudley, Susan, and Melinda Warren. 2012. Growth in the Regulators’ Budget Slowed by Fiscal Stalemate: An Analysis of the U.S. Budget for Fiscal Years 2012 and 2013. St. Louis, MO: Washington Uni- versity in St. Louis, Weidenbaum Center on the Economy, Government, and Public Policy.

Fiorina, Morris P. 1981. Retrospective Voting in American National Elections. New Haven, CT: Yale University Press.

———. 1982. “Legislative Choice of Regulatory Forms: Legal Process or Administrative Process?” Public Choice 39 (1): 233-67.

Gallup. 2013. “Government Agency Ratings and Government Power.” May 20-21. http:// www.gallup.com/poll/162764/americans-views-irs-sharply-negative-2009.aspx (accessed May 3, 2015).

Gans, Herbert J. 1979. Deciding What’s News. New York: Pantheon. Gentzkow, Matthew, and Jesse M. Shapiro. 2010. “What Drives Media Slant? Evidence from the

U.S. Daily Newspapers.” Econometica 78 (1): 35-71. Gilens, Martin, Lynn Vavreck, and Martin Cohen. 2007. “The Mass Media and the Public’s Assess-

ment of Presidential Candidates, 1952-2000.” The Journal of Politics 69 (4): 1160-75. Gomez, Brad T., and J. Matthew Wilson. 2008. “Political Sophistication and Attributions

of Blame in the Wake of Hurricane Katrina.” Publius: The Journal of Federalism 38 (4): 633-50. Government Accountability Office. 2003. Rulemaking: OMB’s Role in Reviews of Agencies’ Draft Rules

and the Transparency of Those Reviews. GAO Report GAO-03-929. http://www.gao.gov/new.i- tems/d03929.pdf (accessed April 17, 2015).

Groseclose, Tim, and Jeffrey Milyo. 2005. “A Measure of Media Bias.” Quarterly Journal of Econom- ics 120 (4): 1191-237.

Hacker, Kenneth L., Maury Giles, and Aja Guerrero. 2003. “The Political Image Management Dynamics of President Bill Clinton.” In Images, Scandal, and Communication Strategies of the Clin- ton Presidency, eds. Robert E. Denton Jr. and Rachel L.Holloway. Westport, CT: Praeger, 1-38.

Harris Interactive. 2010. “Troubles for Traditional Media—Both Print and Television.” October 28. http://www.harrisinteractive.com/NewsRoom/HarrisPolls/tabid/447/mid/1508/articleId/604/ctl/ ReadCustom%20Default/Default.aspx (accessed April 17, 2015).

Ruder / BLAME FOR AGENCY SCANDALS | 537

Hays, Kristin. 2014. “BP Bids in Gulf of Mexico Lease Sale after U.S. Lifts Ban.” Associated Press, March 19.

Huber, John D., and Nolan McCarty. 2004. “Bureaucratic Capacity, Delegation, and Political Reform.” American Political Science Review 98 (3): 481-94.

Kagan, Elena. 2001. “Presidential Administration.” Harvard Law Review 114 (8): 2245-385. Kauffman, Tim. 2010. “Minerals Management Service Chief Resigns.” Federal Times, May 27. Key, V. O. 1966. The Responsible Electorate. Cambridge, MA: Harvard University Press. Larcinese, Valentino, Riccardo Puglisi, and James M. Snyder. 2011. “Partisan Bias in Economic

News: Evidence on the Agenda Setting Behavior of Newspapers.” Journal of Public Economics 95 (9-10): 1178-89.

Lessig, Lawrence, and Cass R Sunstein. 1994. “The President and the Administration.” Columbia Law Review 94 (1): 1-123.

Lewis, David. E. 2003. Presidents and the Politics of Agency Design. Stanford, CA: Stanford University Press.

———. 2008. The Politics of Presidential Appointments: Political Control and Bureaucractic Performance. Princeton, NJ: Princeton University Press.

Lowi, Theodore J. 1979. The End of Liberalism. 2nd ed. New York: W. W. Norton. Maestas, Cherie D., Lonna Rai Atkeson, Thomas Croom, and Lisa A. Bryant. 2008. “Shifting the

Blame: Federalism, Media, and Public Assignment of Blame Following Hurricane Katrina.” Publius: The Journal of Federalism 38 (4): 609-32.

Mainwaring, Scott, and Matthew S. Shugart. 1997. “Juan Linz, Presidentialism, and Democracy: A Critical Appraisal.” Comparative Politics 29 (4): 449-71.

Malhotra, Neil, and Alexander G. Kuo. 2008. “Attributing Blame: The Public’s Response to Hur- ricane Katrina.” The Journal of Politics 70 (1): 120-35.

McCoy, Kevin. 2013. “Consumer Protection Bureau is Under Attack, but Working.” USA Today. June 6.

Moe, Terry M., and Scott A. Wilson. 1994. “Presidents and the Politics of Structure.” Law and Contemproary Problems 57 (Spring): 1-44.

Office of Management and Budget. 2013. “The Mission and Structure of the Office of Management and Budget, Office of Management and Budget, the White House.” www.whitehouse.gov/ omb/organization_mission (accessed December 3, 2013).

Powell, G. Bingham Jr., and Guy D. Whitten. 1993. “A Cross-National Analysis of Economic Voting: Taking Account of the Political Context.” American Journal of Political Science 37 (2): 391-414.

Prat, Andrea, and David Str€omberg. 2011. “The Political Economy of Mass Media.” Working Paper. http://www.columbia.edu/~ap3116/papers/mediasurvey11.pdf (accessed April 17, 2015).

Puglisi, Riccardo. 2011. “Being the New York Times: The Political Behavior of a Newspaper.” The B.E. Journal of Economic Analysis and Policy 11 (1): art. 20.

Rosen, Bernard. 1998. Holding Government Bureaucracies Accountable. 3rd ed. Westport, CT: Praeger. Rottinghaus, Brandon, and Zlata Berenznikova. 2006. “Exorcising Scandal in the White House:

Presidential Polling in Times of Crisis.” Presidential Studies Quarterly 36 (3): 493-505. Ruder, Alex I. 2014. “Institutional Design and the Attribution of Presidential Control.” Quarterly

Journal of Political Science 9 (3): 301-35. Rudolph, Thomas J. 2003. “Institutional Context and the Assignment of Political Responsibility.”

The Journal of Politics 65 (1): 190-215. Sandhaus, Evan. 2008. The New York Times Annotated Corpus. Philadelphia: Linguistic Data

Consortium. Schattschneider, Elmer Eric. 1942. Party Government. New York: Rinehart. Snyder, James M., and David Str€omberg. 2010. “Press Coverage and Political Accountability.”

Journal of Political Economy 118 (2): 355-408. Snyder, James M., and Riccardo Puglisi. 2011. “Newspaper Coverage of Political Scandals.” The

Journal of Politics 73 (3): 931-50. Volden, Craig. 2002. “A Formal Model of the Politics of Delegation in a Separation of Powers

System.” American Journal of Political Science 46 (1): 111-33.

538 | PRESIDENTIAL STUDIES QUARTERLY / September 2015

West, William F. 2005. “The Institutionalization of Regulatory Review: Organizational Stability and Responsive Competence at OIRA.” Presidential Studies Quarterly 35 (1): 76-93.

———. 2006. “Presidential Leadership and Administrative Coordination: Examining the Theory of a Unified Executive.” Presidential Studies Quarterly 36 (3): 433-56.

Wilhoit, G. Cleveland, and David H. Weaver. 1991. The American Journalist: A Portrait of U.S. News People and Their Work. Bloomington: Indiana University Press.

Supporting Information

Additional Supporting Information may be found in the online version of this article at the publisher’s Web site:

Appendix S1. Supplemental Analysis Appendix S2. Survey Instrument Appendix S3. Offshore Lease Statistics Appendix S4. Federal Reserve Appendix S5. Minerals Management Service

Ruder / BLAME FOR AGENCY SCANDALS | 539

Copyright of Presidential Studies Quarterly is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.