How Do Elected Officials Evaluate Performance? Goal Preferences, Governance Preferences, and the Process of Goal Reprioritization

How Do Elected Officials Evaluate Performance? Goal Preferences, Governance Preferences, and the... Abstract Performance data allows politicians to exert accountability over public organizations, even as ideological biases can affect how they interpret such data. However, we know little about how motivated decision-makers prioritize goals when facing multiple pieces of contradictory performance data that reflect the competing goals of public services. Such goal conflict is an inherent aspect of public management. To understand its implications for the use of performance data, we develop a theory of goal reprioritization. We start by assuming that elected officials have preferences between specific policy goals, and about governance processes—such as a preference for public or private service provision. When elected officials face contradictory pieces of performance data, governance preferences drive performance evaluations to the point that they are willing to reweight their goal preferences to minimize cognitive dissonance. We offer experimental evidence of this process, showing that elected officials asked to evaluate school performance reprioritize between two distinct policy goals for schools—test scores and student well-being—to fit with their governance preferences. Reprioritization is an attractive strategy since it allows elected officials to claim they are using performance data, even as underlying governance preferences lead them to set aside the evaluative goal-based criteria by which they would otherwise make performance evaluations. In other words, preferences concerning the nature of government can trump goal preferences when decision-makers use performance data. Introduction We are in “the age of quantified performance” (Van de Walle and Roberts 2011, 215) reflecting the ongoing ubiquity of performance management in governance (van Dooren 2011). Public organizations set goals and collect large amounts of data to evaluate whether these goals have been reached, in the hope of allowing elected officials to allocate resources more effectively, prioritize between different programs and optimize organizational processes (for critical assessments, see Julnes 2011; Kroll 2015; van Dooren et al. 2010). The same underlying logic on how data can improve outcomes also motivates the evidence-based policymaking movement (Heinrich 2007). A truism of the performance movement is that “what gets measured gets managed”, meaning that unmeasured goals are subject to inattention and goal displacement. While the risks of a unidimensional performance system are well documented, as we discuss in the next section, we know much less about how decision-makers manage multidimensional performance data. Even a relatively limited array of data can reflect goal conflict, where multiple performance indicators contradict each other, undermining the potential for a single consistent image of performance to emerge. Since organizations make tradeoffs between goals, goal conflict can be expected to be routine (Chun and Rainey 2005; Hvidman and Andersen 2016; Rainey and Jung 2015; Wright 2004). Political principals can navigate goal conflict if they have a clear and consistent preference for the policy goals they hope to maximize, and manage public organizations accordingly. However, ideological beliefs may undercut the ability to interpret data about policy with consistency (Kunda 1990; Taber and Lodge 2006). An important insight on this point comes from Baekgaard and Serritzlew (2016, 73), who argue that “even objective, clear and unambiguous performance information is subject to biased interpretation depending on whether the information is consistent with the prior beliefs held by those who receive the information.” They asked subjects to rank hospital performance using simple objective measures of operations, and showed that people with strong pro-market or pro-public beliefs were more likely to make errors when the sector affiliation of the hospital was revealed. Related work confirms such tendencies among elected officials (Baekgaard et al. 2017; Nielsen and Moynihan 2017). While prior work alerts us to how political beliefs affect how people process one piece of performance data, it does not tell us the role that such beliefs play as decision-makers manage multiple pieces of performance data, and in particular, weigh tradeoffs in conditions of goal conflict. We therefore develop and test a theory that incorporates the reality of goal conflict. To do so, we adapt a specific subset of psychological theory on motivated reasoning that we characterize as goal reprioritization, which can be understood as the “opportunistic adjustment of the weight-assigned evidence conditional on its conformity” to desired conclusions (Kahan 2016, 1). We explain goal reprioritization in greater detail below, but for now we note that even as it has been recognized as “the signature feature” of identity-protective motivated reasoning (ibid.), it has not been used to explain conditions of goal conflict or other decision contexts in contemporary public administration. To adapt reprioritization theory to the complex but realistic context of elected officials with different preferences assessing public sector performance with multiple and contradictory pieces of information, it is necessary to add a conceptual distinction, not explicit in prior work, between two types of preferences: governance preferences and goal preferences. Governance preferences reflect ideological beliefs about the appropriate processes in and design of the public sector. In our study, we model governance preferences as a preference for public or private provision of services. Goal preferences are the weighted preference for one policy goal over another, and we examine goal preferences within a policy domain. For example, people may want schools to provide both good academic outcomes and well-rounded students, but weigh one goal as more important than the other. Goal preferences form evaluative criteria by which individuals can assess the performance of a service provider. Those who weigh academic achievement as more important than student well-being should do so consistently, thereby providing goal clarity to organizations via the evaluation process. However, our application of goal reprioritization hypothesizes that governance preferences can compete with goal preferences, sometimes causing decision-makers to relegate goal-based evaluative criteria in their assessment of public sector performance. We test this hypothesis using a survey experiment with responses from Danish city councilors. The elected officials were placed in multiple groups that allow both a demonstration of the salience of goal preferences and the influence of governance preferences when the officials evaluate performance data about schools. In control groups, elected officials consistently apply goal preferences as evaluative criteria. However, when they are told the sector affiliation of the schools the goal-based evaluative criteria are set aside. Instead, officials tend to assess the performance data through the lens of their governance preferences for public or private service provision. The results are substantively important as they demonstrate how governance preferences can undercut the ability of governments, even when working within a performance management framework, to clearly and consistently pursue a set of specific policy goals. Both governance and goal preferences are ideological preferences, but they serve different purposes, making it all the more important to conceptually and empirically distinguish their salience for public management. Elected officials may establish a principal-agent chain of accountability between themselves, voters and bureaucrats on a set of promises about both governance and policies. Our theory proposes that commitments to policy goals are fluid enough to be contingent upon assumptions about the administrative machinery in place, and when those assumptions are violated, ideological actors are willing to re-weight their goal preferences. While other work has convincingly made the case that ideology matters in how public performance is evaluated, we add nuance to this claim: ideology matters, but ideology about governance may matter more than ideology about policy goals. In the following sections, we build our argument by drawing from a mix of literatures. First, we review research on performance management and goal ambiguity to make the case that goal conflict is a common feature of public administration. We next examine research that shows that ideological beliefs matter in how people process performance data, and draw from the broader literature on motivated reasoning to develop the concept of goal reprioritization. These literatures provide both the theoretical strands to make our argument, while demonstrating its relevance. But across these literatures, there is no prior work that makes the specific predictions that we do. We next present results, and consider the implications and limitation of our research. Goal Conflict: A Basic Condition of Public Administration The theoretical setting in which we situate our study—goal conflict—is a frequent and salient context of public management (Chun and Rainey 2005; Wright 2004). In this section, we document how performance management systems capture goal conflict. The basic logic for performance management is clear: decision-makers with valid information about the performance of their organizations are better able to make informed, goal-oriented decision making than those who rely on intuition (Holzer and Yang 2004, 16). If having some performance data is good, a variety of perspectives make the case that having more data is better. Performance data provides some basic form of accountability to the public and to policymakers (Behn 2001). Because public organizations pursue multiple goals, a single or small number of measures fail to fully capture the diversity of public sector outcomes (Boyne 2002; Chun and Rainey 2005). More measures should therefore provide better accountability, and better reflect competing goal preferences. For example, Jacobsen, Snyder, and Saultz (2015) argue that a unidimensional approach to performance measurement runs the risk of mischaracterizing how people assess public services. They show that an assessment of school performance based only on academic performance fails to capture the goal preferences of those who prefer a well-rounded education.1 Adding more metrics is a solution to this problem, and to a variety of other observed issues with performance systems, such as the prioritization of efficiency measures over other dimensions of performance, like equity or procedural safeguards (Moynihan et al. 2011), or goal displacement as individuals increase effort on measured goals while reducing effort on unmeasured ones (Bohte and Meier 2000; Holmstrom and Milgrom 1991; Kelman and Friedman 2009; Rainey and Jung 2015). In practice, performance systems accrue measures over time, incrementally evolving into more comprehensive frameworks (Moynihan and Beazley 2016). But more comprehensive performance systems create their own problems, such as increased transaction costs (Bischoff and Blaeschke 2016). More measures are also more likely to reflect and formalize goal conflict. Chun and Rainey (2005, 2) define “direct goal conflict” as a context where “achieving one valued goal directly inhibits achieving another desired goal” while “indirect goal conflict can be observed whenever there are multiple goals at the same hierarchical level since having more goals means having a greater need to split limited resources and thereby exposes more points of potential conflict.” While the more general construct of goal ambiguity can give rise to goal conflict (Wright 2004), goal conflict specifically taps counteracting relationships between goals. For any moderately complex program, there will be multiple goals and multiple measures that reflect those goals, which frequently send conflicting signals to decision-makers. Even if multiple goals do not actually represent trade-offs, policymakers may assume that the achievement of each additional goal draws resources away from other goals.2 The sense of conflict may become stronger if performance of one goal appears to move in opposite direction to others. Goal conflict is not, in itself, a bad thing. It may lead to more difficult decision processes, but perhaps also better decisions, reflecting a fuller appreciation of the nuances of a program (Moynihan 2015). But goal conflict also begets discretion in how to evaluate programs, and individual preferences will influence the use of that discretion. While prior work has suggested that goal conflict can be expected to be a common condition in government, this still leaves open the question of how preferences affect how individuals weigh data about conflicting goals. The following section makes use of psychological literature about information processing to offer a theoretical model to incorporate how ideological beliefs matter in conditions of goal conflict. Preferences and Performance Information Use The literature on motivated reasoning provides a basis for understanding the role of attitudes and beliefs when decision-makers process performance data. People are often motivated to “justify a specific, preselected conclusion” (Lodge and Taber 2000, 186). To do so, they actively seek to interpret information in ways that support desired conclusions (e.g., Groenendyk 2013; Jerit and Barabas 2012; Kunda 1987, 1990; Lavine et al. 2012; Taber and Lodge 2006, 2016). For instance, voters whose party is in charge of government are more forgiving of government failures, and blame factors others than the actions of the government when explaining negative outcomes (Bisgaard 2015; Tilley and Hobolt 2011). A nascent literature has begun to experimentally demonstrate how attitudes and beliefs matter to the selection and processing of performance information about public organizations. Moynihan (2008, 155) hypothesized that: “that the creation, presentation, and use of performance data in the public sphere will reflect the values of those involved, be they partisan, organizational, institutional, or other.” As noted in the introduction, Baekgaard and Serritzlew (2016, 73) offer empirical evidence for this claim by showing that pro-market or pro-public beliefs affect how people process performance data. James and Van Ryzin (2017) show that priming people to think politically affects how they evaluate the usefulness of data when assessing the US Patient Protection and Affordable Care Act, which was passed by the Democratic party. For example, politically primed Democrats evaluate data supportive of the Act as stronger evidence. Elected officials have greater policy expertise than average voters, but are not immune to preferences affecting their judgments. Indeed, they seem to resist attempts to debias faulty evaluations through mechanisms like being asked to justify interpretations of information (Christensen 2017) or through provision of additional evidence at odds with the officials’ governance preferences (Baekgaard et al. 2017). Partisan beliefs also seem to affect who politicians listen to when using performance data for decision making. Thus, Nielsen and Moynihan (2017) found that liberal elected officials were responsive to interpretations of performance data provided by an advocate (a union representative) who shared their ideological beliefs, while conservatives ignored this advocate. This article adds to this growing literature on the psychology of performance information use by adapting theory from psychology to model how preferences affect decision-makers’ weighting of multiple pieces of contradictory evidence about competing policy goals. To do so, it becomes necessary to make explicit some distinctions hinted at in prior empirical work, which has framed ideology as a factor that affects how performance data is processed (e.g., Baekgaard and Serritzlew 2016; Hvidman and Andersen 2016), without accounting for how performance measures themselves also reflect ideological preferences for some goals over others. This becomes more apparent when we attempt to model how decision-makers manage multiple performance goals rather than just one, which allows them to weight data on their preferred goal more heavily than data on other goals. Thus, we expect that goal preferences form evaluative criteria in cases of goal conflict. We also need to distinguish between governance preferences and goal preferences. Goodsell (2004) argues that preferences about governance have been taught to citizens as a coherent package of beliefs, which in recent decades have been defined in negative terms that government agencies, lacking an element of competition, inherently drift into inefficiency and inflexibility, rewarding employees with overly generous terms at the expense of responsiveness to the public or politicians. This results in a preference for less government over more, and private provision over public. Governance preferences may be so ingrained as to form an unconscious bias about public organizations (Marvel 2016). For example, people’s stereotypes of public and private sectors may lead them to assume that private organizations perform better on measures of efficiency or managerial flexibility (Hvidman and Andersen 2016; Marvel 2016). As we explain in the next section, the stability of such governance preferences may affect how people prioritize between conflicting goal preferences about policy. Reprioritization as a Strategy Psychologists have long been aware of a human tendency to downplay the importance of psychologically unpleasant information while increasing the perceived importance of pleasant information. Festinger (1957) described how people often engage in retrospective reweighting of the relative importance they assign to different thoughts in order to reduce cognitive dissonance. Festinger used this insight to explain how people behave after making choices between competing alternatives (e.g., buying certain consumer items instead of others). In such situations, people are motivated to convince themselves that they have made good choices. An effective strategy for doing so, is to “magnify the importance of the good points associated with the chosen alternative and to think of new advantages that [one] hadn’t thought of before” (Festinger 1957, 45), while downplaying disadvantages with the chosen options and advantages with unchosen ones. Simon et al. (1995) use the term “trivialization” to describe an equivalent reasoning strategy. More recently, Groenendyk (2013) has applied similar logic to understand politically motivated reasoning among voters, using the term “issue reprioritization” to describe how partisans systematically alter the importance they assign to different issues depending on what makes their own political party look best (Groenendyk 2013, 49). As a reasoning strategy, reprioritization is psychologically appealing. It is simple to execute but opaque enough to be defensible. More blatant forms of motivated reasoning, such as ignoring or misinterpreting information, leave people at risk of looking irrational and closedminded, unattractive personal traits we generally seek to avoid (Klar and Krupnikov 2016). Reprioritization, on the other hand, allows people to maintain an “appearance of objectivity” (Groenendyk 2013, 50) by acknowledging the existence of inconvenient information without having to make politically inconvenient conclusions. Thus, people can leverage the ambiguity of the information environment to come up with reasonable arguments to support desired conclusions. We adapt issue reprioritization to the public management context of competing goal preferences, and we therefore use the term goal reprioritization. There are some differences in the application of the two concepts. Issue reprioritization has been primarily used to explain how partisans reweight the importance of individual issues when given information. However, the literature on issue reprioritization does not tell us whether such processes also apply to elected officials, the processing of performance data, and the weighting of different goals in the same policy area. As such, this is an adaption rather than a straightforward application of the theory. Parallels to goal reprioritization can be found in Wildavsky’s description of “retrospection” (Wildavsky 1979), a tendency to “attribute new motivational meaning to what we have done” (ibid., 136) and to “reformulate our problems” (ibid., 137) in ways that allow us to “make what we have done conform to reason as we understand it” (ibid.). However, while the “opportunistic adjustment of the weight-assigned evidence conditional on its conformity to” desired conclusions has in psychological literature been recognized as “the signature feature” of identity-protective motivated reasoning (Kahan 2016, 1), insights about this kind of reasoning have been largely overlooked in contemporary public administration settings. This oversight is remarkable given the ready-made utility of goal reprioritization to understanding public sector conditions of multiple and competing goals. Thus, with this article we investigate the usefulness of goal reprioritization in understanding how decision-makers deal with goal conflict. We expect that decision-makers will tend to use goal reprioritization as a strategy to make performance evaluations align with their governance preferences whenever there is a conflict between these preferences and the decision-makers’ goal preferences. Hypothesis: When facing contradictory performance data, elected officials will use goal reprioritization as a strategy to make their use of performance information fit with governance preferences for public or private service provision. To illustrate how this works in practice, imagine a political principal who has a goal preference for test scores over student well-being. To send a consistent signal to agents, the principal should then use performance data to identify and reward schools that perform well on test scores. In other words, the weights in the principal’s evaluation function should be consistently determined by the principal’s goal preferences. A school that performs poorly on test scores, but well on student well-being, should be evaluated as poorly performing. This process of goal weighting is shown in the first illustration of Figure 1. Figure 1. View largeDownload slide The Process of Goal Reprioritization Figure 1. View largeDownload slide The Process of Goal Reprioritization Yet, if the political principal also has a governance preference for public provision of education, and the data shows public schools perform better on student well-being rather than test scores, our hypothesis predicts that the political principal will interpret the performance data in a way that favors public schools. Specifically, we expect the principal to reprioritize the goals by increasing the weight assigned to student well-being and reducing the weight assigned to test-scores, consistent with the second illustration of Figure 1. The next section explains how we test this hypothesis in greater detail. Analytical Strategy Our theory examines the influence of decision-makers’ governance preferences and goal preferences in the interpretation of ambiguous performance information. In order to statistically test this, we need data on how a large number of decision-makers with varying governance preferences and goal preferences interpret comparable pieces of performance information. Specifically, in order to test our expectation of goal reprioritization, we need data allowing for an investigation of a relationship between X: (in)congruence between governance preferences and goal preferences in performance information and Y: to what extent decision-makers change evaluative criteria when evaluating the performance data. This is difficult to do with an observational model with actual decision-makers, because we cannot control or fully observe the information presented to them. We are also interested in how actual decision-makers respond to information. We therefore sample actual elected officials: Danish city councilors who make decisions about core welfare services like public schools, employment activities, childcare, and elderly care.3 We conducted our survey experiment in November and December 2014. Emails with invitations to participate were sent to each of Denmark’s 2,445 city councilors on 18th of November and when the data collection was closed on 19th of December, 988 city councilors had provided answers to our experimental items. Thus, our study has a response rate of 40.49%. The decision-makers we survey are at the frontline of the debate about public and private service provision, since they regularly decide whether or not to contract out specific services. Our analytic strategy assesses relationships between decision-makers’ governance preferences and their interpretations of performance data by comparing situations where governance preferences are, and are not, triggered. Following Baekgaard and Serritzlew (2016), governance preferences are captured by asking respondents about their preferences for public or private delivery of public services.4 We then conduct a survey experiment where we ask respondents to respond to experimentally varied performance information about a pair of fictional schools. Some are presented to information where governance preferences are not triggered (the schools are called “school A” and “school B”) while others are told that the schools are public and private.5 Information is provided about performance on two different performance indicators: test scores and student well-being. While both are desirable, these represent two distinct goals, often thought to be in conflict with one another. For example, an over-emphasis on academic achievement is often claimed to come at the expense of broader student well-being (Jacobsen, Snyder, and Saultz 2015).6 We then ask respondents to assess the performance of the schools. Our survey experiment is summarized in Figure 2. Respondents were randomly assigned to one of the four conditions in the figure. As can be seen, the respondents were presented to information that was cognitively undemanding, making it easy to see which school performed best on each indicator.7 Furthermore, the information conveyed goal conflict; the school that performed best on test scores performed worst on student well-being, and the other way around. Figure 2. View largeDownload slide Experimental Groups Figure 2. View largeDownload slide Experimental Groups Groups 1–2 serve as control groups allowing us to measure goal preferences in absence of public or private sector information. Thus, for these groups, no information was given that could be expected to trigger governance preferences, only data about school outcomes for academic achievement and student well-being. This gives us a baseline against which we can compare the effects of governance preferences. Groups 3–4 received the same information as groups 1–2, but were also told that the schools were public and private. All else equal, if goal preferences were stable, we should see little difference between the control and treatment groups (i.e., evaluations should not be affected by information about schools’ sector affiliation). However, we expect the sector information to trigger governance preferences, causing respondents in groups 3–4 to engage in goal reprioritization in order to justify their governance preferences based on the available information. To sum up, respondents were randomly assigned to one of four groups that were all asked to indicate which of two schools performed best, based on test scores and student well-being data. Because ours is a between-subject design, one limitation is that we cannot directly observe the same set of people changing their behavior as a result of the treatment. Instead, we infer that any difference in behavior of equivalent subjects is the result of the experimental treatment. More specifically, we propose that the groups told that the schools were public and private (groups 3–4) will behave differently relative to the groups not given this information (groups 1–2). A balance test (Table A1) shows that all groups balance in terms of age, gender, education, years of membership in current party, years of membership in a city council, size of municipality, the tendency to go to private school as a primary school student in respondents’ municipalities, the proportion of the groups consisting of members of a municipal financial committee, and the proportion of the groups consisting of parents to children who go to school. Members of group 2 are slightly more in favor of the public sector than the rest of the sample, which is also reflected in more members of group 2 being members of a left-wing party. Furthermore, the average member of group 4 is from a slightly wealthier municipality (measured as tax base per citizen) than the rest of the sample. However, including controls such as party affiliation and tax base does not alter our substantial results and therefore, there is no reason to believe that our findings result from imbalances between our groups. Results We do not have clear a priori expectations about decision-makers’ goal preferences between student well-being and academic achievements in the outset (i.e., in our control groups). However, we expect that politicians who have governance preferences for the public sector will tend to give higher priority to academic achievements (test scores) in group 3 where the public school performs better on this metric, and higher priority to student well-being in group 4 compared to the corresponding control groups. The reversed pattern is expected to exist among people who prefer the private sector. Our hypothesis is tested in Table 1, where we estimate the expected moderating effect of providing information about schools’ sector affiliation on the relationship between governance preferences and goal preferences. As can be seen from the positive and significant coefficients for the “pro-public preference” variable, a positive relationship exists in the outset (i.e., among respondents in the control groups) between preferring the public sector and prioritizing student well-being over academic achievements. Thus, respondents who prefer the public sector tend to find student well-being more important than academic achievements, whereas respondents who prefer the private sector find academic achievements more important. Table 1. Influence of Governance Preferences on Assessments of Performance   Model 1  Model 2  Model 3  Model 4  Pro public preference  2.95*** (0.46)  3.67*** (0.70)  1.96*** (0.42)  1.88** (0.66)  Sector revealed (treatment)  1.92*** (0.41)  2.09*** (0.45)  −1.71*** (0.47)  −1.49** (0.49)  Pro-public preference × Sector revealed  −5.55*** (0.65)  −6.15*** (0.73)  3.96*** (0.81)  3.83*** (0.85)  Intercept  −1.04*** (0.28)  −0.88 (1.10)  −0.80** (0.30)  −0.22 (1.02)  LR χ2  130.46***  157.39***  160.21***  155.83***  N  487  443  501  462  Controls included  No  Yes  No  Yes  Data  Group 1 + 3  Group 1 + 3  Group 2 + 4  Group 2 + 4  Best well-being  Private / B  Private / B  Public / A  Public / A    Model 1  Model 2  Model 3  Model 4  Pro public preference  2.95*** (0.46)  3.67*** (0.70)  1.96*** (0.42)  1.88** (0.66)  Sector revealed (treatment)  1.92*** (0.41)  2.09*** (0.45)  −1.71*** (0.47)  −1.49** (0.49)  Pro-public preference × Sector revealed  −5.55*** (0.65)  −6.15*** (0.73)  3.96*** (0.81)  3.83*** (0.85)  Intercept  −1.04*** (0.28)  −0.88 (1.10)  −0.80** (0.30)  −0.22 (1.02)  LR χ2  130.46***  157.39***  160.21***  155.83***  N  487  443  501  462  Controls included  No  Yes  No  Yes  Data  Group 1 + 3  Group 1 + 3  Group 2 + 4  Group 2 + 4  Best well-being  Private / B  Private / B  Public / A  Public / A  Note: Two-sided significance tests. Entries are logistic regression coefficients. Standard errors in parentheses. Dependent variable is respondents’ tendency to prioritize student wellbeing over grades when ranking school performance. Controls: Member of left-wing party (members of the Social Democrats, the Socialist People’s Party, and the Red-Green Alliance have been coded as members of a left-wing party; members of the Danish Social Liberal Party, Liberal Alliance, the Christian Democrats, the Danish People’s Party, and Venstre, Liberal Party of Denmark have been coded as not being members of a left-wing party; members of local parties and city councilors who have not indicated a party affiliation have been coded as missing); gender; age; higher education; parent of child in school, years of membership in current political party; years of membership on city council; member of finance committee; size of municipality (number of citizens in thousands); tax base per citizen in municipality; proportion of municipality students in private schools. Last three items 2014–data from www.noegletal.dk which is run by the Danish Ministry for Economic Affairs and the Interior. ***p < .001; **p < .01; *p < .05. View Large The positive relationship between preferring the public sector and prioritizing student well-being is also illustrated in Figure 3, which provides predicted probabilities of prioritizing well-being over test scores. Figure 3. View largeDownload slide Predicted Probability of Prioritizing Well-Being over Test-Scores Figure 3. View largeDownload slide Predicted Probability of Prioritizing Well-Being over Test-Scores The significant coefficients for the interaction term “pro-public preference × sector visible” provide evidence that is consistent with our hypothesis: respondents’ do engage in goal reprioritization: Table A2 in the appendix shows that using party identification as a proxy for governance preferences does not alter the substantial results. Thus, the negative coefficient for the interaction term in Model 1 shows that the relationship between preferring the public sector and giving high priority to student well-being shifts in a negative direction for respondents who learn that a public school performs better on academic achievements but worse on well-being. By contrast, the positive coefficient for the interaction term in Model 2 shows that the relationship between preferring the public sector and giving high priority to student well-being becomes even stronger when respondents learn that a public school performs better on well-being but worse on academic achievements compared to a private school. This relationship also becomes apparent in the treatment groups (group 3 and 4) in Figure 3, as the association between preferring the public sector and prioritizing student well-being turns negative when a private school performs well on well-being and a public school performs well on test scores (group 3). Furthermore, Figure 3 shows how the association becomes even more positive when a public school performs well on well-being and a private school performs well on test scores (group 4). The nature and scale of our results is most intuitively presented in Figure 4, which shows the proportions of respondents who prioritized student well-being when evaluating the performance of the schools. To illustrate how goal reprioritization occurs, we divide respondents into three groups: those with pro-public governance preferences, those with pro-private preferences, and those who are agnostic. Figure 4. View largeDownload slide Proportions Who Chose School with Highest Student Well-Being as Best Performing Figure 4. View largeDownload slide Proportions Who Chose School with Highest Student Well-Being as Best Performing Figure 4 shows how decision-makers are willing to make quite dramatic changes in their evaluative criteria in order to justify their governance preferences. This is clearest when we look at those with pro-public preferences (Figure 4A). Consistent with the information in Table 1, this group generally prioritizes student well-being as a goal preference. However, when schools’ sector affiliation is visible, it is also clear from Figure 4A that those who prefer the public sector are affected by the content of the information. Take school B in group 1 and the private school in group 3. In both cases, these schools out-perform the other school on student well-being to exactly the same degree. When respondents who strongly prefer the public sector are presented with this data but without sector information, 82% select school B as best performing. However, when told that the school is private, only 19% select this school as best performing. Simply revealing that the provider of the school is private leads many elected officials to set aside their goal preferences for student well-being. In order to justify a desired conclusion (that a public school performs better than a private one), decision-makers reprioritize their evaluative criteria by lowering the importance they place on student well-being, while redirecting the perceived importance to academic achievements instead. For the pro-public group, the tendency to find well-being important becomes even stronger when a public school is shown to perform well on student well-being and less well on academic achievements. Thus, in group 4, no less than 94% of those who strongly prefer the public sector point to the public school as the school that performed best, compared to 71% in group 2, where the politicians did not receive information about the schools’ sector affiliation. Turning to respondents who strongly prefer the private sector (Figure 4C) the evidence is again supportive of our hypothesis, though weaker as the effects of providing information on schools’ sector affiliation is only marginally significant. Thus, it seems that people who prefer the private sector are affected by our sector cue, but not to the same degree than those who prefer the public sector in our study. The finding of weaker results among people with pro-private preferences is consistent with prior research (Baekgaard and Serritzlew 2016; Baekgaard et al. 2017). Our design does not allow for a test of possible reasons for this difference, but there are a number of intriguing possibilities that merit additional investigation. One is that governance preferences may be generally stronger among people who prefer the public than among people who prefer the private sector (see e.g., Taber and Lodge 2006 on the effects of attitude strength). Another possibility is that the strength of governance preferences is contingent on context. Education is traditionally perceived as a public function. As a result, those holding pro-public governance preferences may view public provision of education as the default setting, a status quo to be defended, thereby strengthening the intensity of their preferences. If this logic holds, those with pro-private provisions would feel more intensely about defending historically private functions from government takeover. In our case, the concerns of those with pro-public governance preferences may be further intensified by a perception that their preferences are under threat. In Denmark, as elsewhere, the percentage of students attending private schools has been on the increase, going from 12.9% in 2004 to 16.7% in 2014, the year of our survey.8 Another possibility is that strong pro-public preferences are more likely in countries like Denmark, which feature both a large public sector and relatively high trust in government. The range of possible explanations call for further research to better understand how governance preferences generate symmetrical or asymmetrical responses, and under what circumstances. Another way of considering our hypothesis is to see what happens to those who are less motivated to reach certain conclusions, that is, those who declare themselves to be agnostic about public or private provision. To behave consistently with our hypothesis, this group should not alter their evaluative criteria even when provided with information about the schools’ sector affiliations. The results from Figure 4B fit with these expectations. There is not a clear directional or large effect of providing sector information to those who say they are agnostic about sector. This finding not only provides support for our hypothesis; it also suggests that those without strong prior governance preferences are more likely to consistently apply goal preferences to evaluate public sector performance. Discussion This article offers a number of theoretical contributions. First, we map insights from motivated reasoning theory to the contours of public management, examining how they play out in conditions of goal conflict. Second, we fashion and test useful concepts for our theoretical goals. In adapting psychological insights about goal reprioritization to the context of organizational goals, we offer a new avenue for studying how political and organizational actors are willing to reweight the importance they assign to policy goals relative to governance preferences. Whereas researchers of voting behavior observe that partisan alignments trump policy issues (Groenendyk 2013), we offer evidence that governance preferences trump goal preferences in the context of public management and decision-making. The findings are not just of theoretical interest—they have substantive importance. Our results undercut the idea that goal preferences are stable or dominant forces in shaping assessments of organziational performance; instead they are subservient to deeper ideological preferences about governance. In our study, political principals are motivated more by normative beliefs about the appropriate nature of the public sector than by a consistent set of goal preferences when using performance data to evaluate providers of public services. Goal preferences, applied consistently, provide goal clarity for public organizations, and some form of performance-based accountability to the public. Governance preferences do not—they devalue both the notion of consistency in policy goals, and the notion of performance more generally, in favor of one vision of governance over another. In the most basic of terms, if managers understand that governance preferences matter more than performance or goal preferences, they have little reason to expect that their efforts to improve performance will count for much. The results raise an intruiging question for practice: is it possible—or even desirable—to enforce consistency in goal preferences? It might be possible to design mechanisms to limit goal repriortization. For example, political principals might be forced to publicly rank goal preferences in a predecisional phase, and then be reminded of those commitments when data becomes available. Such a mechanism might curb the willingness of policymakers to selectively reweight goal preferences to fit with governance preferences, enforcing greater consistency in commitments to goals. But some might see such commitment mechanisms as overly technocratic, robbing policymakers of discretion to incorporate a consideration of governance preferences that they were also elected to pursue. The theory and findings both offer insight and complicate the study of performance information use. They add to a growing body of work that suggests that ideology matters. But while that prior work has offered empirical evidence for the claim that motivated reasoning alters how people interpret performance data, we show something different: that performance data will be used to justify switching between policy goals to justify governance preferences. Goal conflict is therefore convenient for motivated reasoners with strong governance preferences, in that it gives them an opportunity to make a data-based jutification for a predetermined ideological end. The claim that decision-makers will selectively use performance data to justify their arguments is not entirely new (Moynihan 2008), but we offer a specific theoretical logic for how such a process happens, as well as a compelling test for this claim. Our approach has some limitations. Because we rely on a between-subjects design we cannot definitively prove that the difference between groups is because of the goal reprioritization mechanism we specify. A within-subjects design could test the causal mechanism more conclusively. As with every experiment, we are limited in external generalizability because of the need to simplify conditions to identify causal processes. More research is needed in other settings and other policy areas where governance preferences might work differently, and with other groups of decision-makers. For example, we cannot assume that our study of elected officials extends to bureaucrats, who may approach decisions quite differently (Meier and O’Toole 2006). However, our survey experiment provides some value to practice in two ways. First, it engages actual elected officials with plausible performance data and goal preferences. Within the necessarily artificial nature of experiments, the design is not unrealistic. Second, the design is valuable precisely because it deliberately incorporates more of the complexity of actual decision scenarios than has been the case in other studies of how performance data is used. The theoretical conditions we sketch out—multiple measures tied to competing goal preferences, and decision-makers with strong governance preferences—are both more realistic and more complex than experiments that examine how individuals process only a single piece of performance data around a single goal. While in actual practice there will be even more competing goal preferences and governance preferences, our theory is premised on acknowledging this complexity in a way that is absent from much of prior research. In practice, we might assume that having a greater array of performance data, and more competing goal preferences would make the process of goal reprioritization even more likely to occur. The results add nuance to the debate about the relative virtues of simple versus comprehensive performance data systems. One implication of our result is that the provision of more information will not necessarily improve goal-based learning. We noted that the level of ambiguity in performance information can be expected to increase as more performance indicators are included in a performance measurement system. As more data creates goal conflict, it also opens up the opportunity for a biased use of data. Goal reprioritization becomes one additional reason why more data does not actually generate agreement, instead reinforcing disagreement in polarized settings. However, our results also show that those without strong governance preferences do rely on goal preferences, suggesting that performance data may be used with greater consistency in institutions—such as bureaucracies—where governance preferences can be muted (Moynihan 2008). While our study is largely an exercise in adapting relevant theory from political psychology to the context of public administration, it also offers insights of broader interest to psychology and motivated reasoning. First, the results help to expand the scope conditions of issue reprioritization, moving beyond its original setting of voters using ideology to reweight the importance of competing policy issues, to incorporate a new setting of elected officials engaged in performance evaluation within a single policy area. Second, the results shed light on whether expertise and commitment can overcome motivated reasoning. Much of political psychology and psychology relies on surveys of students or members of the public on issues where they may have marginal knowledge or commitment. By contrast, we examine a context with subjects with deeper policy expertise and more crystalized goal preferences. We show that even under these relatively demanding conditions, processes of reprioritization still hold. Conclusion Processes of what we call goal reprioritization are salient in psychology but only hinted at in public administration. In this article, we show that political decision-makers use goal reprioritization as a strategy for motivated reasoning when interpreting ambiguous information about the performance of organizations. Elected officials change the evaluative criteria they use to assess performance depending on what the data actually says about the outcome. In more concrete terms, political decision-makers seem to “shop around” for the dimensions of performance that can best be used to justify their prior governance preferences regarding public versus private service provision. In particular, those who support public provision of services use the dimensions of performance that portray public schools in the most favorable light. Given the lack of prior attention to processes of goal reprioritization, there is much additional work that could be pursued, some of which we have already discussed. Perhaps most obviously, we do not observe if these processes emerge through deliberate strategy or unconcious bias. The answer to this question might have implications in terms of designing interventions to debias decision-makers’ reasoning. Additional work could also usefully examine if the insights we develop about elected officials applying goal reprioritization under conditions of goal conflict extend to other groups or conditions. The original theory of issue reprioritization was based on citizens, and so it should be expected to be salient for citizens assessing outcomes. On the other hand, we have no evidence on whether public managers engage in goal reprioritization. Future research should test if similar patterns of goal repriortization hold among career officials. The concept of governance preferences is broad, but has particular salience to public settings, qualities that make it an example of a concept ripe for deeper theoretical development to bridge disciplinary research and public administration, while also making it useful for empirical testing of behavioral implications (James, Jilke, and Van Ryzin 2017). There is also significant room to consider how other forms of governance preferences—such as attitudes toward public employment, politicization, transparency, the size of government, or other public service values—might drive goal reprioritization or other preferences or decisions. References Baekgaard, Martin, Julian Christensen, Casper M. Dahlmann, Asbjørn Mathiasen, and Niels B. G. Petersen. 2017. The role of evidence in politics: Motivated reasoning and persuasion among political decision-makers. British Journal of Political Science . doi: 10.1017/S0007123417000084 Baekgaard, Martin, and Søren Serritzlew. 2016. Interpreting performance information: Motivated reasoning or unbiased comprehension. Public Administration Review  76: 73– 82. Google Scholar CrossRef Search ADS   Behn, Robert D. 2001. Rethinking democratic accountability . Washington, DC: Brookings Institution Press. Bischoff, Ivo, and Frédéric Blaeschke. 2016. Performance budgeting: Incentives and social waste from window dressing. Journal of Public Administration Research and Theory  26: 344– 58. Google Scholar CrossRef Search ADS   Bisgaard, Martin. 2015. Bias will find a way: Economic perceptions, attributions of blame, and partisan-motivated reasoning during crisis. The Journal of Politics  77: 849– 60. Google Scholar CrossRef Search ADS   Bohte, John, and Kenneth J. Meier. 2000. Goal displacement : Assessing the motivation for organizational cheating. Public Administration Review  60: 173– 82. Google Scholar CrossRef Search ADS   Boyne, George A. 2002. Concepts and indicators of local authority performance: An evaluation of the statutory frameworks in England and Wales. Public Money & Management  22: 17– 24. Google Scholar CrossRef Search ADS   Christensen, Julian. 2017. Justification requirements as a debiasing measure against motivated interpretations of factual information? In 2017 Annual Meeting of American Political Science Association (APSA) . San Fransisco. Chun, Young Han, and Hal G. Rainey. 2005. Goal ambiguity and organizational performance in U.S. Federal agencies. Journal of Public Administration Research and Theory  15: 529– 57. Google Scholar CrossRef Search ADS   Festinger, Leon. 1957. A theory of cognitive dissonance . Stanford Univ. Press. Goodsell, Charles T. 2004. The Case for Bureaucracy: A Public Administration Polemic . 4th ed. Washington, DC: CQ Press. Government of Denmark. 2013. Agreement between the Danish Government (the Social Democrats, the Social-Liberal Party and the Socialist People’s Party), the Liberal Party of Denmark and the Danish People’s Party on an improvement of standards in the Danish public school (primary and lower secondary education). http://eng.uvm.dk/-/media/filer/uvm/english/pdf/131007-folkeskolereformaftale-eng-red.pdf?la=en (accessed January 19, 2018). Groenendyk, Eric W. 2013. Competing motives in the partisan mind: How loyalty and responsiveness shape party identification and democracy . New York: Oxford Univ. Press. Google Scholar CrossRef Search ADS   Heinrich, Carolyn J. 2007. Evidence-based policy and performance management. The American Review of Public Administration  37: 255– 77. Google Scholar CrossRef Search ADS   Holmstrom, Bengt, and Paul Milgrom. 1991. Multitask principal-agent analyses: Incentive contracts, asset ownership, and job design. Journal of Law, Economics, & Organization  7 (Special Issue: [Papers from the Conference on the New Science of Organization, January 1991]): 24– 52. Google Scholar CrossRef Search ADS   Holzer, Marc, and Kaifeng Yang. 2004. Performance measurement and improvement: An assessment of the state of the art. International Review of Administrative Sciences  70: 15– 31. Google Scholar CrossRef Search ADS   Hvidman, Ulrik, and Simon Calmar Andersen. 2016. Perceptions of public and private performance: Evidence from a survey experiment. Public Administration Review  76: 111– 20. Google Scholar CrossRef Search ADS   Jacobsen, Rebecca, Jeffrey W. Snyder, and Andrew Saultz. 2015. Understanding satisfaction with schools: The role of expectations. Journal of Public Administration Research and Theory  25: 831– 48. Google Scholar CrossRef Search ADS   James, Oliver, and Gregg G Van Ryzin. 2017. Motivated reasoning about public performance: An experimental study of how citizens judge obamacare. Journal of Public Administration Research and Theory  27: 197– 209. Google Scholar CrossRef Search ADS   James, Oliver, Sebastian R. Jilke, and Gregg G. Van Ryzin (eds). 2017. Experiments in public management research: Challenges and contributions . Cambridge: Cambridge University Press. Google Scholar CrossRef Search ADS   Jerit, Jennifer, and Jason Barabas. 2012. Partisan perceptual bias and the information environment. The Journal of Politics  74: 672– 84. Google Scholar CrossRef Search ADS   Julnes, Patria de Lancer. 2011. Performance measurement beyond instrumental use. In Performance information in the public sector: How it is used , ed. Wouter van Dooren and Steven van de Walle, 62– 75. New York: Palgrave Macmillan. Kahan, Dan M. 2016. The politically motivated reasoning paradigm, Part 2: Unanswered questions. In Emerging trends in the social and behavioral sciences: An interdisciplinary, searchable, and linkable resource, 1–15. John Wiley & Sons. http://onlinelibrary.wiley.com/doi/10.1002/9781118900772.etrds0418/abstract (accessed January 19, 2018). Kelman, Steven, and John N. Friedman. 2009. Performance improvement and performance dysfunction: An empirical examination of distortionary impacts of the emergency room wait-time target in the English National Health Service. Journal of Public Administration Research and Theory  19: 917– 46. Google Scholar CrossRef Search ADS   Klar, Samara, and Yanna Krupnikov. 2016. Independent politics: How American disdain for parties leads to political inaction . New York: Cambridge Univ. Press. Google Scholar CrossRef Search ADS   Kroll, Alexander. 2015. Drivers of performance information use: Systematic literature review and directions for future research. Public Performance & Management Review  38: 459– 86. Google Scholar CrossRef Search ADS   Kunda, Ziva. 1987. Motivated inference: Self-serving generation and evaluation of causal theories. Journal of Personality and Social Psychology  53: 636– 47. Google Scholar CrossRef Search ADS   Kunda, Ziva. 1990. The case for motivated reasoning. Psychological Bulletin  108: 480– 98. Google Scholar CrossRef Search ADS PubMed  Lavine, Howard G., Christopher D. Johnston, and Marco R. Steenbergen. 2012. The ambivalent partisan: How critical loyalty promotes democracy . New York: Oxford Univ. Press. Google Scholar CrossRef Search ADS   Lodge, Milton, and Charles S. Taber. 2000. Three steps toward a theory of motivated political reasoning. In Elements of reason , edited by Arthur Lupia, Mathew D McCubbins, and Samuel L Popkin, 183– 213. Cambridge: Cambridge University Press. Google Scholar CrossRef Search ADS   Marvel, John D. 2016. Unconscious bias in citizens’ evaluations of public sector performance. Journal of Public Administration Research and Theory  26: 143– 58. Meier, Kenneth J., and Laurence J. O’Toole. 2006. Bureaucracy in a Democratic State: A Governance Perspective . Baltimore, MD: Johns Hopkins University Press. Meier, Kenneth J., Robert J. Wrinkle, and J. L. Polinard. 1999. Representative bureaucracy and distributional equity: Addressing the hard question. The Journal of Politics  61: 1025– 39. Google Scholar CrossRef Search ADS   Moynihan, Donald P. 2008. The dynamics of performance management: Constructing information and reform . Washington, DC: Georgetown University Press. Moynihan, Donald P. 2015. Uncovering the circumstances of performance information use findings from an experiment. Public Performance & Management Review  39: 33– 57. Google Scholar CrossRef Search ADS   Moynihan, Donald P., and Ivor Beazley. 2016. Toward next-generation performance budgeting. Lessons from the experiences of seven reforming countries . Washington, DC: World Bank. Google Scholar CrossRef Search ADS   Moynihan, Donald P., Sergio Fernandez, Soonhee Kim, Kelly M. LeRoux, Suzanne J. Piotrowski, Bradley E. Wright, and Kaifeng Yang. 2011. Performance regimes amidst governance complexity. Journal of Public Administration Research and Theory  21 ( Supp 1): 141– 55. Google Scholar CrossRef Search ADS   Nielsen, Poul A., and Donald P. Moynihan. 2017. How do politicians attribute bureaucratic responsibility for performance? Negativity bias and interest group advocacy. Journal of Public Administration Research and Theory  27: 269– 83. Rainey, Hal G., and Chan Su Jung. 2015. A conceptual framework for analysis of goal ambiguity in public organizations. Journal of Public Administration Research and Theory  25: 71– 99. Google Scholar CrossRef Search ADS   Serritzlew, Søren. 2006. Linking budgets to activity: A test of the effect of output-purchase budgeting. Public Budgeting & Finance  26: 101– 20. Google Scholar CrossRef Search ADS   Simon, Linda, Jeff Greenberg, and Jack Brehm. 1995. Trivialization: The forgotten mode of dissonance reduction. Journal of Personality and Social Psychology  68: 247– 60. Google Scholar CrossRef Search ADS   Taber, Charles S., and Milton Lodge. 2016. The illusion of choice in democratic politics: The unconscious impact of motivated political reasoning. Political Psychology  37: 61– 85. Google Scholar CrossRef Search ADS   Taber, Charles S., and Milton Lodge. 2006. Motivated skepticism in the evaluation of political beliefs. American Journal of Political Science  50: 755– 69. Google Scholar CrossRef Search ADS   Tilley, James, and Sara B Hobolt. 2011. Is the government to blame? An experimental test of how partisanship shapes perceptions of performance and responsibility. Journal of Politics  73: 316– 30. Google Scholar CrossRef Search ADS   Van de Walle, Steven, and Alasdair Roberts. 2011. Publishing performance information: An illusion of control? In Performance information in the public sector: How it is used , edited by Wouter Van Dooren and Steven Van de Walle, 215– 30. New York: Palgrave Macmillan. Google Scholar CrossRef Search ADS   Van Dooren, Wouter. 2011. Nothing new under the sun? Change and continuity in the twentieth-century performance movements. In Performance information in the public sector: How it is used , ed. Wouter Van Dooren and Steven Van de Walle, 15– 27. New York: Palgrave Macmillan. Google Scholar CrossRef Search ADS   Van Dooren, Wouter Van, Geert Bouckaert, and John Halligan. 2010. Performance management in the public sector . Abingdon, Oxton: Routledge. Wildavsky, Aaron. 1979. Speaking truth to power: The art and craft of policy analysis . Boston: Little, Brown and Company. Google Scholar CrossRef Search ADS   Wright, Bradley E. 2004. The role of work context in work motivation: A public sector application of goal and social cognitive theories. Journal of Public Administration Research and Theory  14: 59– 78. Google Scholar CrossRef Search ADS   Footnotes 1 Jacobsen, Snyder, and Saultz (2015) use the term goal expectations for what we describe as goal preferences here. 2 Our model does not assume that tradeoffs and therefore goal conflict are inevitable: there may also be cases where performance on one goal is positively correlated with performance on another (Meier, Wrinkle, and Polinard 1999). Instead, we merely claim that such tradeoffs, and hence the condition of goal conflict, are relatively common. 3 In addition to our experiment, respondents participated in other survey experiments that were designed to test other but related research questions. However, the survey experiment reported in this article was placed before the remaining experiments in the survey, and therefore the participation in additional experiments did not have the potential to affect our results. Results from the additional experiments are reported in Baekgard et al. (2017) and Christensen (2017), which are mentioned in the text. 4 The wording of the items was: “To what extent do you agree or disagree with the following statements? (1) Many public activities could be produced both better and cheaper by private providers. (2) We should to a larger degree outsource public services (such as child care, elderly care, and hospital treatments). (3) The public sector is best at providing public services.” Possible answers were “completely agree,” “partly agree,” “neither agree nor disagree,” “partly disagree,” “completely disagree,” and “don’t know.” A factor analysis revealed factor scores above 0.8 for all items and an index was therefore constructed ranging from 0 to 1, where higher values correspond to a stronger preference for the public sector. Cronbach’s alpha was 0.92. Mean = 0.62 and standard deviation = 0.33. 5 In Denmark, private schools are financed through a combination of user charges and government subsidies whereas public schools are financed through municipal taxes (Serritzlew 2006: 108). Furthermore, private schools differ from public schools by being operated by school boards with greater autonomy regarding approaches to learning and pedagogy. The autonomy of public school boards is more limited and depends on delegation from the city council (ibid.). 6 There was enough of an ongoing concern about such tensions that in 2013, shortly before our data collection, a majority of the Danish national parliament mandated, as a part of a major reform of the Danish public school system, that each school collect student well-being data in addition to test scores (Government of Denmark 2013). Thus, in the policy setting we study, the tensions between the two dimensions of performance are both plausible and salient. 7 The use of cognitively undemanding data is different from analyses such as Baekgaard et al. (2017) and Baekgaard and Serritzlew (2016) who ask respondents to compute performance using more complex data. For our purposes, making the performance data as clear as possible is beneficial because it excludes cognitive complexity as a reason why policymakers struggle to evaluate the performance of schools. 8 Data from the Danish Ministry for Economic Affairs and the Interior (www.noegletal.dk). Appendix Table A1. Balance Checks of Randomization: Each Group Against Rest of Sample   Group 1  Group 2  Group 3  Group 4  Women (pct)  32.8  32.6  28.6  27.1  Age (years, mean)  51.9  53.1  52.7  50.9  Pro-public preference (mean)  0.61  0.67*  0.62  0.60.  Higher education (pct)  71.9  70.2  67.2  69.3  Parent to child in school (pct)  25.7  32.9  33.2  28.3  Years of membership in current party (mean)  24.3  26.9  24.8  26.7  Years of membership in city council (mean)  20.8  21.5  22.3  21.3  Member of Finance Committee (pct)  35.2  29.1  33.2  32.0  Member of left-wing partya (pct)  45.6  55.6**  45.0  46.1  Proportion of municipality’s students who are in private schoolsb (mean pct)  16.1  16.6  15.8  16.9  Tax base per citizenc (mean)  172.5  171.9  171.3  176.3*  Size of municipalityd (mean)  67.0  73.9  62.8  61.1    Group 1  Group 2  Group 3  Group 4  Women (pct)  32.8  32.6  28.6  27.1  Age (years, mean)  51.9  53.1  52.7  50.9  Pro-public preference (mean)  0.61  0.67*  0.62  0.60.  Higher education (pct)  71.9  70.2  67.2  69.3  Parent to child in school (pct)  25.7  32.9  33.2  28.3  Years of membership in current party (mean)  24.3  26.9  24.8  26.7  Years of membership in city council (mean)  20.8  21.5  22.3  21.3  Member of Finance Committee (pct)  35.2  29.1  33.2  32.0  Member of left-wing partya (pct)  45.6  55.6**  45.0  46.1  Proportion of municipality’s students who are in private schoolsb (mean pct)  16.1  16.6  15.8  16.9  Tax base per citizenc (mean)  172.5  171.9  171.3  176.3*  Size of municipalityd (mean)  67.0  73.9  62.8  61.1  Note: Significant differences are identified by comparing each group with the rest of the sample using two-tailed t-tests and tests of proportions. aMembers of the Social Democrats, the Socialist People’s Party, and the Red-Green Alliance have been coded as members of a left-wing party. Members of the Danish Social Liberal Party, Liberal Alliance, the Christian Democrats, the Danish People’s Party, and Venstre, Liberal Party of Denmark have been coded as not being members of a left-wing party. Members of local parties and city councilors who have not indicated a party affiliation have been coded as missing. bProportion of all primary school students in a respondent’s municipality who was in a private school (2014 numbers drawn from www.noegletal.dk which is run by the Danish Ministry for Economic Affairs and the Interior). 3Tax base per citizen measures the 2014 tax base (in thousands of Danish Kroner) per citizen in a respondent’s municipality (numbers drawn from www.noegletal.dk). 4Size of municipality measures the number of citizens (in thousands) in a respondent’s municipality as of January 1, 2014 (numbers drawn from www.noegletal.dk). ***p < .001; **p < .01; *p < .05. View Large Table A2. Influence of Governance Preferences on Assessments of Performance, Using Party ID Instead of Stated Preferences   Model 1  Model 2  Left winga  1.69*** (0.32)  0.89** (0.27)  Sector revealed (treatment)  0.03 (0.25)  −0.62* (0.26)  Left wing × Sector revealed  −3.41*** (0.45)  3.03*** (0.60)  Intercept  −3.67e−16 (0.18)  −0.02 (0.19)  LR χ2  110.51***  126.90***  N  468  491  Data  Group 1 + 3  Group 2 + 4  Best well-being  Private / B  Public / A    Model 1  Model 2  Left winga  1.69*** (0.32)  0.89** (0.27)  Sector revealed (treatment)  0.03 (0.25)  −0.62* (0.26)  Left wing × Sector revealed  −3.41*** (0.45)  3.03*** (0.60)  Intercept  −3.67e−16 (0.18)  −0.02 (0.19)  LR χ2  110.51***  126.90***  N  468  491  Data  Group 1 + 3  Group 2 + 4  Best well-being  Private / B  Public / A  Note: Two-sided significance tests. Entries are logistic regression coefficients. Standard errors in parentheses. Dependent variable is respondents’ tendency to prioritize student wellbeing over grants when ranking school performance. aMembers of the Social Democrats, the Socialist People’s Party, and the Red-Green Alliance have been coded as members of a left-wing party (left wing = 1). Members of the Danish Social Liberal Party, Liberal Alliance, the Christian Democrats, the Danish People’s Party, and Venstre, Liberal Party of Denmark have been coded as not being member of a left-wing party (left wing = 0). Members of local parties and city councilors who have not indicated a party affiliation have been coded as missing. ***p < .001; **p < .01; *p < .05; †p < .1. View Large © The Author(s) 2018. Published by Oxford University Press on behalf of the Public Management Research Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Public Administration Research and Theory Oxford University Press

How Do Elected Officials Evaluate Performance? Goal Preferences, Governance Preferences, and the Process of Goal Reprioritization

Loading next page...
 
/lp/ou_press/how-do-elected-officials-evaluate-performance-goal-preferences-1uD0D4xTkK
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Public Management Research Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
1053-1858
eISSN
1477-9803
D.O.I.
10.1093/jopart/muy001
Publisher site
See Article on Publisher Site

Abstract

Abstract Performance data allows politicians to exert accountability over public organizations, even as ideological biases can affect how they interpret such data. However, we know little about how motivated decision-makers prioritize goals when facing multiple pieces of contradictory performance data that reflect the competing goals of public services. Such goal conflict is an inherent aspect of public management. To understand its implications for the use of performance data, we develop a theory of goal reprioritization. We start by assuming that elected officials have preferences between specific policy goals, and about governance processes—such as a preference for public or private service provision. When elected officials face contradictory pieces of performance data, governance preferences drive performance evaluations to the point that they are willing to reweight their goal preferences to minimize cognitive dissonance. We offer experimental evidence of this process, showing that elected officials asked to evaluate school performance reprioritize between two distinct policy goals for schools—test scores and student well-being—to fit with their governance preferences. Reprioritization is an attractive strategy since it allows elected officials to claim they are using performance data, even as underlying governance preferences lead them to set aside the evaluative goal-based criteria by which they would otherwise make performance evaluations. In other words, preferences concerning the nature of government can trump goal preferences when decision-makers use performance data. Introduction We are in “the age of quantified performance” (Van de Walle and Roberts 2011, 215) reflecting the ongoing ubiquity of performance management in governance (van Dooren 2011). Public organizations set goals and collect large amounts of data to evaluate whether these goals have been reached, in the hope of allowing elected officials to allocate resources more effectively, prioritize between different programs and optimize organizational processes (for critical assessments, see Julnes 2011; Kroll 2015; van Dooren et al. 2010). The same underlying logic on how data can improve outcomes also motivates the evidence-based policymaking movement (Heinrich 2007). A truism of the performance movement is that “what gets measured gets managed”, meaning that unmeasured goals are subject to inattention and goal displacement. While the risks of a unidimensional performance system are well documented, as we discuss in the next section, we know much less about how decision-makers manage multidimensional performance data. Even a relatively limited array of data can reflect goal conflict, where multiple performance indicators contradict each other, undermining the potential for a single consistent image of performance to emerge. Since organizations make tradeoffs between goals, goal conflict can be expected to be routine (Chun and Rainey 2005; Hvidman and Andersen 2016; Rainey and Jung 2015; Wright 2004). Political principals can navigate goal conflict if they have a clear and consistent preference for the policy goals they hope to maximize, and manage public organizations accordingly. However, ideological beliefs may undercut the ability to interpret data about policy with consistency (Kunda 1990; Taber and Lodge 2006). An important insight on this point comes from Baekgaard and Serritzlew (2016, 73), who argue that “even objective, clear and unambiguous performance information is subject to biased interpretation depending on whether the information is consistent with the prior beliefs held by those who receive the information.” They asked subjects to rank hospital performance using simple objective measures of operations, and showed that people with strong pro-market or pro-public beliefs were more likely to make errors when the sector affiliation of the hospital was revealed. Related work confirms such tendencies among elected officials (Baekgaard et al. 2017; Nielsen and Moynihan 2017). While prior work alerts us to how political beliefs affect how people process one piece of performance data, it does not tell us the role that such beliefs play as decision-makers manage multiple pieces of performance data, and in particular, weigh tradeoffs in conditions of goal conflict. We therefore develop and test a theory that incorporates the reality of goal conflict. To do so, we adapt a specific subset of psychological theory on motivated reasoning that we characterize as goal reprioritization, which can be understood as the “opportunistic adjustment of the weight-assigned evidence conditional on its conformity” to desired conclusions (Kahan 2016, 1). We explain goal reprioritization in greater detail below, but for now we note that even as it has been recognized as “the signature feature” of identity-protective motivated reasoning (ibid.), it has not been used to explain conditions of goal conflict or other decision contexts in contemporary public administration. To adapt reprioritization theory to the complex but realistic context of elected officials with different preferences assessing public sector performance with multiple and contradictory pieces of information, it is necessary to add a conceptual distinction, not explicit in prior work, between two types of preferences: governance preferences and goal preferences. Governance preferences reflect ideological beliefs about the appropriate processes in and design of the public sector. In our study, we model governance preferences as a preference for public or private provision of services. Goal preferences are the weighted preference for one policy goal over another, and we examine goal preferences within a policy domain. For example, people may want schools to provide both good academic outcomes and well-rounded students, but weigh one goal as more important than the other. Goal preferences form evaluative criteria by which individuals can assess the performance of a service provider. Those who weigh academic achievement as more important than student well-being should do so consistently, thereby providing goal clarity to organizations via the evaluation process. However, our application of goal reprioritization hypothesizes that governance preferences can compete with goal preferences, sometimes causing decision-makers to relegate goal-based evaluative criteria in their assessment of public sector performance. We test this hypothesis using a survey experiment with responses from Danish city councilors. The elected officials were placed in multiple groups that allow both a demonstration of the salience of goal preferences and the influence of governance preferences when the officials evaluate performance data about schools. In control groups, elected officials consistently apply goal preferences as evaluative criteria. However, when they are told the sector affiliation of the schools the goal-based evaluative criteria are set aside. Instead, officials tend to assess the performance data through the lens of their governance preferences for public or private service provision. The results are substantively important as they demonstrate how governance preferences can undercut the ability of governments, even when working within a performance management framework, to clearly and consistently pursue a set of specific policy goals. Both governance and goal preferences are ideological preferences, but they serve different purposes, making it all the more important to conceptually and empirically distinguish their salience for public management. Elected officials may establish a principal-agent chain of accountability between themselves, voters and bureaucrats on a set of promises about both governance and policies. Our theory proposes that commitments to policy goals are fluid enough to be contingent upon assumptions about the administrative machinery in place, and when those assumptions are violated, ideological actors are willing to re-weight their goal preferences. While other work has convincingly made the case that ideology matters in how public performance is evaluated, we add nuance to this claim: ideology matters, but ideology about governance may matter more than ideology about policy goals. In the following sections, we build our argument by drawing from a mix of literatures. First, we review research on performance management and goal ambiguity to make the case that goal conflict is a common feature of public administration. We next examine research that shows that ideological beliefs matter in how people process performance data, and draw from the broader literature on motivated reasoning to develop the concept of goal reprioritization. These literatures provide both the theoretical strands to make our argument, while demonstrating its relevance. But across these literatures, there is no prior work that makes the specific predictions that we do. We next present results, and consider the implications and limitation of our research. Goal Conflict: A Basic Condition of Public Administration The theoretical setting in which we situate our study—goal conflict—is a frequent and salient context of public management (Chun and Rainey 2005; Wright 2004). In this section, we document how performance management systems capture goal conflict. The basic logic for performance management is clear: decision-makers with valid information about the performance of their organizations are better able to make informed, goal-oriented decision making than those who rely on intuition (Holzer and Yang 2004, 16). If having some performance data is good, a variety of perspectives make the case that having more data is better. Performance data provides some basic form of accountability to the public and to policymakers (Behn 2001). Because public organizations pursue multiple goals, a single or small number of measures fail to fully capture the diversity of public sector outcomes (Boyne 2002; Chun and Rainey 2005). More measures should therefore provide better accountability, and better reflect competing goal preferences. For example, Jacobsen, Snyder, and Saultz (2015) argue that a unidimensional approach to performance measurement runs the risk of mischaracterizing how people assess public services. They show that an assessment of school performance based only on academic performance fails to capture the goal preferences of those who prefer a well-rounded education.1 Adding more metrics is a solution to this problem, and to a variety of other observed issues with performance systems, such as the prioritization of efficiency measures over other dimensions of performance, like equity or procedural safeguards (Moynihan et al. 2011), or goal displacement as individuals increase effort on measured goals while reducing effort on unmeasured ones (Bohte and Meier 2000; Holmstrom and Milgrom 1991; Kelman and Friedman 2009; Rainey and Jung 2015). In practice, performance systems accrue measures over time, incrementally evolving into more comprehensive frameworks (Moynihan and Beazley 2016). But more comprehensive performance systems create their own problems, such as increased transaction costs (Bischoff and Blaeschke 2016). More measures are also more likely to reflect and formalize goal conflict. Chun and Rainey (2005, 2) define “direct goal conflict” as a context where “achieving one valued goal directly inhibits achieving another desired goal” while “indirect goal conflict can be observed whenever there are multiple goals at the same hierarchical level since having more goals means having a greater need to split limited resources and thereby exposes more points of potential conflict.” While the more general construct of goal ambiguity can give rise to goal conflict (Wright 2004), goal conflict specifically taps counteracting relationships between goals. For any moderately complex program, there will be multiple goals and multiple measures that reflect those goals, which frequently send conflicting signals to decision-makers. Even if multiple goals do not actually represent trade-offs, policymakers may assume that the achievement of each additional goal draws resources away from other goals.2 The sense of conflict may become stronger if performance of one goal appears to move in opposite direction to others. Goal conflict is not, in itself, a bad thing. It may lead to more difficult decision processes, but perhaps also better decisions, reflecting a fuller appreciation of the nuances of a program (Moynihan 2015). But goal conflict also begets discretion in how to evaluate programs, and individual preferences will influence the use of that discretion. While prior work has suggested that goal conflict can be expected to be a common condition in government, this still leaves open the question of how preferences affect how individuals weigh data about conflicting goals. The following section makes use of psychological literature about information processing to offer a theoretical model to incorporate how ideological beliefs matter in conditions of goal conflict. Preferences and Performance Information Use The literature on motivated reasoning provides a basis for understanding the role of attitudes and beliefs when decision-makers process performance data. People are often motivated to “justify a specific, preselected conclusion” (Lodge and Taber 2000, 186). To do so, they actively seek to interpret information in ways that support desired conclusions (e.g., Groenendyk 2013; Jerit and Barabas 2012; Kunda 1987, 1990; Lavine et al. 2012; Taber and Lodge 2006, 2016). For instance, voters whose party is in charge of government are more forgiving of government failures, and blame factors others than the actions of the government when explaining negative outcomes (Bisgaard 2015; Tilley and Hobolt 2011). A nascent literature has begun to experimentally demonstrate how attitudes and beliefs matter to the selection and processing of performance information about public organizations. Moynihan (2008, 155) hypothesized that: “that the creation, presentation, and use of performance data in the public sphere will reflect the values of those involved, be they partisan, organizational, institutional, or other.” As noted in the introduction, Baekgaard and Serritzlew (2016, 73) offer empirical evidence for this claim by showing that pro-market or pro-public beliefs affect how people process performance data. James and Van Ryzin (2017) show that priming people to think politically affects how they evaluate the usefulness of data when assessing the US Patient Protection and Affordable Care Act, which was passed by the Democratic party. For example, politically primed Democrats evaluate data supportive of the Act as stronger evidence. Elected officials have greater policy expertise than average voters, but are not immune to preferences affecting their judgments. Indeed, they seem to resist attempts to debias faulty evaluations through mechanisms like being asked to justify interpretations of information (Christensen 2017) or through provision of additional evidence at odds with the officials’ governance preferences (Baekgaard et al. 2017). Partisan beliefs also seem to affect who politicians listen to when using performance data for decision making. Thus, Nielsen and Moynihan (2017) found that liberal elected officials were responsive to interpretations of performance data provided by an advocate (a union representative) who shared their ideological beliefs, while conservatives ignored this advocate. This article adds to this growing literature on the psychology of performance information use by adapting theory from psychology to model how preferences affect decision-makers’ weighting of multiple pieces of contradictory evidence about competing policy goals. To do so, it becomes necessary to make explicit some distinctions hinted at in prior empirical work, which has framed ideology as a factor that affects how performance data is processed (e.g., Baekgaard and Serritzlew 2016; Hvidman and Andersen 2016), without accounting for how performance measures themselves also reflect ideological preferences for some goals over others. This becomes more apparent when we attempt to model how decision-makers manage multiple performance goals rather than just one, which allows them to weight data on their preferred goal more heavily than data on other goals. Thus, we expect that goal preferences form evaluative criteria in cases of goal conflict. We also need to distinguish between governance preferences and goal preferences. Goodsell (2004) argues that preferences about governance have been taught to citizens as a coherent package of beliefs, which in recent decades have been defined in negative terms that government agencies, lacking an element of competition, inherently drift into inefficiency and inflexibility, rewarding employees with overly generous terms at the expense of responsiveness to the public or politicians. This results in a preference for less government over more, and private provision over public. Governance preferences may be so ingrained as to form an unconscious bias about public organizations (Marvel 2016). For example, people’s stereotypes of public and private sectors may lead them to assume that private organizations perform better on measures of efficiency or managerial flexibility (Hvidman and Andersen 2016; Marvel 2016). As we explain in the next section, the stability of such governance preferences may affect how people prioritize between conflicting goal preferences about policy. Reprioritization as a Strategy Psychologists have long been aware of a human tendency to downplay the importance of psychologically unpleasant information while increasing the perceived importance of pleasant information. Festinger (1957) described how people often engage in retrospective reweighting of the relative importance they assign to different thoughts in order to reduce cognitive dissonance. Festinger used this insight to explain how people behave after making choices between competing alternatives (e.g., buying certain consumer items instead of others). In such situations, people are motivated to convince themselves that they have made good choices. An effective strategy for doing so, is to “magnify the importance of the good points associated with the chosen alternative and to think of new advantages that [one] hadn’t thought of before” (Festinger 1957, 45), while downplaying disadvantages with the chosen options and advantages with unchosen ones. Simon et al. (1995) use the term “trivialization” to describe an equivalent reasoning strategy. More recently, Groenendyk (2013) has applied similar logic to understand politically motivated reasoning among voters, using the term “issue reprioritization” to describe how partisans systematically alter the importance they assign to different issues depending on what makes their own political party look best (Groenendyk 2013, 49). As a reasoning strategy, reprioritization is psychologically appealing. It is simple to execute but opaque enough to be defensible. More blatant forms of motivated reasoning, such as ignoring or misinterpreting information, leave people at risk of looking irrational and closedminded, unattractive personal traits we generally seek to avoid (Klar and Krupnikov 2016). Reprioritization, on the other hand, allows people to maintain an “appearance of objectivity” (Groenendyk 2013, 50) by acknowledging the existence of inconvenient information without having to make politically inconvenient conclusions. Thus, people can leverage the ambiguity of the information environment to come up with reasonable arguments to support desired conclusions. We adapt issue reprioritization to the public management context of competing goal preferences, and we therefore use the term goal reprioritization. There are some differences in the application of the two concepts. Issue reprioritization has been primarily used to explain how partisans reweight the importance of individual issues when given information. However, the literature on issue reprioritization does not tell us whether such processes also apply to elected officials, the processing of performance data, and the weighting of different goals in the same policy area. As such, this is an adaption rather than a straightforward application of the theory. Parallels to goal reprioritization can be found in Wildavsky’s description of “retrospection” (Wildavsky 1979), a tendency to “attribute new motivational meaning to what we have done” (ibid., 136) and to “reformulate our problems” (ibid., 137) in ways that allow us to “make what we have done conform to reason as we understand it” (ibid.). However, while the “opportunistic adjustment of the weight-assigned evidence conditional on its conformity to” desired conclusions has in psychological literature been recognized as “the signature feature” of identity-protective motivated reasoning (Kahan 2016, 1), insights about this kind of reasoning have been largely overlooked in contemporary public administration settings. This oversight is remarkable given the ready-made utility of goal reprioritization to understanding public sector conditions of multiple and competing goals. Thus, with this article we investigate the usefulness of goal reprioritization in understanding how decision-makers deal with goal conflict. We expect that decision-makers will tend to use goal reprioritization as a strategy to make performance evaluations align with their governance preferences whenever there is a conflict between these preferences and the decision-makers’ goal preferences. Hypothesis: When facing contradictory performance data, elected officials will use goal reprioritization as a strategy to make their use of performance information fit with governance preferences for public or private service provision. To illustrate how this works in practice, imagine a political principal who has a goal preference for test scores over student well-being. To send a consistent signal to agents, the principal should then use performance data to identify and reward schools that perform well on test scores. In other words, the weights in the principal’s evaluation function should be consistently determined by the principal’s goal preferences. A school that performs poorly on test scores, but well on student well-being, should be evaluated as poorly performing. This process of goal weighting is shown in the first illustration of Figure 1. Figure 1. View largeDownload slide The Process of Goal Reprioritization Figure 1. View largeDownload slide The Process of Goal Reprioritization Yet, if the political principal also has a governance preference for public provision of education, and the data shows public schools perform better on student well-being rather than test scores, our hypothesis predicts that the political principal will interpret the performance data in a way that favors public schools. Specifically, we expect the principal to reprioritize the goals by increasing the weight assigned to student well-being and reducing the weight assigned to test-scores, consistent with the second illustration of Figure 1. The next section explains how we test this hypothesis in greater detail. Analytical Strategy Our theory examines the influence of decision-makers’ governance preferences and goal preferences in the interpretation of ambiguous performance information. In order to statistically test this, we need data on how a large number of decision-makers with varying governance preferences and goal preferences interpret comparable pieces of performance information. Specifically, in order to test our expectation of goal reprioritization, we need data allowing for an investigation of a relationship between X: (in)congruence between governance preferences and goal preferences in performance information and Y: to what extent decision-makers change evaluative criteria when evaluating the performance data. This is difficult to do with an observational model with actual decision-makers, because we cannot control or fully observe the information presented to them. We are also interested in how actual decision-makers respond to information. We therefore sample actual elected officials: Danish city councilors who make decisions about core welfare services like public schools, employment activities, childcare, and elderly care.3 We conducted our survey experiment in November and December 2014. Emails with invitations to participate were sent to each of Denmark’s 2,445 city councilors on 18th of November and when the data collection was closed on 19th of December, 988 city councilors had provided answers to our experimental items. Thus, our study has a response rate of 40.49%. The decision-makers we survey are at the frontline of the debate about public and private service provision, since they regularly decide whether or not to contract out specific services. Our analytic strategy assesses relationships between decision-makers’ governance preferences and their interpretations of performance data by comparing situations where governance preferences are, and are not, triggered. Following Baekgaard and Serritzlew (2016), governance preferences are captured by asking respondents about their preferences for public or private delivery of public services.4 We then conduct a survey experiment where we ask respondents to respond to experimentally varied performance information about a pair of fictional schools. Some are presented to information where governance preferences are not triggered (the schools are called “school A” and “school B”) while others are told that the schools are public and private.5 Information is provided about performance on two different performance indicators: test scores and student well-being. While both are desirable, these represent two distinct goals, often thought to be in conflict with one another. For example, an over-emphasis on academic achievement is often claimed to come at the expense of broader student well-being (Jacobsen, Snyder, and Saultz 2015).6 We then ask respondents to assess the performance of the schools. Our survey experiment is summarized in Figure 2. Respondents were randomly assigned to one of the four conditions in the figure. As can be seen, the respondents were presented to information that was cognitively undemanding, making it easy to see which school performed best on each indicator.7 Furthermore, the information conveyed goal conflict; the school that performed best on test scores performed worst on student well-being, and the other way around. Figure 2. View largeDownload slide Experimental Groups Figure 2. View largeDownload slide Experimental Groups Groups 1–2 serve as control groups allowing us to measure goal preferences in absence of public or private sector information. Thus, for these groups, no information was given that could be expected to trigger governance preferences, only data about school outcomes for academic achievement and student well-being. This gives us a baseline against which we can compare the effects of governance preferences. Groups 3–4 received the same information as groups 1–2, but were also told that the schools were public and private. All else equal, if goal preferences were stable, we should see little difference between the control and treatment groups (i.e., evaluations should not be affected by information about schools’ sector affiliation). However, we expect the sector information to trigger governance preferences, causing respondents in groups 3–4 to engage in goal reprioritization in order to justify their governance preferences based on the available information. To sum up, respondents were randomly assigned to one of four groups that were all asked to indicate which of two schools performed best, based on test scores and student well-being data. Because ours is a between-subject design, one limitation is that we cannot directly observe the same set of people changing their behavior as a result of the treatment. Instead, we infer that any difference in behavior of equivalent subjects is the result of the experimental treatment. More specifically, we propose that the groups told that the schools were public and private (groups 3–4) will behave differently relative to the groups not given this information (groups 1–2). A balance test (Table A1) shows that all groups balance in terms of age, gender, education, years of membership in current party, years of membership in a city council, size of municipality, the tendency to go to private school as a primary school student in respondents’ municipalities, the proportion of the groups consisting of members of a municipal financial committee, and the proportion of the groups consisting of parents to children who go to school. Members of group 2 are slightly more in favor of the public sector than the rest of the sample, which is also reflected in more members of group 2 being members of a left-wing party. Furthermore, the average member of group 4 is from a slightly wealthier municipality (measured as tax base per citizen) than the rest of the sample. However, including controls such as party affiliation and tax base does not alter our substantial results and therefore, there is no reason to believe that our findings result from imbalances between our groups. Results We do not have clear a priori expectations about decision-makers’ goal preferences between student well-being and academic achievements in the outset (i.e., in our control groups). However, we expect that politicians who have governance preferences for the public sector will tend to give higher priority to academic achievements (test scores) in group 3 where the public school performs better on this metric, and higher priority to student well-being in group 4 compared to the corresponding control groups. The reversed pattern is expected to exist among people who prefer the private sector. Our hypothesis is tested in Table 1, where we estimate the expected moderating effect of providing information about schools’ sector affiliation on the relationship between governance preferences and goal preferences. As can be seen from the positive and significant coefficients for the “pro-public preference” variable, a positive relationship exists in the outset (i.e., among respondents in the control groups) between preferring the public sector and prioritizing student well-being over academic achievements. Thus, respondents who prefer the public sector tend to find student well-being more important than academic achievements, whereas respondents who prefer the private sector find academic achievements more important. Table 1. Influence of Governance Preferences on Assessments of Performance   Model 1  Model 2  Model 3  Model 4  Pro public preference  2.95*** (0.46)  3.67*** (0.70)  1.96*** (0.42)  1.88** (0.66)  Sector revealed (treatment)  1.92*** (0.41)  2.09*** (0.45)  −1.71*** (0.47)  −1.49** (0.49)  Pro-public preference × Sector revealed  −5.55*** (0.65)  −6.15*** (0.73)  3.96*** (0.81)  3.83*** (0.85)  Intercept  −1.04*** (0.28)  −0.88 (1.10)  −0.80** (0.30)  −0.22 (1.02)  LR χ2  130.46***  157.39***  160.21***  155.83***  N  487  443  501  462  Controls included  No  Yes  No  Yes  Data  Group 1 + 3  Group 1 + 3  Group 2 + 4  Group 2 + 4  Best well-being  Private / B  Private / B  Public / A  Public / A    Model 1  Model 2  Model 3  Model 4  Pro public preference  2.95*** (0.46)  3.67*** (0.70)  1.96*** (0.42)  1.88** (0.66)  Sector revealed (treatment)  1.92*** (0.41)  2.09*** (0.45)  −1.71*** (0.47)  −1.49** (0.49)  Pro-public preference × Sector revealed  −5.55*** (0.65)  −6.15*** (0.73)  3.96*** (0.81)  3.83*** (0.85)  Intercept  −1.04*** (0.28)  −0.88 (1.10)  −0.80** (0.30)  −0.22 (1.02)  LR χ2  130.46***  157.39***  160.21***  155.83***  N  487  443  501  462  Controls included  No  Yes  No  Yes  Data  Group 1 + 3  Group 1 + 3  Group 2 + 4  Group 2 + 4  Best well-being  Private / B  Private / B  Public / A  Public / A  Note: Two-sided significance tests. Entries are logistic regression coefficients. Standard errors in parentheses. Dependent variable is respondents’ tendency to prioritize student wellbeing over grades when ranking school performance. Controls: Member of left-wing party (members of the Social Democrats, the Socialist People’s Party, and the Red-Green Alliance have been coded as members of a left-wing party; members of the Danish Social Liberal Party, Liberal Alliance, the Christian Democrats, the Danish People’s Party, and Venstre, Liberal Party of Denmark have been coded as not being members of a left-wing party; members of local parties and city councilors who have not indicated a party affiliation have been coded as missing); gender; age; higher education; parent of child in school, years of membership in current political party; years of membership on city council; member of finance committee; size of municipality (number of citizens in thousands); tax base per citizen in municipality; proportion of municipality students in private schools. Last three items 2014–data from www.noegletal.dk which is run by the Danish Ministry for Economic Affairs and the Interior. ***p < .001; **p < .01; *p < .05. View Large The positive relationship between preferring the public sector and prioritizing student well-being is also illustrated in Figure 3, which provides predicted probabilities of prioritizing well-being over test scores. Figure 3. View largeDownload slide Predicted Probability of Prioritizing Well-Being over Test-Scores Figure 3. View largeDownload slide Predicted Probability of Prioritizing Well-Being over Test-Scores The significant coefficients for the interaction term “pro-public preference × sector visible” provide evidence that is consistent with our hypothesis: respondents’ do engage in goal reprioritization: Table A2 in the appendix shows that using party identification as a proxy for governance preferences does not alter the substantial results. Thus, the negative coefficient for the interaction term in Model 1 shows that the relationship between preferring the public sector and giving high priority to student well-being shifts in a negative direction for respondents who learn that a public school performs better on academic achievements but worse on well-being. By contrast, the positive coefficient for the interaction term in Model 2 shows that the relationship between preferring the public sector and giving high priority to student well-being becomes even stronger when respondents learn that a public school performs better on well-being but worse on academic achievements compared to a private school. This relationship also becomes apparent in the treatment groups (group 3 and 4) in Figure 3, as the association between preferring the public sector and prioritizing student well-being turns negative when a private school performs well on well-being and a public school performs well on test scores (group 3). Furthermore, Figure 3 shows how the association becomes even more positive when a public school performs well on well-being and a private school performs well on test scores (group 4). The nature and scale of our results is most intuitively presented in Figure 4, which shows the proportions of respondents who prioritized student well-being when evaluating the performance of the schools. To illustrate how goal reprioritization occurs, we divide respondents into three groups: those with pro-public governance preferences, those with pro-private preferences, and those who are agnostic. Figure 4. View largeDownload slide Proportions Who Chose School with Highest Student Well-Being as Best Performing Figure 4. View largeDownload slide Proportions Who Chose School with Highest Student Well-Being as Best Performing Figure 4 shows how decision-makers are willing to make quite dramatic changes in their evaluative criteria in order to justify their governance preferences. This is clearest when we look at those with pro-public preferences (Figure 4A). Consistent with the information in Table 1, this group generally prioritizes student well-being as a goal preference. However, when schools’ sector affiliation is visible, it is also clear from Figure 4A that those who prefer the public sector are affected by the content of the information. Take school B in group 1 and the private school in group 3. In both cases, these schools out-perform the other school on student well-being to exactly the same degree. When respondents who strongly prefer the public sector are presented with this data but without sector information, 82% select school B as best performing. However, when told that the school is private, only 19% select this school as best performing. Simply revealing that the provider of the school is private leads many elected officials to set aside their goal preferences for student well-being. In order to justify a desired conclusion (that a public school performs better than a private one), decision-makers reprioritize their evaluative criteria by lowering the importance they place on student well-being, while redirecting the perceived importance to academic achievements instead. For the pro-public group, the tendency to find well-being important becomes even stronger when a public school is shown to perform well on student well-being and less well on academic achievements. Thus, in group 4, no less than 94% of those who strongly prefer the public sector point to the public school as the school that performed best, compared to 71% in group 2, where the politicians did not receive information about the schools’ sector affiliation. Turning to respondents who strongly prefer the private sector (Figure 4C) the evidence is again supportive of our hypothesis, though weaker as the effects of providing information on schools’ sector affiliation is only marginally significant. Thus, it seems that people who prefer the private sector are affected by our sector cue, but not to the same degree than those who prefer the public sector in our study. The finding of weaker results among people with pro-private preferences is consistent with prior research (Baekgaard and Serritzlew 2016; Baekgaard et al. 2017). Our design does not allow for a test of possible reasons for this difference, but there are a number of intriguing possibilities that merit additional investigation. One is that governance preferences may be generally stronger among people who prefer the public than among people who prefer the private sector (see e.g., Taber and Lodge 2006 on the effects of attitude strength). Another possibility is that the strength of governance preferences is contingent on context. Education is traditionally perceived as a public function. As a result, those holding pro-public governance preferences may view public provision of education as the default setting, a status quo to be defended, thereby strengthening the intensity of their preferences. If this logic holds, those with pro-private provisions would feel more intensely about defending historically private functions from government takeover. In our case, the concerns of those with pro-public governance preferences may be further intensified by a perception that their preferences are under threat. In Denmark, as elsewhere, the percentage of students attending private schools has been on the increase, going from 12.9% in 2004 to 16.7% in 2014, the year of our survey.8 Another possibility is that strong pro-public preferences are more likely in countries like Denmark, which feature both a large public sector and relatively high trust in government. The range of possible explanations call for further research to better understand how governance preferences generate symmetrical or asymmetrical responses, and under what circumstances. Another way of considering our hypothesis is to see what happens to those who are less motivated to reach certain conclusions, that is, those who declare themselves to be agnostic about public or private provision. To behave consistently with our hypothesis, this group should not alter their evaluative criteria even when provided with information about the schools’ sector affiliations. The results from Figure 4B fit with these expectations. There is not a clear directional or large effect of providing sector information to those who say they are agnostic about sector. This finding not only provides support for our hypothesis; it also suggests that those without strong prior governance preferences are more likely to consistently apply goal preferences to evaluate public sector performance. Discussion This article offers a number of theoretical contributions. First, we map insights from motivated reasoning theory to the contours of public management, examining how they play out in conditions of goal conflict. Second, we fashion and test useful concepts for our theoretical goals. In adapting psychological insights about goal reprioritization to the context of organizational goals, we offer a new avenue for studying how political and organizational actors are willing to reweight the importance they assign to policy goals relative to governance preferences. Whereas researchers of voting behavior observe that partisan alignments trump policy issues (Groenendyk 2013), we offer evidence that governance preferences trump goal preferences in the context of public management and decision-making. The findings are not just of theoretical interest—they have substantive importance. Our results undercut the idea that goal preferences are stable or dominant forces in shaping assessments of organziational performance; instead they are subservient to deeper ideological preferences about governance. In our study, political principals are motivated more by normative beliefs about the appropriate nature of the public sector than by a consistent set of goal preferences when using performance data to evaluate providers of public services. Goal preferences, applied consistently, provide goal clarity for public organizations, and some form of performance-based accountability to the public. Governance preferences do not—they devalue both the notion of consistency in policy goals, and the notion of performance more generally, in favor of one vision of governance over another. In the most basic of terms, if managers understand that governance preferences matter more than performance or goal preferences, they have little reason to expect that their efforts to improve performance will count for much. The results raise an intruiging question for practice: is it possible—or even desirable—to enforce consistency in goal preferences? It might be possible to design mechanisms to limit goal repriortization. For example, political principals might be forced to publicly rank goal preferences in a predecisional phase, and then be reminded of those commitments when data becomes available. Such a mechanism might curb the willingness of policymakers to selectively reweight goal preferences to fit with governance preferences, enforcing greater consistency in commitments to goals. But some might see such commitment mechanisms as overly technocratic, robbing policymakers of discretion to incorporate a consideration of governance preferences that they were also elected to pursue. The theory and findings both offer insight and complicate the study of performance information use. They add to a growing body of work that suggests that ideology matters. But while that prior work has offered empirical evidence for the claim that motivated reasoning alters how people interpret performance data, we show something different: that performance data will be used to justify switching between policy goals to justify governance preferences. Goal conflict is therefore convenient for motivated reasoners with strong governance preferences, in that it gives them an opportunity to make a data-based jutification for a predetermined ideological end. The claim that decision-makers will selectively use performance data to justify their arguments is not entirely new (Moynihan 2008), but we offer a specific theoretical logic for how such a process happens, as well as a compelling test for this claim. Our approach has some limitations. Because we rely on a between-subjects design we cannot definitively prove that the difference between groups is because of the goal reprioritization mechanism we specify. A within-subjects design could test the causal mechanism more conclusively. As with every experiment, we are limited in external generalizability because of the need to simplify conditions to identify causal processes. More research is needed in other settings and other policy areas where governance preferences might work differently, and with other groups of decision-makers. For example, we cannot assume that our study of elected officials extends to bureaucrats, who may approach decisions quite differently (Meier and O’Toole 2006). However, our survey experiment provides some value to practice in two ways. First, it engages actual elected officials with plausible performance data and goal preferences. Within the necessarily artificial nature of experiments, the design is not unrealistic. Second, the design is valuable precisely because it deliberately incorporates more of the complexity of actual decision scenarios than has been the case in other studies of how performance data is used. The theoretical conditions we sketch out—multiple measures tied to competing goal preferences, and decision-makers with strong governance preferences—are both more realistic and more complex than experiments that examine how individuals process only a single piece of performance data around a single goal. While in actual practice there will be even more competing goal preferences and governance preferences, our theory is premised on acknowledging this complexity in a way that is absent from much of prior research. In practice, we might assume that having a greater array of performance data, and more competing goal preferences would make the process of goal reprioritization even more likely to occur. The results add nuance to the debate about the relative virtues of simple versus comprehensive performance data systems. One implication of our result is that the provision of more information will not necessarily improve goal-based learning. We noted that the level of ambiguity in performance information can be expected to increase as more performance indicators are included in a performance measurement system. As more data creates goal conflict, it also opens up the opportunity for a biased use of data. Goal reprioritization becomes one additional reason why more data does not actually generate agreement, instead reinforcing disagreement in polarized settings. However, our results also show that those without strong governance preferences do rely on goal preferences, suggesting that performance data may be used with greater consistency in institutions—such as bureaucracies—where governance preferences can be muted (Moynihan 2008). While our study is largely an exercise in adapting relevant theory from political psychology to the context of public administration, it also offers insights of broader interest to psychology and motivated reasoning. First, the results help to expand the scope conditions of issue reprioritization, moving beyond its original setting of voters using ideology to reweight the importance of competing policy issues, to incorporate a new setting of elected officials engaged in performance evaluation within a single policy area. Second, the results shed light on whether expertise and commitment can overcome motivated reasoning. Much of political psychology and psychology relies on surveys of students or members of the public on issues where they may have marginal knowledge or commitment. By contrast, we examine a context with subjects with deeper policy expertise and more crystalized goal preferences. We show that even under these relatively demanding conditions, processes of reprioritization still hold. Conclusion Processes of what we call goal reprioritization are salient in psychology but only hinted at in public administration. In this article, we show that political decision-makers use goal reprioritization as a strategy for motivated reasoning when interpreting ambiguous information about the performance of organizations. Elected officials change the evaluative criteria they use to assess performance depending on what the data actually says about the outcome. In more concrete terms, political decision-makers seem to “shop around” for the dimensions of performance that can best be used to justify their prior governance preferences regarding public versus private service provision. In particular, those who support public provision of services use the dimensions of performance that portray public schools in the most favorable light. Given the lack of prior attention to processes of goal reprioritization, there is much additional work that could be pursued, some of which we have already discussed. Perhaps most obviously, we do not observe if these processes emerge through deliberate strategy or unconcious bias. The answer to this question might have implications in terms of designing interventions to debias decision-makers’ reasoning. Additional work could also usefully examine if the insights we develop about elected officials applying goal reprioritization under conditions of goal conflict extend to other groups or conditions. The original theory of issue reprioritization was based on citizens, and so it should be expected to be salient for citizens assessing outcomes. On the other hand, we have no evidence on whether public managers engage in goal reprioritization. Future research should test if similar patterns of goal repriortization hold among career officials. The concept of governance preferences is broad, but has particular salience to public settings, qualities that make it an example of a concept ripe for deeper theoretical development to bridge disciplinary research and public administration, while also making it useful for empirical testing of behavioral implications (James, Jilke, and Van Ryzin 2017). There is also significant room to consider how other forms of governance preferences—such as attitudes toward public employment, politicization, transparency, the size of government, or other public service values—might drive goal reprioritization or other preferences or decisions. References Baekgaard, Martin, Julian Christensen, Casper M. Dahlmann, Asbjørn Mathiasen, and Niels B. G. Petersen. 2017. The role of evidence in politics: Motivated reasoning and persuasion among political decision-makers. British Journal of Political Science . doi: 10.1017/S0007123417000084 Baekgaard, Martin, and Søren Serritzlew. 2016. Interpreting performance information: Motivated reasoning or unbiased comprehension. Public Administration Review  76: 73– 82. Google Scholar CrossRef Search ADS   Behn, Robert D. 2001. Rethinking democratic accountability . Washington, DC: Brookings Institution Press. Bischoff, Ivo, and Frédéric Blaeschke. 2016. Performance budgeting: Incentives and social waste from window dressing. Journal of Public Administration Research and Theory  26: 344– 58. Google Scholar CrossRef Search ADS   Bisgaard, Martin. 2015. Bias will find a way: Economic perceptions, attributions of blame, and partisan-motivated reasoning during crisis. The Journal of Politics  77: 849– 60. Google Scholar CrossRef Search ADS   Bohte, John, and Kenneth J. Meier. 2000. Goal displacement : Assessing the motivation for organizational cheating. Public Administration Review  60: 173– 82. Google Scholar CrossRef Search ADS   Boyne, George A. 2002. Concepts and indicators of local authority performance: An evaluation of the statutory frameworks in England and Wales. Public Money & Management  22: 17– 24. Google Scholar CrossRef Search ADS   Christensen, Julian. 2017. Justification requirements as a debiasing measure against motivated interpretations of factual information? In 2017 Annual Meeting of American Political Science Association (APSA) . San Fransisco. Chun, Young Han, and Hal G. Rainey. 2005. Goal ambiguity and organizational performance in U.S. Federal agencies. Journal of Public Administration Research and Theory  15: 529– 57. Google Scholar CrossRef Search ADS   Festinger, Leon. 1957. A theory of cognitive dissonance . Stanford Univ. Press. Goodsell, Charles T. 2004. The Case for Bureaucracy: A Public Administration Polemic . 4th ed. Washington, DC: CQ Press. Government of Denmark. 2013. Agreement between the Danish Government (the Social Democrats, the Social-Liberal Party and the Socialist People’s Party), the Liberal Party of Denmark and the Danish People’s Party on an improvement of standards in the Danish public school (primary and lower secondary education). http://eng.uvm.dk/-/media/filer/uvm/english/pdf/131007-folkeskolereformaftale-eng-red.pdf?la=en (accessed January 19, 2018). Groenendyk, Eric W. 2013. Competing motives in the partisan mind: How loyalty and responsiveness shape party identification and democracy . New York: Oxford Univ. Press. Google Scholar CrossRef Search ADS   Heinrich, Carolyn J. 2007. Evidence-based policy and performance management. The American Review of Public Administration  37: 255– 77. Google Scholar CrossRef Search ADS   Holmstrom, Bengt, and Paul Milgrom. 1991. Multitask principal-agent analyses: Incentive contracts, asset ownership, and job design. Journal of Law, Economics, & Organization  7 (Special Issue: [Papers from the Conference on the New Science of Organization, January 1991]): 24– 52. Google Scholar CrossRef Search ADS   Holzer, Marc, and Kaifeng Yang. 2004. Performance measurement and improvement: An assessment of the state of the art. International Review of Administrative Sciences  70: 15– 31. Google Scholar CrossRef Search ADS   Hvidman, Ulrik, and Simon Calmar Andersen. 2016. Perceptions of public and private performance: Evidence from a survey experiment. Public Administration Review  76: 111– 20. Google Scholar CrossRef Search ADS   Jacobsen, Rebecca, Jeffrey W. Snyder, and Andrew Saultz. 2015. Understanding satisfaction with schools: The role of expectations. Journal of Public Administration Research and Theory  25: 831– 48. Google Scholar CrossRef Search ADS   James, Oliver, and Gregg G Van Ryzin. 2017. Motivated reasoning about public performance: An experimental study of how citizens judge obamacare. Journal of Public Administration Research and Theory  27: 197– 209. Google Scholar CrossRef Search ADS   James, Oliver, Sebastian R. Jilke, and Gregg G. Van Ryzin (eds). 2017. Experiments in public management research: Challenges and contributions . Cambridge: Cambridge University Press. Google Scholar CrossRef Search ADS   Jerit, Jennifer, and Jason Barabas. 2012. Partisan perceptual bias and the information environment. The Journal of Politics  74: 672– 84. Google Scholar CrossRef Search ADS   Julnes, Patria de Lancer. 2011. Performance measurement beyond instrumental use. In Performance information in the public sector: How it is used , ed. Wouter van Dooren and Steven van de Walle, 62– 75. New York: Palgrave Macmillan. Kahan, Dan M. 2016. The politically motivated reasoning paradigm, Part 2: Unanswered questions. In Emerging trends in the social and behavioral sciences: An interdisciplinary, searchable, and linkable resource, 1–15. John Wiley & Sons. http://onlinelibrary.wiley.com/doi/10.1002/9781118900772.etrds0418/abstract (accessed January 19, 2018). Kelman, Steven, and John N. Friedman. 2009. Performance improvement and performance dysfunction: An empirical examination of distortionary impacts of the emergency room wait-time target in the English National Health Service. Journal of Public Administration Research and Theory  19: 917– 46. Google Scholar CrossRef Search ADS   Klar, Samara, and Yanna Krupnikov. 2016. Independent politics: How American disdain for parties leads to political inaction . New York: Cambridge Univ. Press. Google Scholar CrossRef Search ADS   Kroll, Alexander. 2015. Drivers of performance information use: Systematic literature review and directions for future research. Public Performance & Management Review  38: 459– 86. Google Scholar CrossRef Search ADS   Kunda, Ziva. 1987. Motivated inference: Self-serving generation and evaluation of causal theories. Journal of Personality and Social Psychology  53: 636– 47. Google Scholar CrossRef Search ADS   Kunda, Ziva. 1990. The case for motivated reasoning. Psychological Bulletin  108: 480– 98. Google Scholar CrossRef Search ADS PubMed  Lavine, Howard G., Christopher D. Johnston, and Marco R. Steenbergen. 2012. The ambivalent partisan: How critical loyalty promotes democracy . New York: Oxford Univ. Press. Google Scholar CrossRef Search ADS   Lodge, Milton, and Charles S. Taber. 2000. Three steps toward a theory of motivated political reasoning. In Elements of reason , edited by Arthur Lupia, Mathew D McCubbins, and Samuel L Popkin, 183– 213. Cambridge: Cambridge University Press. Google Scholar CrossRef Search ADS   Marvel, John D. 2016. Unconscious bias in citizens’ evaluations of public sector performance. Journal of Public Administration Research and Theory  26: 143– 58. Meier, Kenneth J., and Laurence J. O’Toole. 2006. Bureaucracy in a Democratic State: A Governance Perspective . Baltimore, MD: Johns Hopkins University Press. Meier, Kenneth J., Robert J. Wrinkle, and J. L. Polinard. 1999. Representative bureaucracy and distributional equity: Addressing the hard question. The Journal of Politics  61: 1025– 39. Google Scholar CrossRef Search ADS   Moynihan, Donald P. 2008. The dynamics of performance management: Constructing information and reform . Washington, DC: Georgetown University Press. Moynihan, Donald P. 2015. Uncovering the circumstances of performance information use findings from an experiment. Public Performance & Management Review  39: 33– 57. Google Scholar CrossRef Search ADS   Moynihan, Donald P., and Ivor Beazley. 2016. Toward next-generation performance budgeting. Lessons from the experiences of seven reforming countries . Washington, DC: World Bank. Google Scholar CrossRef Search ADS   Moynihan, Donald P., Sergio Fernandez, Soonhee Kim, Kelly M. LeRoux, Suzanne J. Piotrowski, Bradley E. Wright, and Kaifeng Yang. 2011. Performance regimes amidst governance complexity. Journal of Public Administration Research and Theory  21 ( Supp 1): 141– 55. Google Scholar CrossRef Search ADS   Nielsen, Poul A., and Donald P. Moynihan. 2017. How do politicians attribute bureaucratic responsibility for performance? Negativity bias and interest group advocacy. Journal of Public Administration Research and Theory  27: 269– 83. Rainey, Hal G., and Chan Su Jung. 2015. A conceptual framework for analysis of goal ambiguity in public organizations. Journal of Public Administration Research and Theory  25: 71– 99. Google Scholar CrossRef Search ADS   Serritzlew, Søren. 2006. Linking budgets to activity: A test of the effect of output-purchase budgeting. Public Budgeting & Finance  26: 101– 20. Google Scholar CrossRef Search ADS   Simon, Linda, Jeff Greenberg, and Jack Brehm. 1995. Trivialization: The forgotten mode of dissonance reduction. Journal of Personality and Social Psychology  68: 247– 60. Google Scholar CrossRef Search ADS   Taber, Charles S., and Milton Lodge. 2016. The illusion of choice in democratic politics: The unconscious impact of motivated political reasoning. Political Psychology  37: 61– 85. Google Scholar CrossRef Search ADS   Taber, Charles S., and Milton Lodge. 2006. Motivated skepticism in the evaluation of political beliefs. American Journal of Political Science  50: 755– 69. Google Scholar CrossRef Search ADS   Tilley, James, and Sara B Hobolt. 2011. Is the government to blame? An experimental test of how partisanship shapes perceptions of performance and responsibility. Journal of Politics  73: 316– 30. Google Scholar CrossRef Search ADS   Van de Walle, Steven, and Alasdair Roberts. 2011. Publishing performance information: An illusion of control? In Performance information in the public sector: How it is used , edited by Wouter Van Dooren and Steven Van de Walle, 215– 30. New York: Palgrave Macmillan. Google Scholar CrossRef Search ADS   Van Dooren, Wouter. 2011. Nothing new under the sun? Change and continuity in the twentieth-century performance movements. In Performance information in the public sector: How it is used , ed. Wouter Van Dooren and Steven Van de Walle, 15– 27. New York: Palgrave Macmillan. Google Scholar CrossRef Search ADS   Van Dooren, Wouter Van, Geert Bouckaert, and John Halligan. 2010. Performance management in the public sector . Abingdon, Oxton: Routledge. Wildavsky, Aaron. 1979. Speaking truth to power: The art and craft of policy analysis . Boston: Little, Brown and Company. Google Scholar CrossRef Search ADS   Wright, Bradley E. 2004. The role of work context in work motivation: A public sector application of goal and social cognitive theories. Journal of Public Administration Research and Theory  14: 59– 78. Google Scholar CrossRef Search ADS   Footnotes 1 Jacobsen, Snyder, and Saultz (2015) use the term goal expectations for what we describe as goal preferences here. 2 Our model does not assume that tradeoffs and therefore goal conflict are inevitable: there may also be cases where performance on one goal is positively correlated with performance on another (Meier, Wrinkle, and Polinard 1999). Instead, we merely claim that such tradeoffs, and hence the condition of goal conflict, are relatively common. 3 In addition to our experiment, respondents participated in other survey experiments that were designed to test other but related research questions. However, the survey experiment reported in this article was placed before the remaining experiments in the survey, and therefore the participation in additional experiments did not have the potential to affect our results. Results from the additional experiments are reported in Baekgard et al. (2017) and Christensen (2017), which are mentioned in the text. 4 The wording of the items was: “To what extent do you agree or disagree with the following statements? (1) Many public activities could be produced both better and cheaper by private providers. (2) We should to a larger degree outsource public services (such as child care, elderly care, and hospital treatments). (3) The public sector is best at providing public services.” Possible answers were “completely agree,” “partly agree,” “neither agree nor disagree,” “partly disagree,” “completely disagree,” and “don’t know.” A factor analysis revealed factor scores above 0.8 for all items and an index was therefore constructed ranging from 0 to 1, where higher values correspond to a stronger preference for the public sector. Cronbach’s alpha was 0.92. Mean = 0.62 and standard deviation = 0.33. 5 In Denmark, private schools are financed through a combination of user charges and government subsidies whereas public schools are financed through municipal taxes (Serritzlew 2006: 108). Furthermore, private schools differ from public schools by being operated by school boards with greater autonomy regarding approaches to learning and pedagogy. The autonomy of public school boards is more limited and depends on delegation from the city council (ibid.). 6 There was enough of an ongoing concern about such tensions that in 2013, shortly before our data collection, a majority of the Danish national parliament mandated, as a part of a major reform of the Danish public school system, that each school collect student well-being data in addition to test scores (Government of Denmark 2013). Thus, in the policy setting we study, the tensions between the two dimensions of performance are both plausible and salient. 7 The use of cognitively undemanding data is different from analyses such as Baekgaard et al. (2017) and Baekgaard and Serritzlew (2016) who ask respondents to compute performance using more complex data. For our purposes, making the performance data as clear as possible is beneficial because it excludes cognitive complexity as a reason why policymakers struggle to evaluate the performance of schools. 8 Data from the Danish Ministry for Economic Affairs and the Interior (www.noegletal.dk). Appendix Table A1. Balance Checks of Randomization: Each Group Against Rest of Sample   Group 1  Group 2  Group 3  Group 4  Women (pct)  32.8  32.6  28.6  27.1  Age (years, mean)  51.9  53.1  52.7  50.9  Pro-public preference (mean)  0.61  0.67*  0.62  0.60.  Higher education (pct)  71.9  70.2  67.2  69.3  Parent to child in school (pct)  25.7  32.9  33.2  28.3  Years of membership in current party (mean)  24.3  26.9  24.8  26.7  Years of membership in city council (mean)  20.8  21.5  22.3  21.3  Member of Finance Committee (pct)  35.2  29.1  33.2  32.0  Member of left-wing partya (pct)  45.6  55.6**  45.0  46.1  Proportion of municipality’s students who are in private schoolsb (mean pct)  16.1  16.6  15.8  16.9  Tax base per citizenc (mean)  172.5  171.9  171.3  176.3*  Size of municipalityd (mean)  67.0  73.9  62.8  61.1    Group 1  Group 2  Group 3  Group 4  Women (pct)  32.8  32.6  28.6  27.1  Age (years, mean)  51.9  53.1  52.7  50.9  Pro-public preference (mean)  0.61  0.67*  0.62  0.60.  Higher education (pct)  71.9  70.2  67.2  69.3  Parent to child in school (pct)  25.7  32.9  33.2  28.3  Years of membership in current party (mean)  24.3  26.9  24.8  26.7  Years of membership in city council (mean)  20.8  21.5  22.3  21.3  Member of Finance Committee (pct)  35.2  29.1  33.2  32.0  Member of left-wing partya (pct)  45.6  55.6**  45.0  46.1  Proportion of municipality’s students who are in private schoolsb (mean pct)  16.1  16.6  15.8  16.9  Tax base per citizenc (mean)  172.5  171.9  171.3  176.3*  Size of municipalityd (mean)  67.0  73.9  62.8  61.1  Note: Significant differences are identified by comparing each group with the rest of the sample using two-tailed t-tests and tests of proportions. aMembers of the Social Democrats, the Socialist People’s Party, and the Red-Green Alliance have been coded as members of a left-wing party. Members of the Danish Social Liberal Party, Liberal Alliance, the Christian Democrats, the Danish People’s Party, and Venstre, Liberal Party of Denmark have been coded as not being members of a left-wing party. Members of local parties and city councilors who have not indicated a party affiliation have been coded as missing. bProportion of all primary school students in a respondent’s municipality who was in a private school (2014 numbers drawn from www.noegletal.dk which is run by the Danish Ministry for Economic Affairs and the Interior). 3Tax base per citizen measures the 2014 tax base (in thousands of Danish Kroner) per citizen in a respondent’s municipality (numbers drawn from www.noegletal.dk). 4Size of municipality measures the number of citizens (in thousands) in a respondent’s municipality as of January 1, 2014 (numbers drawn from www.noegletal.dk). ***p < .001; **p < .01; *p < .05. View Large Table A2. Influence of Governance Preferences on Assessments of Performance, Using Party ID Instead of Stated Preferences   Model 1  Model 2  Left winga  1.69*** (0.32)  0.89** (0.27)  Sector revealed (treatment)  0.03 (0.25)  −0.62* (0.26)  Left wing × Sector revealed  −3.41*** (0.45)  3.03*** (0.60)  Intercept  −3.67e−16 (0.18)  −0.02 (0.19)  LR χ2  110.51***  126.90***  N  468  491  Data  Group 1 + 3  Group 2 + 4  Best well-being  Private / B  Public / A    Model 1  Model 2  Left winga  1.69*** (0.32)  0.89** (0.27)  Sector revealed (treatment)  0.03 (0.25)  −0.62* (0.26)  Left wing × Sector revealed  −3.41*** (0.45)  3.03*** (0.60)  Intercept  −3.67e−16 (0.18)  −0.02 (0.19)  LR χ2  110.51***  126.90***  N  468  491  Data  Group 1 + 3  Group 2 + 4  Best well-being  Private / B  Public / A  Note: Two-sided significance tests. Entries are logistic regression coefficients. Standard errors in parentheses. Dependent variable is respondents’ tendency to prioritize student wellbeing over grants when ranking school performance. aMembers of the Social Democrats, the Socialist People’s Party, and the Red-Green Alliance have been coded as members of a left-wing party (left wing = 1). Members of the Danish Social Liberal Party, Liberal Alliance, the Christian Democrats, the Danish People’s Party, and Venstre, Liberal Party of Denmark have been coded as not being member of a left-wing party (left wing = 0). Members of local parties and city councilors who have not indicated a party affiliation have been coded as missing. ***p < .001; **p < .01; *p < .05; †p < .1. View Large © The Author(s) 2018. Published by Oxford University Press on behalf of the Public Management Research Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.

Journal

Journal of Public Administration Research and TheoryOxford University Press

Published: Apr 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Unlimited reading

Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.

Stay up to date

Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.

Organize your research

It’s easy to organize your research with our built-in tools.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

Monthly Plan

  • Read unlimited articles
  • Personalized recommendations
  • No expiration
  • Print 20 pages per month
  • 20% off on PDF purchases
  • Organize your research
  • Get updates on your journals and topic searches

$49/month

Start Free Trial

14-day Free Trial

Best Deal — 39% off

Annual Plan

  • All the features of the Professional Plan, but for 39% off!
  • Billed annually
  • No expiration
  • For the normal price of 10 articles elsewhere, you get one full year of unlimited access to articles.

$588

$360/year

billed annually
Start Free Trial

14-day Free Trial