Political Ideology in the Bureaucracy
Political ideology in the bureaucracy (def.). A condensed representation of the policy preferences held by individuals or groups of individuals (e.g., agencies) tasked with carrying out the orders of policymakers.
In political science, research studying the public bureaucracy has focused heavily on whether public bureaucrats prioritize their own ideological goals over politicians’ policy directives (Miller 2005). In public administration, scholarship has examined whether public bureaucrats passively (via shared personal traits) or actively (via the pursuit of shared ideological aims) represent the communities they serve (Lim 2006; Meier 1975; Meier and Nigro 1976; Krislov 1974). By concentrating on these topics, research has moved ideology to the heart of the conversation about public bureaucracy. Yet, in so doing, these literatures have exposed a troublesome empirical problem: measuring the ideologies of public bureaucrats is difficult. This entry describes the challenge of measuring bureaucrats’ ideologies and the recent methodological developments that have facilitated the measurement of public bureaucrats’ ideologies.
What Is Ideology?
Generations of political scientists have sought to define ideology (Hamilton 1987), but no universally agreed-upon definition of ideology exists. Indeed, some challenge the concept’s distinctiveness and usefulness (Mullins 1972). Thus, although various existing definitions provide a sense of what is meant by “ideology” (notably, Bell (1960, p. 393; Althusser 1971), no definition can claim superiority over others.
With this state of the literature as a backdrop, students of public bureaucracy have turned to the definition of ideology employed by rational choice scholars to lay the theoretical foundation on which to build their empirical estimates. Scholars drawing on the paradigm of rational choice have characterized ideology as a reflection of the policy preferences an individual possesses (Poole 2005). These policy preferences are represented by a mathematical model that describes preferences based on their placement in an abstract space containing all possible policies that could be adopted. In this model, a specific individual’s policy preferences can be represented by a location in this space. That pinpointed location is the individual’s single, most-preferred policy (the “ideal point”) within the entire range of possible policies. Any policy that deviates from an individual’s ideal point is valued less by the individual than the policy at the ideal point; the policy’s value declines as it grows more distant from the ideal point. With this conception of ideology in tow, individuals are then assumed to act in ways that maximize the utility they gain from adopted policies. In sum, to the rational choice scholar, individuals’ ideologies consist of their ideal points, as well as the distance between those ideal points and other points in the policy space; furthermore, individuals act on their ideological leanings so as to maximize the utility they derive from policy.
Viewing ideology via this spatial model of preferences has facilitated empirical efforts aimed at measuring ideology and it did so initially in topical areas outside the study of public bureaucracy. This spatial model enables the empirical estimation of preferences because it implies that an individual’s behavior indicates her preferences, due to the assumption of utility maximization. Additionally, this behavior, due to the relational nature of preferences in the paradigm, can be compared with the behavior of others in order to place individuals on a common ideological scale (Poole and Rosenthal 1991, 1997). Locating actors on that ideological scale can be achieved by data reduction techniques that analyze political actors’ recurrent decisions and summarize patterns in those data with numerical estimates. Accordingly, to derive ideological estimates, scholars have scaled patterns of voting by legislators, decisions by judges, and stances on legislative votes by executives as a means of placing political actors’ preferences on quantitative scales (see discussion and citations in McCarty 2011). These quantitative measures, in turn, could be compared and incorporated into broader analyses testing theories of political behavior and institutions.
Recently, students of public bureaucracy have sought to follow in the footsteps of this research tradition in order to develop quantifiable estimates of the ideologies of public bureaucrats. However, as initial forays into this line of research indicated, unique features of bureaucrats’ work environments made it difficult to directly transfer previously-used methods into the sphere of public bureaucracy research.
The Challenge of Measuring Bureaucratic Ideology
Measuring the ideologies of public bureaucrats, as Clinton et al. (2012) have noted, proves to be more challenging than estimating the ideal points of legislators, judges, or political executives due to the fact that bureaucrats perform a heterogeneous set of activities. Even when looking within a given governmental context, researchers can see this heterogeneity vividly. The activities performed by a manager in the U.S. Department of Labor, for instance, differ from those performed by a manager in the U.S. Department of Health and Human Services and that manager’s activities often differ from those of an attorney in the U.S. Department of Justice or a forest ranger in the National Park Service. Given these incongruent activities, one cannot directly employ the scaling methods used to estimate legislative, executive, or judicial preferences. After all, those methods rest on a comparison of common behaviors in order to infer individuals’ underlying ideologies.
Given these difficulties, early studies of bureaucratic ideology offered rich descriptive information about the various policy positions that bureaucrats held, but refrained from attempts to characterize the ideologies that might underlie those positions. For instance, Meier (1975) turned to survey data concerning policy positions and avowed vote choices in the 1972 U.S. Presidential election in order to understand the policy viewpoints of public bureaucrats working in federal, state, and local governments. Meier uncovered a leftward lean in bureaucrats policy preferences, relative to other survey respondents; however, only in their support for “Trade with Communists” and the “Legalization of Marijuana” did the positions of public bureaucrats differ significantly (alpha=0.10) from the policy preferences of non-bureaucrats (Meier 1975, Table 7). The differences in the distribution of responses between bureaucrats and non-bureaucrats for survey items such as self-reported voting for Nixon, affiliation with the Republican Party, and a range of other policy issues could not be distinguished from mere sampling variation (Meier 1975, Table 7). Moreover, in subtle defiance of the notion that demographic representation translated into ideological representation in the public bureaucracy, Meier (1975) indicated that the bureaucrats’ demographic attributes correlated more weakly with their policy positions than did their agency affiliations. These findings were of intrinsic interest, given the limited information available about public bureaucrats’ ideologies at the time, yet they gave only an episodic view of bureaucrats’ policy preferences (i.e., no attempt was made to consider intra-individual correlations across issue items) and they did not offer any means of comparing bureaucrats’ preferences to other political actors (such as judges or legislators), unless those actors were in the pool of survey respondents. Such efforts continued in the decades thereafter as scholars offered summary statistics about specific behavioral patterns that indicated U.S. public bureaucrats’ views on specific political matters (e.g., Johnson and Libecap 1994).
However, as those methods were being employed, a theoretical literature that sought to understand interactions between public institutions – such as legislatures, the courts, and the bureaucracy (Ferejohn and Shipan 1990) – began to develop (de Figueiredo et al 2008). To test the models developed in this literature on the separation of powers, scholars had to develop measures of bureaucratic ideology that could be compared with the rich set of ideological measures available in the study of legislatures, the courts, and the presidency (again, see McCarty 2011 on the latter measures).
Measures of Bureaucratic Ideology
Given that theoretical models in the separation of powers literature focus on the interaction of institutions, with early models implying that a single ideal point could capture the ideology of a given institution (Ferejohn and Shipan 1990), a first attempt at measuring bureaucratic ideology involved using the historical attributes or policy missions of agencies as surrogates for ideology. This approach rested on a reasonable premise: political factors shape the origins and purpose of agencies (Howell and Lewis 2002), thus categorical variables relating to those structural features can capture an agency’s ideological lean. For example, using the partisan control of institutions during the time of an agency’s founding provides a stand-in for that agency’s ideology (see Lewis 2007 and Lewis 2008 for examples of this approach).
This stand-in for ideology, however, contained many limitations. First, the method of implementing this approach often confounds the ideological information contained in the measure with other, unrelated factors. Consider a measure of agency ideology that indicates if an agency was created under a president affiliated with the Democratic Party or the Republican Party. Any variable correlated with when Democratic Party politicians have held the Presidency – for instance, economic trends – would confound the relationship between the indicator of agency ideology and an outcome variable of interest. Furthermore, such a measure does not convey the ideology of the public bureaucrats in the agency per se unless the structural factors promote self-selection into the agency by ideologically like-minded bureaucrats (as the evidence in Meier 1975 suggests; also, in a related vein, see Gailmard and Patty 2007). Absent the possibility of self-selection, the measure might fail to capture the actual ideological leanings of individuals working in the agency. Finally, the measure does not facilitate tests of models that make claims about the ideological distance of on institution from another (e.g., Ferejohn and Shipan 1990). Such models rest at the heart of understanding the problem of bureaucratic discretion and political control (see Epstein and O’Halloran 1999; Huber and Shipan 2002).
Clinton and Lewis (2008) offered a solution to the latter problem by examining survey data. However, unlike earlier efforts to use survey data to understand bureaucratic ideology (Meier 1975), Clinton and Lewis (2008) conducted a survey that asked academics and journalists familiar with the U.S. public bureaucracy to rate the ideological proclivities of 82 federal agencies. Clinton and Lewis (2008) then integrated these ratings with further information about the agencies themselves to model the latent – that is, unobserved in the data – ideology of the agency. This measure captured subtle differences in the ideological leanings of agencies. However, a later team of scholars – which included both Clinton and Lewis – acknowledged that, from this method, “[p]roblems arise if experts are limited in their knowledge of lesser-known agencies or make similar mistakes in categorizing agencies as liberal or conservative” (Clinton et al. 2012, p. 2). Also, even when setting aside those methodological issues, Clinton and Lewis (2008) presented a measure that – just as with the previous approach of using nominal variables to proxy for ideology – did not offer insight into the ideologies of individual bureaucrats, nor did it enable comparison with ideological measures of other institutions (Clinton et al. 2012, p. 2).
Perhaps the first method of providing estimates of individual bureaucrats’ ideologies was ingeniously devised by David C. Nixon (2004). Nixon (2004) recognized that a non-trivial portion of top-ranking public bureaucrats served in Congress prior to their entry into the U.S. federal service. This career trajectory meant that those officials’ ideologies had been measured via other scholars’ efforts to measure the ideologies of members of the U.S. Congress (Poole and Rosenthal 1991, 1997). Accordingly, one could simply pluck those ideological estimates from prior analyses and incorporate them into empirical tests that sought to understand interactions between public bureaucrats and the Congress (Nixon 2004). In sum, the approach yielded individual estimates that could be compared across institutions. The only problem, however, was that the approach could only yield estimates of bureaucratic ideology for the subset of bureaucrats who had served in Congress – a small subset of the public bureaucracy.
To widen the number of bureaucratic officials for whom ideological estimates were available, while also providing a time-varying measure of bureaucratic ideology, Bertelli and Grose (2007, 2009, 2011) devised a method that could use the public statements issued by top officials to measure bureaucratic ideology. The method collected the testimonies of high-level U.S. federal bureaucrats from 1991 to 2004 and identified testimonies that spoke to legislation subject to roll call voting. Coding procedures then rated these top officials support or opposition for that legislation as if they had issued a vote in a roll call. Integrating these responses with the actual votes of members of Congress and, then, performing scaling analyses allowed for the estimation of ideological estimates that could be compared across institutions. Although it clearly offered an insightful means of measuring the ideologies of top officials, the method still could not extend to a wider set of public bureaucrats. Only the relatively few public bureaucrats who had issued public statements could have their ideologies estimated by the method.
The formation of a leading team of researchers in this area – consisting of Clinton, Bertelli, Grose, Lewis, and Nixon – led to the development of ideological estimates covering a wider number of bureaucrats (Clinton et al. 2012). Clinton et al. (2012) sent a survey to thousands of public bureaucrats during the presidential administration of George W. Bush. The survey inquired about these bureaucrats’ policy preferences by asking respondents about their views on policy issues reflected in legislation subject to Congressional roll call votes; responses to these survey items then were scaled (Clinton et al. 2012). The method, in other words, combined the survey methods of past research (Meier 1975; Clinton and Lewis 2008) with the use of legislation subject to roll-call votes (Bertelli and Grose 2009) – in a framework that placed bureaucrats and legislators in a common ideological scale (Nixon 2004) – to estimate bureaucratic ideology. The method offered a rich set of individual-level and aggregate-level ideal points. However, the survey methods limited the scope of the analysis and made the prospect of updating the ideal point estimates painstaking. Only roughly 33 % of individuals receiving a survey request responded, with appointees even less likely to respond (Clinton et al. 2012); furthermore, the resulting estimates only covered the years 2005–2006, thus requiring the procedures to be updated frequently to ensure wider temporal scope.
To develop estimates that covered a larger number of public bureaucrats and that spanned a longer time period, Chen (2010) and Chen and Johnson (2011, 2015) studied a common activity in which a substantial number of public bureaucrats engage: donating contributions to political campaigns. Using the universe of federal campaign contribution data, Chen and Johnson (2015) identified all donors who (1) issued campaign contributions of $200 (USD) or more to incumbent federal politicians or their political action committees (PACs). Then, given that federal election laws require that donors reveal their employers when making a contribution, Chen and Johnson (2015) identified all campaign contributors who listed a U.S. federal agency as their employer. Subsequently, for every contribution issued by a federal bureaucrat, Chen and Johnson (2015) identified the ideal point estimates of the donation’s recipient as measured via the method of Poole and Rosenthal (1991, 1997). Then, to estimate an agency’s ideology, the researchers took the ideal points associated with contributions issued by employees of a given agency and, weighting these ideal points by the dollar value of each donation, estimated the mean of the ideal points. This led to a measure of agency ideology based on a common activity – campaign contributions – that spanned over two decades and covered myriad public bureaucrats. Moreover, the ability to use dollar amounts as a weighting device allowed the authors to emphasize larger contributions likely to have been issued by wealthier, top-level officials, plus the approach could be readily automated in order to update estimates frequently and it could generate individual-level estimates (see Bonica et al. 2015).
The method, however, was not without potential problems. First, the method fails to offer valid ideological estimates if campaign contributions do not reflect a genuine ideological expression. Ansolabahere et al. (2003) provide evidence to that effect, showing that individuals’ contributions can only be interpreted as an ideologically motivated consumption good. Moreover, Gimpel et al. (2008) report that contributions issued by individuals who will not be represented by a candidate (i.e., out-of-district contributors) appear ideologically motivated. However, Gordon and Hafer (2005) present evidence of strategic giving, thereby raising questions that the method calculates the sincere ideological orientation of public bureaucrats. Furthermore, the weighting scheme used in the estimation assumes that higher campaign contributions come from higher-paid bureaucrats and those bureaucrats’ ideologies figure more prominently in agency activities. These assumptions might not be valid and, to a degree, they defy the widely-held view that agency problems in the public bureaucracy mean that even low-level bureaucrats can distort politicians’ policy objectives.
The conceptualization of public officials’ political ideologies looms large in both political science’s and public administration’s studies of bureaucracy. Yet, because public bureaucrats perform highly-variable activities, distilling ideological patterns from their behavior proves to be difficult. Early studies of bureaucratic ideology circumvented this problem by simply supplying summary statistics about bureaucrats’ preferences for specific policies and political candidates (e.g., Meier 1975). However, with the development of theoretical frameworks that offered general models of how institutions interacted to produce policy, the need for a measure of bureaucratic ideology – not just individual issue viewpoints – became apparent. Scholars responded initially by using categorical variables to convey the policy orientation of agencies (e.g., Lewis 2007), yet this method was quickly eclipsed by methods that produced ideological estimates from survey methods (Clinton and Lewis 2008; Clinton et al. 2012), analyses of bureaucrats who worked in other institutions (Nixon 2004), the study of public messages delivered by bureaucrats (Bertelli and Grose 2007, 2009, 2011), and the investigation of bureaucrats’ campaign contributions (Chen 2010; Chen and Johnson 2011). Despite disadvantages to each of these methods, scholars now possess measures of bureaucratic ideology that can be readily incorporated into empirical tests of models concerning the public bureaucracy.
- Althusser L (1971) Ideology and ideological state apparatuses. In: Brewster B (ed) Lenin and philosophy and other essays. Monthly Review Press, New York, pp 121–176Google Scholar
- Bell D (1960) The end of ideology. Harvard University Press, Cambridge, MAGoogle Scholar
- Chen J (2010) Estimating bureaucratic ideal points with federal campaign contributions. Paper presented at the annual meeting of the American Political Science Association, Washington, DCGoogle Scholar
- Chen J, Johnson T (2011) Unionization and Presidential control of the bureaucracy. Paper presented at the annual meeting of the Midwest Political Science Association, ChicagoGoogle Scholar
- Chen J, Johnson T (2015) Federal employee unionization and presidential control of the bureaucracy: estimating and explaining ideological change in executive agencies. J Theor Polit 101(4):657–676Google Scholar
- de Figueiredo RJP, Jacobi T, Weingast BR (2008) The new separation-of-powers approach to American politics. In: Wittman DA, Weingast BR (eds) The Oxford handbook of political economy. Oxford University Press, Oxford, pp 199–222Google Scholar
- Krislov S (1974) Representative bureaucracy. Prentice Hall, Englewood CliffsGoogle Scholar
- McCarty N (2011) Measuring legislative preferences. In: Edwards GC III, Lee FE, Schickler E (eds) The Oxford handbook of the American Congress. Oxford University Press, Oxford, pp 66–94Google Scholar
- Poole KT, Rosenthal H (1997) Congress: a political economic history of rollcall voting. Oxford University Press, OxfordGoogle Scholar