DETERMINANTS AND INFLUENCE OF EXPECTATIONS OF GOVERNMENT PERFORMANCE By Aliyah J. McIlwain A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Political Science – Doctor of Philosophy 2023 ABSTRACT How are expectations of government performance developed and why do they matter? The literature on government expectations typically focuses on expectations of candidate behavior even though scholarship on policy feedback suggests that policy influences how and which constituents participate in the political system. I argue that these bodies of scholarship have failed to acknowledge the relevancy of citizens expectations of government performance—in this case, how well government provides the services that citizens expect from it. In a series of three essays, I explore how experience and identity influence expectations of policy outcomes, and finally how expectations impact political participation. To Laura, Arnold, and Cathy iii ACKNOWLEDGMENTS The completion of this journey simply would not be possible without a whole entourage of cheerleaders. I am grateful to my family and closest friends who may not have understood what I was going through, but because you love and believe in me you listen and support me. And, to those who know this journey well, your guidance has been invaluable. I may not have ever considered graduate school if not for Steven Thomas and the Summer Research Opportunities Program (SROP). My work has benefited the financial, literary, and social support of the Alliance for the Graduate Education and the Professoriate (AGEP). I hold SROP and AGEP near and dear. I am thankful to my committee, Sarah Reckhow, Valentina Bali, Jennifer Wolak, and Josh Sapotichne. You all’s enthusiasm for my ideas was instrumental to my confidence as a scholar. Thank you for every meeting, piece of feedback, pontification, and opportunity for space on a survey. To my advisor, Sarah, I could not have asked for a more supportive or kinder mentor through this journey. I will miss our weekly check-ins. I also want to extend a thank you to Matt Grossman for letting me try out survey questions on the State of the State Survey. And, to Nazita Lajevardi and Eric Juenke, thank you for creating safe space for the students of color in our department—the work that you do does not go unnoticed. The knowledge that I gained working at the Education Policy Innovation Collaborative (EPIC) has been instrumental to my growth as a scholar. Erica Harbatkin and Katharine Strunk, you two have taught me to communicate importance science to broad audiences and to do so with rigor and high standards. I will take these lessons wherever I go. Katharine, thank you for taking a chance on the computer scientist who didn’t know anything about ed policy yet. To my fellow grad students, especially Kesicia Dickinson, Shayla Olson, Erika Vallejo, iv Ha -Eun Choi, and Gerson Guevara, you all have been my tribe over the last few years. Our various weekly writing groups and social outings have been a place of support, productivity, and great scholarship. I hope that we continue to support each other, and I wish you all great success. v TABLE OF CONTENTS CHAPTER 1: INTRODUCTION…………………………………………………………………1 CHAPTER 2: RECONSTRUCTIVE REPARATIONS? A SURVEY OF GOVERNMENT PERFORMANCE AND PERCEPTIONS IN SOUTHEASTERN MICHIGAN………….……...4 CHAPTER 3: WHO’S IN CHARGE? HOW REPRESENTATION IN GOVERNMENT INFLUENCES EXPECTATION OF LOCAL SERVICES…..…………………………………21 CHAPTER 4: PARTICIPATION AS A FUNCTION OF EXPECTATIONS AND SATISFACTION ………………………………………………..………………………………35 CHAPTER 5: CONCLUSION…………………………………………………………………..49 BIBLIOGRAPHY………………………………………………………………………………..52 APPENDIX A: SUPPLEMENTAL MATERIAL FOR CHAPTER 2…………………………..58 APPENDIX B: SUPPLEMENTAL MATERIAL FOR CHAPTER 3…………………………..61 APPENDIX C: SUPPLEMENTAL MATERIAL FOR CHAPTER 4…………………………..67 vi CHAPTER 1: INTRODUCTION This dissertation is a study of how the experience and identity of American citizens influence their expectations of government performance and what expectations could mean for the health of our democratic system. Typical studies of government expectations in the political science discipline seem far removed from the actual product that government produces—policies—yet the policy feedback scholarship acknowledges that policies influence the ways that citizens interact with the political system. Both bodies of scholarship are missing considerations of expectations in important ways. Studies of expectations often focus on how candidates and elected officials should behave in office as opposed to whether we expect them to produce policy that aligns with our needs (Davidson 1970; Griffin and Flavin 2007; Kimball and Patterson 1997; Seyd 2015; Waterman et al. 1999). While policy feedback studies often begin to explore this relationship in the aftermath of a policy experience, forgoing consideration for a citizen’s expectations for a policy experience (Beland 2010; Campbell 2002, 2003; Lowi 1972; Mettler 2002; Mettler and Soss 2004; Pierson 1993; Soss 1999). In this project, I explore the relationship between expectations of government, satisfaction with policy experiences, and political participation. This dissertation will focus on two aspects of this relationship: the social and political factors that shape expectations and the influence of expectations on satisfaction. I aim to understand how expectations of policy outcomes influence satisfaction and how these expectations might develop. I rely on a series of survey experiments to explore these aspects through the following research questions: • Do expectations of government differ by historical context? • Do expectations of government differ across race and class? • How does the descriptive identity of political officials influence expectations? 1 • What is the relationship between predictive expectations and experienced satisfaction? And how do changes between pre-existing expectations and levels of satisfaction after a policy experience influence civic and political participation? By answering these questions, I hope to highlight the important but overlooked function of expectations in our existing models of satisfaction and participation. In the following chapters, I explore some of the ways citizens develop different expectations of their local policy experience and what these expectations mean for political participation. I rely on three original survey experiments. The first part of the dissertation uses an original experiment conducted in the 86th Michigan State University State of the State Survey. I leverage the historical and racial context in Detroit, Michigan to understand the expectations of a real policy initiative in Michigan. I find the Black Wayne County residents are distinct in their perceptions of policy from non-Wayne County residents and non-Black Wayne County residents, suggesting that personally salient historical knowledge has influence over citizens expectations of future policy. The second part of my dissertation builds on previous studies of candidate preferences and perceptions to understand how identity influences perceptions of candidate quality and how those perceptions might be related to expectations of local services. Conducting a candidate conjoint experiment in the 2022 Congressional Election Study, I find that constituents expect to be more satisfied with local public services if a candidate is also perceived as understanding and having the solution for people like themselves. Perceptions of understanding and ability to problem solve are influenced by both candidate and constituent identity. The findings also suggest that traditional operationalizations of qualification for office might not be sufficient. Finally, I examine the impact of expected satisfaction and experienced satisfaction on 2 political participation, using a survey experiment that presents with respondents a hypothetical experience with either a positive or negative outcome involving local government services. I find that expected satisfaction does influence experienced satisfaction with a policy. Changes between expected and experienced satisfaction do influence the ways people choose to participate. 3 CHAPTER 2: RECONSTRUCTIVE REPARATIONS? A SURVEY OF GOVERNMENT PERFORMANCE AND PERCEPTIONS IN SOUTHEASTERN MICHIGAN Distrust of American Government is strong and persistent. Historically, many marginalized communities have been harmed by or had disappointing experiences with government. We can look to several large-scale events like The Indian Removal Act of 1830, Japanese Internment during World War II, Jim Crow Laws, The Syphilis Study at Tuskegee, Hurricane Katrina, Hurricane Maria, and The Flint Water Crisis as examples of times citizens have been harmed by American government. Surely these events along with smaller everyday interactions have had lasting influence on perceptions of government. For example, if we consider Coronavirus Virus (COVID- 19) vaccine hesitancy among Black Americans, many factors could influence public trust in medical institutions and procedures; however, for Black Americans, the Syphilis Study at Tuskegee is a substantially painful moment in history that influenced many Black Americans’ choice on whether to be vaccinated against COVID-19. Policy feedback scholars have explored how policies influence citizen engagement with the political system. This work focuses on the impacts of policies on citizens, but policy feedback research often fails to consider the influence of pre-existing perceptions of government. Overlooking these expectations falsely suggests that government gets a new chance to prove itself for every policy experience. It is imperative to understand how citizens make their evaluations of government if perception of government is ever to improve. The historical, racial, and policy context of unequal treatment by government present an opportunity to assess the relationship between historical circumstance, personal identity, and policy expectations. Do expectations of government differ by historical context? Do expectations of government differ by race? I investigate the relationship between historical circumstance, personal identity, and policy 4 attitudes relying on the rich racial and historical context in Detroit, Michigan. The I-375 replacement project in Detroit presents an opportunity to understand the expectations of a policy before it happens within a racial and historical context. The project is very high profile and widely relevant to Metro-Detroiters1, offering a real-world case to analyze the expectations of residents with a wide array of racial, ethnic, and socio-economic backgrounds. I assess how views change or don’t change in response to informational cues regarding the historical context surrounding the proposed I-375 replacement project. I use a survey experiment to show that views are racially and geographically distinct and that citizens use their experience to develop attitudes about policies before they happen. I find that Black Wayne County residents (predominately Detroit residents) hold distinct views about the I-375 renovation project from both Black residents across the rest of the state and non-Black residents of Wayne County. This study aides our understanding of different how citizens might express contending views regarding the benefits of a policy. Historical Context: The Destruction (and revitalization) of Black Bottom and Paradise Valley Black Bottom and the adjacent Paradise Valley neighborhoods are a culturally important historically Black section of Detroit, MI. The Black Bottom neighborhood was bounded by Brush Street, Gratiot Avenue, and the Grand Trunk railroad tracks. The area was named from the dark, fertile, topsoil that was a part of the river than was buried below it. Paradise Valley, sitting just north of Black Bottom bound by Gratiot Avenue, John R Street, Mack Avenue, and Hasting Street, served as entertainment corridor to Black Bottom. Paradise Valley was home to prominent music venues, bars, a bowling alley, and hotels. In the late 1950s and early 60s the thriving Black neighborhoods were demolished to build Interstate 375 through eminent domain 1 In this context, Metro-Detroiters refers to the tri-county surrounding Detroit-Michigan: Wayne County, Oakland County, and Macomb County. See Appendix Figure A.1 5 policy on what was Hasting Street. This area was chosen because it would not affect many white businesses. The population in this area was displaced on short notice with few resources. The loss of these neighborhoods is considered a massive historical and cultural loss for Detroiters (Grant 2022; Lawerence 2022). FIGURE 2.1 MAP OF PARADISE VALLEY AND BLACK BOTTOM Note: Map of Paradise Valley and Black Bottom locations within Modern-Day Downtown Detroit. Source: DetroitIsIt Seventy years later there are new efforts to replace Interstate 375 (I-375) with a street- level boulevard to create development opportunities, particularly for Black businesses, in the area that was formerly Black Bottom and Paradise Valley (Lawerence 2022). The project has been touted as a priority by Michigan Governor Gretchen Whitmer, Detroit Mayor Mike Duggan, and United States Transportation Secretary Pete Buttigieg, and others. The project was approved by the Michigan Department of Transportation (MDOT) and is set to receive $105 million from a competitive federal grant. During a September 2022 press release, Secretary Buttigieg highlighted the revitalization goal of the project: This stretch of I-375 cuts like a gash through the neighborhood, one of the many examples I have seen in communities across the country where a piece of infrastructure 6 has become a barrier. With these funds, we’re now partnering with the state and the community to transform it into a road that will connect rather than divide. (Whitmer Sec Buttigieg and Local Leaders Celebrate Historic I-375 Project, 2022) While Governor Whitmer spoke directly about the historical context: In the 1950s, I-375 paved through two prosperous Black communities and displaced over 130,000 Michiganders and hundreds of businesses. While we cannot change the past, we can work together to build a more just future, and that’s exactly what today’s grant empowers us to do. (Whitmer Sec Buttigieg and Local Leaders Celebrate Historic I-375 Project, 2022) FIGURE 2.2 MICHIGAN DEPARTMENT OF TRANSPORTATION INTERSTATE 375 PROJECT AREA RENDERING The city began to see declines in population following the end of World War II. Most of the residents who exited the city were white due to discriminatory housing policy, looming desegregation, and changing industry. This created an even starker separation of blacks from whites and higher income from lower income in the region. Naturally, as people spread beyond city boundaries, they needed ways to come back in. A bustling regional transit system was developing until the 1980s when commuter rail service from Pontiac and Ann Arbor was 7 terminated. A combination of racial and political tension along with investment from the automotive industry have continuously thwarted efforts for regional public transportation. Instead, Metro-Detroit is left with highways that rip through the majority black city and communities that remain. While the highways in Southeastern Michigan make accessing the city easy for those living in metro-Detroit suburbs they run right through neighborhoods within the city. This has created a city that is not walkable and difficult to navigate for those who cannot afford to have a personal vehicle. Declining city population and dysfunctional regional transit authorities weaken the effectiveness of public transit for those who need it within city lines. For these reasons, metro-Detroiter’s may have different perspectives on the I-375 replacement project depending on race, income, and where they reside within Southeast Michigan. Policy feedback and Participation Policy feedback scholars study the influence of policy on political participation (Beland 2010; Campbell 2002, 2003; Lowi 1972; Mettler 2002; Mettler and Soss 2004; Pierson 1993; Soss 1999). Policies can either deter or foster future participation through implementation and consequences of the policy change (Campbell 2002, 2003; Schneider and Ingram 1993). Pierson (1993) suggests that policies provide resource and/or interpretive effects that impact civic engagement. Resource effects are effects caused by the provision or removal of resources from a policy’s target population. Interpretive effects are messages conveyed through policy implementation that suggest how government might feel about target groups, influencing how the general public feels about target groups, additionally altering how the target group may see themselves and their place in society (Schneider and Ingram 1993; Shklar 1995; Skocpol 1991; Soss 1999; Wilson 1991). Experiencing empowering policies should be associated with greater participation. A 8 program like welfare can provide a recipient with resources, yet it has also been shown to have particularly negative effects on participation (Soss, 1999). Some scholars have suggested that these types of programs can lower personal motivation or that the resources provided through these programs do not create a sense of deservingness among beneficiaries, so they are deterred from participating when they otherwise would have (Edelman 1985; Mead 1997; Michener 2018). Brady, Verba and Schlozman (1995) and Rosenstone and Hansen (1993) suggest that lower income citizens, like those who might use welfare, may not have time, financial resources, or civic skills to participate. These findings focus on the political effects of the resources that distributive programs provide; however, other scholars present works that focus more on how the institutional design of these programs can empower or deter beneficiaries (Lawless and Fox 2001; Mettler and Soss 2004; Michener 2018; Soss 1999). Mettler and Soss (2004) suggest that policy design can forge political cohesion and group divisions, build or undermine civic capacities, and structure political participation. Other scholars suggest that these programs require onerous application processes and surveillance by government agencies, making participants feel undeserving and retract from participation (Michener, 2018). These byproducts focus on the effect of a citizen’s political experience with policy programs on their future political participation. The Influence of Expectations on Political Experience While much of the policy feedback literature focuses on the influence of an individual policy experience, I argue that perceptions about a policy experience is not only related to the policy itself (as discussed above), but also pre-existing expectations regarding a policy. This argument incorporates the expectancy-disconfirmation model of satisfaction from public administration literature. The expectancy-disconfirmation model suggests that satisfaction is a function of 9 expectations and perceived performance (Favero and Kim 2021; Van Ryzin 2004; Hjortskov 2019). Incorporating the expectancy-disconfirmation model into the policy feedback framework then suggests that policy feedback is dependent on expectations. Race, Class, and Representation: How Experiences Could Shape Expectations While public administration scholars have made some discussion of the cognitive development of expectations2 there is a lack of discussion about the ways in which identity and personal experience influence expectations. This lack of discussion makes sense given that public administration scholars tend to focus on the development of satisfaction as a straightforward measure of government performance. However, I argue that citizen satisfaction with government services is often not solely dependent on the quality of services. Many of the same effects of policy discussed above, like interpretive effects and targeted population construction, should be expected to influence expectations of policy experiences. As discussed above, American government often systematically imposes policies that impact specific groups of people. As such it is important to consider all the ways citizens develop expectations. Historically, policy experiences have influenced specific targeted groups. This paper focuses on Black Detroiters as a group that had a negative social construction as a targeted group during the construction of Interstate-375. In this paper I explore the influence of past policy experience for a negatively targeted population on expectations about future policies and services. While this paper focuses on the experiences of Black Detroiters, the main argument could apply in other localities or for other targeted cultural groups, like other races/ethnicities, women, students, or those that identify as LGBT. 2 Favero and Kim (2021) find that expectations can be influenced by the information provided to respondents and Hjortskov (2019) finds that expectations are influenced by past expectations, satisfaction, and performance. 10 The influence of previous policy experience (for negatively targeted groups) on perceptions Given the past acts of government that have systematically disadvantaged specific groups, I ask: do perceptions vary systematically across race and historical influence? Answering this question would aide our understanding of how past policies can influence future perceptions. I expect that communities that have experienced harmful policies in the past will have more negative perceptions of policy. I expect that these perceptions will depend on the degree to which historical context is salient to a respondent. In this study, I am using race and locality as a proxy for historical salience. In other words, individuals who share membership with the racial and/or community group(s) impacted by past policies should hold more negative perceptions of future policy. Hence, I expect that: H1: Black Detroiters will have more negative perceptions than everyone else in the state. H2: Black respondents will have more negative perceptions than non-Black respondents. H3: Detroiters will have more negative perceptions than non-Detroiters. To assess perceptions regarding the I-375 project, I ask specifically about support for the project, expected satisfaction, and perceived benefits to Detroiters, Metro-Detroiters, and the families of the displaced. While expected satisfaction and benefits are more directly about expectations, support for the project could be similarly influenced by past policy experience. Data and Experimental Design I conducted an experiment as part of the 86th Michigan State University State of the State Survey (SOSS). The quarterly phone and online survey provides a stratified random sample of 1,000 Michigan residents that is representative in terms of gender, age, race, and education. The survey was fielded December 9-19, 2022. 11 TABLE 2.1 SAMPLE AND POPULATION DEMOGRAPHICS ACS 2021 LOCATION TOTAL POP. PERCENT OF WHOLE BLACK POP. PERCENT BLACK Metro-Detroit 3,940,887 39.16 951,574 24.15 178 PERCENT OF WHOLE 17.8 BLACK POP. PERCENT BLACK 73 41.01 SURVEY TOTAL POP. Wayne County 1,789,781 17.79 676,504 37.80 172 17.2 73 42.44 Detroit 645,658 6.42 503,197 77.94 78 7.8 59 75.64 Statewide 10,062,512 100.00 1,368,177 13.60 1,000 100.00 145 14.50 In the experiment, along with several questions assessing demographic information, political knowledge, and partisanship, I ask citizens (1) “How strongly do you support or oppose this project”; (2) “How satisfied do you expect to be with this project”; and (3-5) “How beneficial do you think the I-375 replacement project is” for residents of Metro-Detroit, the City of Detroit, and the families and descendants of those that were displaced during the development of I-375. The response choices for the support question were “strongly support, somewhat support, neither support or oppose, somewhat oppose, and strongly oppose”; scaled such that a 1 indicates opposition and 3 indicates support. The response choices for the satisfaction question were “extremely dissatisfied, dissatisfied, neither dissatisfied or satisfied, satisfied, and extremely satisfied”; scaled such that a 1 indicates dissatisfaction and 3 indicates satisfaction. The response options for the benefits questions were “extremely beneficial, beneficial, neither beneficial or harmful, harmful, and extremely harmful”; scaled such that a 1 indicates harmful and 3 indicates beneficial. All respondents were told: “You may have heard that state and local leaders in southeast Michigan are developing plans to remove and replace Interstate 375 in Detroit with a boulevard and business corridor” before answering the support, benefits, and historical knowledge 12 questions. The control condition gave no other information (Low information). In the historical context condition, respondents were also told: “The proposed business corridor is intended to support Black Businesses in what was a prominent historically Black area—Black Bottom and Paradise Valley—that was demolished in the early 1960s to make space for Interstate 375, displacing 130,000 residents.” The low information condition was included as a comparison to understand the extent to which respondents may experience priming effects from the historical context condition. It is possible that by mentioning race or benefits, I may be priming respondents to think about race and history more than they otherwise would.3 Each condition provides entirely true information to respondents. However, the historical context condition highlights the way that federal, state, and local policymakers have discussed the project. To further separate priming effects from personal historical knowledge, I also ask respondents “What is the first name of Mayor Young of Detroit?” Mayor Coleman A. Young was a prominent and influential former mayor of the city of Detroit; the city municipal building is named in his honor. As such, respondents who correctly answer this question have some existing of knowledge of local politics in Detroit. In sum, I compare the effect on the outcome variables, expectations, support, and benefits, by treatment, locality (in this case Wayne County residency status), race (Black versus non-Black), and historical knowledge. Naturally this analysis suffers from small sample sizes as it relates to Black respondents as well as respondents who have historical knowledge outside of Wayne County and those who do not have historical knowledge within Wayne County (See Table 2.2). As such, I will focus on the estimated effects 3 Table 2.2 shows the percentage of respondents in each race x location group that received each version of the treatment and has historical knowledge. It is possible that the historical treatment influenced respondents’ memory, reminding them of what they knew. I do see a higher percentage of respondents with historical knowledge based on the Mayor Young question after they have received the historical treatment. However, my central analysis is to compare treatment effects within prior knowledge groups, so this should have very little influence on the understanding of the study. 13 for each group with less focus on statistical significance in an effort to highlight the nuanced differences. LOCATION RACE TREATMENT TABLE 2.2 BALANCE TABLE HAS N HISTORICAL KNOWLEDGE NO HISTORICAL KNOWLEDGE PERCENT WITH HISTORICAL KNOWLEDGE OUTSIDE WAYNE COUNTY Non- Black Low Information (control) 365 Historical Context (treatment) 391 Black Low Information (control) 43 Historical Context (treatment) 29 WAYNE COUNTY Non- Black Low Information (control) 42 Historical Context (treatment) 57 Black Low Information (control) 36 Historical Context (treatment) 37 149 154 15 15 25 36 22 29 216 40.82 237 39.39 28 14 17 21 14 8 34.88 51.72 59.52 63.16 61.11 78.38 Results Expected Satisfaction for I-375 Replacement Figure 2.3 reports the predicted level of expected satisfaction by historical knowledge, treatment, race, and locality. I examined respondents’ expected satisfaction for the I-375 replacement project on a 3-category scale from 1 to 3. Non-Black respondents living outside of Wayne County hold neutral expectation about the policy that are resistant to the influence of both the historical context provided by the vignettes and a respondent’s historical knowledge about Detroit. 14 Black respondents experienced distinct treatment effects. Black respondents both in and out of Wayne County that have historical knowledge exhibit a parallel relationship with the treatments such that the additional historical context increases expected satisfaction for the project. For respondents receiving new information—that is having low prior knowledge and receiving the historical treatment—those outside of Wayne County expressed greater support than their Black Wayne County counterparts. Black respondents residing in Wayne County experience negative feelings when they receive what is likely new contextual information.4 In sum, the historical context treatment is particularly influential for respondents that have a personal connection to the impacted policy group. Results follow closely for estimated support for the I-375 project (See Appendix Figure A.2). 4 Although, this is a very small subset of the sample (N=8). These feelings are similarly reflected in support (See Appendix Figure A.2). It is possible that the new information was particularly shocking to Black Wayne County residents who had low prior knowledge as they might have experienced a stronger emotional reaction to the negative aspect of the treatment than Black respondents living outside of Wayne County. Black respondents living outside of Wayne County also live outside of the Metro-Detroit area within this sample which furthers their personal proximity to the historical context. 15 FIGURE 2.3 PREDICTED EXPECTED SATISFACTION Note: Expected Satisfaction with the I-375 replacement project is coded as a three-category 1-3 scale from dissatisfied to satisfied. Expected satisfaction is predicted on the interaction between living in Wayne County, being Black, historical knowledge (answering correctly former Mayor Young’s first name), and vignette treatment. Confidence intervals are at the 0.95 confidence level. In addition to expected satisfaction for the project, I also asked respondents how beneficial they think the project is for the families of the displaced, Detroiters, and Metro-Detroiters. Figure 2.4 shows the estimated level of benefit for each group by prior knowledge, treatment, race, and 16 locality.5 In all cases, respondents expressed that the displaced are the least likely to benefit from the I-375 replacement project. Black Detroiters expectation of benefits to Detroiters has a decreasing relationship with how much information they have. Those with low prior information and the low information treatment expressed greater benefits than those with the same background knowledge that received the historical treatment and those with prior historical knowledge regardless of treatment. Even more interesting is the difference in which beneficiary group has the greatest estimated benefit among the comparison groups with prior historical knowledge. Wayne County residents who received the low information control expressed that the project benefited Metro-Detroit more than Detroiters, while those outside of Wayne County expressed that the project was essentially equally beneficial to Detroiters and Metro-Detroiters. Among those with prior historical knowledge who received the historical context treatment, Black Wayne County residents – nearly 81 percent of which are Detroiters in this sample—were the only group not to express that Detroiters were most benefited by the project. While everyone else with historical knowledge displays a pattern of opinion that suggests that they can be influenced through policy messaging to perceive the project as beneficial towards Detroiters, Black Detroiters with historical knowledge do not see themselves as the primary beneficiary of the replacement project regardless of messaging they received about the project. While this study lacks sufficient sample size to suggest any statistical certainty, future studies of increasing size and new local contexts with similar directional findings would suggest that personal proximity to a known political injustice creates persistent negative perceptions that cannot be shifted through 5 Here I focus again on overall patterns as opposed to statistical significance as the study is underpowered for the comparisons being made (See Table 2.2). 17 manipulative messaging and permeate perceptions of public policy before it is even enacted.6 FIGURE 2.4 EXPECTED BENEFIT OF I-375 REPLACEMENT Note: Benefit level of I-375 replacement for the displaced, Detroiters, and Metro-Detroiters is coded as a three-category 1 to 3 scale from harmful to beneficial. Benefit is predicted on the interaction between living in Wayne County, being Black, historical knowledge (answering correctly former Mayor Young’s first name), and vignette treatment. Confidence intervals are at the 0.95 confidence level. Discussion The results presented in this paper make an important contribution to our understanding of how history and identity influence perceptions of government. While this study suffers from a 6 In this case personal proximity is membership to a specific harmed community. Black Wayne County residents/Detroiters are more a part of the harmed community than Black Michiganders outside of Wayne County (in this sample specifically also outside of Metro-Detroit) and non-Black Wayne County residents. 18 lack of statistical power due to small sample sizes within the relevant comparison groups. I argue that the relationships shown are consistent across groups and therefore still important to highlight. My findings suggest that Black Detroiters have a different perception of a major public policy that has been touted to be in their benefit from other citizens both in their racial in-group and local community. Different groups have distinct perceptions of government that are based on historical context and personal identity (proximity to being a targeted group). Replication of the relationships shown would demonstrate that when a particular group has been harmed by government actions in the past their perceptions of public policies are influenced by these outcomes. Furthermore, this study has broader implications for policy feedback research. Most policy feedback studies start at the end of a policy experience and fail to consider citizens pre-existing expectations for policies. However, this study demonstrates that past policies influence expectations about future policies. Though, it is often difficult to capture citizen expectations about policies before they happen, researchers should make a greater effort to consider these expectations and how they might influence final perceptions about policy experiences and hence future participation. Additionally, this study also adds nuance to our knowledge about political messaging. These findings demonstrate that there are cases when positive messaging can have unintended adverse effects. Despite policy makers best efforts to communicate that this project is in their best interest (as quoted in the introduction) and meant to respond to harms created by policies, this project shows that such messaging may not be enough. The referenced beneficiaries of the replacement project—Black Detroiters—were particularly resistant to the positive framing 19 regarding the reconciling of past harms, especially when these respondents had prior historical knowledge. Further investigation should be done to understand in what ways communities are resistant to or responsive to framing about public policies. Although this study focuses on a single locality and a single historical context, I would expect to find similar results in other places with past identity related harms where policy framing relies on symbolic olive branches. However, it is unclear how history and identity influence perceptions of government and policy more generally. It is entirely possible that these relationships are only relevant where the policy context directly relates to the historical harm. 20 CHAPTER 3: WHO’S IN CHARGE? HOW REPRESENTATION IN GOVERNMENT INFLUENCES EXPECTATION OF LOCAL SERVICES I have previously explored how personal experience and history might influence expectations of government; in this paper I explore how institutional representation influences citizen expectations of government. Two important aspects of representation and identity in political institutions are race and class. While education, political experience, and policy preferences are also important characteristics of political leaders, race and class are lenses through which citizens might better relate to the political elite and may even provide proxy information to citizens about a candidate’s policy preference. As such, this paper focuses on how race and class representation influence expected satisfaction with local services. To test this relationship, I conducted a candidate choice experiment embedded in a nationally representative survey in the United States. Following Carnes and Lupu’s (2016) study, I use a conjoint design to ask constituents to choose between two hypothetical candidates, randomly varying candidates’ characteristics, including race and whether they were a lawyer or retail-worker. While the survey experiment does ask about vote choice, this paper is focused on expected satisfaction with government and perceptions of how capable a candidate is of understanding people’s problems and solving them. This study provides a more nuanced discussion of how constituents view of potential elected officials might influence their perception of government performance. The results of this candidate choice experiment show that constituents expect to be more satisfied with local public services if a candidate is also perceived as understanding and having the solution to problems for people like themselves, which is in turn dependent on how congruent a candidate is to the constituent with respect to racial and socio-economic identities. 21 The Influence of Institutional Composition on Expectations Citizens have been shown to have a more trusting and positive views of representatives’ decision making and responsiveness (Arnesen, Duell, and Johannesson 2019; Bobo and Gilliam 1990; Fenno 1978; Pitkin 1967). And there is vast evidence that politicians are more responsive to those that share their personal characteristics at varying levels of government (Burden 2007; Butler and Broockman 2011; Carnes 2012; Sances and You 2017; Whitby 1997). For example, David Broockman’s 2013 study focusing on state legislators intrinsic and motivation to advance constituents’ interests finds that Black legislators showed a greater intrinsic motivation to help a Black constituent than non-Black legislators, even when the extrinsic reward for doing so was low. As such citizens should prefer candidates that are most descriptively like themselves. While there is quite a bit of evidence suggesting that support for descriptively similar representatives is higher and that these elected officials are more representative it is not clear how citizen and representative identities influence perceptions of representative capacity to govern effectively. Typical studies on representation focus on how representation influences trust and vote choice (See Bobo and Gilliam 1990; Tate 2001; Manzano and Sanchez 2010). Meanwhile, studies related to qualification are lacking in important ways. Scholars have defined candidate quality is comprising of competence, empathy, integrity, intelligence, leadership, caring about constituents, past political experience, financial resources, professional connections, and technical experience (Carmichael 1966; Carnes 2016, Funk 1999, Graves and Lee 2000). Studies regarding candidate qualifications either focus on narrow—and quite objective-- markers of qualifications or define the level of qualification for citizens, like in Manzano and Sanchez’s study of Latino candidates and co-ethnic identity, despite evidence that voters are increasingly considering candidates personal traits (Rosenberg et al 1986). It should follow that, perceived 22 qualification for office may also be influenced by an underlying personal element. I expect that candidates who share racial and class identity with a given constituent will be perceived as more qualified for government by the constituent. Focusing on components of qualification for office that are more about capacity to govern—understanding and being able to solve the problems of the constituent—I expect that candidates with a shared racial and class identity will be perceived as more understanding and solution oriented. Data and Experimental Design Candidate choice experiments are useful because they avoid the drawbacks of examining observational data on elections, where a candidate’s social class background might be correlated with many other factors that influence the results of the election. I want to know what voters’ expectations are of candidate performance and if those expectations vary according to the voters’ identity. Conjoint candidate choice experiments—in which researchers ask voters to choose between two hypothetical candidates, randomizing certain aspects of the candidates’ backgrounds or positions—give us one way to identify the causal effect of a candidate’s identity on how voters evaluate the candidate (Hainmueller, Hangartner, and Yamamoto 2015; Hainmueller, Hopkins, and Yamamoto 2014). Closely following Carnes and Lupu’s ( 2016) survey experiment design for understanding voter biases regarding working class candidates, I fielded a candidate choice experiment in July 2022 to a random subset of 836 U.S. respondents in the Cooperative Congressional Election Study, a 50,000-person national stratified sample survey administered by YouGov/Polimetrix. Each respondent received the randomized experiment three times, providing a sample of 2,508 elections. In the candidate choice experiment, survey respondents were presented with a table 23 containing candidate traits for two hypothetical candidates running for city council. Five characteristics were randomly varied: the candidate’s gender (male or female), race/ethnicity (white, black, or Latinx), occupation (retail worker or lawyer), level of political experience (volunteer experience or elected to the school board), and level of training (participated in a candidate training program for three weeks, six months, or never participated). The political party of the candidate is the same as the respondents’ selected political party to alleviate party- based selections. TABLE 3.1 CANDIDATE CONJOINT TREATMENT EXAMPLE Consider the following two [RESPONDENT’S PARTY] hypothetical candidates for city council. Gender Race/ethnicity Job experience Candidate training experience Political experience Candidate A Man Black Lawyer Participated in a three-week training program for people interested in running for office Local School Board Member Volunteers for local charities Candidate B Man White Lawyer None Each trait was randomly varied independently for each of the two candidates. This allows for simultaneous measurement and comparison of the independent effect of each characteristic (Hainmueller et al. 2014). That is, by randomizing each candidate’s occupational background, gender, race, political experience, and training I can compare the effect of each characteristic independently. I can also more closely compare the identity of candidates to respondents. Randomizing each attribute independently also ensures that respondents were not conflating different attributes. After showing respondents the candidate traits, we then ask respondents to select which candidate (1) they would vote for, (2) their neighbor would vote for, (3) better understands the problems facing people like themselves, (4) is more likely to have solutions to solve those 24 problems, and (5) which candidate they thought was more qualified for political office. Specifically, the questions asked, “If you had to make a choice without knowing more, which of the two do you think you would be more likely to vote for?”, “Which of the two would you guess better understands the problems facing people like you?”, “Which of the two would you guess is more qualified for local office?” , and “Which of the two would you expect to work on finding the solution for the problems facing people like you?” I additionally randomized which candidate won the election and asked respondents to “Suppose [Candidate A/Candidate B] has won the election. If you had to guess, how satisfied would you expect to be local public services?” While this experiment’s design closely mimics that of Carnes and Lupu (2016), its end goal and overarching research question differs in an important way. Carnes and Lupu aimed to understand voter preferences regarding working- class candidates, while I aim to add an additional level of nuance in understanding how candidate identity influences perceptions of future policy outcomes. I am most interested in knowing how respondents viewed candidates based on who was more qualified, understanding of problems, and likely to have solutions for their problems, and if these perceptions vary by respondent characteristics. These questions allow me to better understand the development of citizen expectations regarding candidates and potential office holders. I follow the analytical recommendations of Hainmueller et al. (2014) and Carnes and Lupu (2016), treating each hypothetical candidate in each experiment as a unique case. I estimated ordinary least squares regression models relating the outcome variables to indicators for whether the candidate was randomly assigned to have participated in a candidate training program, be a worker, woman, less educated, black, Latinx, or an experienced politician. To 25 account for the nested nature of the candidates (in a two-person election) I cluster standard errors by election (respondent ID x election). Results How do Candidates Influence Expected Satisfaction with Government? Table 3.2 shows the influence of perceiving a candidate as understanding of one’s problems, having solutions to those problems, and individual candidate characteristics on expected satisfaction with local public services for a randomized winning candidate (Model 4). Here, we see that being perceived to understand a citizen’s problems and have to ability to find solutions matter for expected satisfaction with local public services. These models also show evidence that while perceptions of candidate qualification for office influences the expected satisfaction with local services on its own (Model 2), it’s influence cannot be disentangled from that of perceptions of understanding and solution development ability (Model 3). And in most cases, other than for Black candidates and candidates who have participated in a candidate training program, candidate traits have negligible influence on expected satisfaction when accounting for candidate perceptions. In what follows, I explore how these traits influence candidate perceptions directly. 26 Woman Latinx Black Lawyer TABLE 3.2 CANDIDATE PERCEPTIONS INFLUENCE EXPECTED SATISFACTION WITH LOCAL SERVICES (2) Expected Satisfaction 0.0361 (0.0318) -0.0078 (0.0394) 0.0868* (0.0379) -0.0149 (0.0321) 0.0278 (0.0404) 0.1257** (0.0383) 0.0056 (0.0314) 0.2175*** (0.0325) -- (1) Expected Satisfaction 0.0268 (0.0321) 0.0031 (0.0399) 0.0959* (0.0384) 0.0221 (0.0322) 0.0978* (0.0402) 0.1676*** (0.0388) 0.0016 (0.0319) -- (4) Expected Satisfaction 0.0193 (0.0310) 0.0036 (0.0385) 0.0806* (0.0366) 0.0223 (0.0311) 0.0198 (0.0400) 0.1233** (0.0384) -0.0205 (0.0307) -- -- Three Week Training Volunteer Six Month Training Understands Qualified Has Solutions -- -- Constant N Adjusted R2 R2 2.268*** (0.0477) 2,506 0.01444 0.01719 2.212*** (0.0480) 2,504 0.04004 0.04311 0.1735*** (0.0438) 0.1669*** (0.0445) 2.157*** (0.0470) 2,505 0.07117 0.07451 (3) Expected Satisfaction 0.0208 (0.0313) 0.0010 (0.0385) 0.0798* (0.0366) 0.0160 (0.0317) 0.0114 (0.0397) 0.1183** (0.0380) -0.0178 (0.0307) 0.0415 (0.0448) 0.1679*** (0.0442) 0.1470** (0.0516) 2.155*** (0.0473) 2,503 0.07143 0.07514 Standard errors in parentheses. Standard Errors Clustered by Election + p<0.10, * p<0.05, ** p<0.01, *** p<0.001 What Determines Perceptions of Understanding, Solution Development, and Qualifications? While this candidate choice experiment manipulates several candidate characteristics, this paper focuses on how descriptive representation, particularly regarding race and class, influences our perception of candidates.7 Figures two and three show how the interaction between a candidate and citizen's race and a candidate and citizen’s class, respectively, influence 7 See future works from my colleagues Kesicia Dickinson and Erika Vallejo for further discussion of other candidate traits. 27 perceptions about understanding, solution development, and qualifications.8 In Figure 3.1 we see that candidate racial match matters significantly for nonwhite respondents. White respondents are no more likely to say a candidate is qualified, has solutions, or understands their problems based on candidate race. Hispanic respondents prefer Latino candidates to white candidates. Black respondents feel similarly about Black candidates, showing greater distinction between their perception of Black and Latinx candidates than other nonwhite respondents. Other nonwhite respondents also preferred nonwhite candidates to white candidates. FIGURE 3.1 CANDIDATE PERCEPTIONS ARE INFLUENCED BY RACIAL REPRESENTATION Note: Values represent the respondents’ average estimated perception of a hypothetical candidate based on the interaction between a candidate and respondent’s race for women candidates with school board political experience, no candidate training, and lawyer occupation. Error bars represent the 95% confidence interval estimated using standard errors clustered by unique election. Estimates are based on ordinary least squares regression reported in Appendix Table B.2 Panel A: N= 5,012. Panel B: 5,014. Panel C: 5,016. 8 For a general analysis of candidate traits influence on their perceptions, see Appendix Table A.1 28 FIGURE 3.2 PERCEPTIONS OF CANDIDATE QUALITY AND UNDERSTANDING ARE INFLUENCED BY CANDIDATE CLASS Note: Values represent the respondents’ average estimated perception of a hypothetical candidate based on the interaction between a candidate and respondent’s class (occupation x median income) for women candidates with school board political experience, and no candidate training controlling for candidate respondent racial match. Error bars represent the 95% confidence interval estimated using standard errors clustered by unique election. Estimates are based on ordinary least squares regression reported in Appendix Table B.3 Panel A: N= 5,006. Panel B: N= 5,010. 29 Continuing to explore the relationship between class and candidate perceptions, Figure 3.2 panels A and B show the likelihood of saying a candidate is qualified and understands problems by candidate and respondent class and respondent candidate racial match (Appendix Table B.3). Figure 3.3 shows the likelihood of saying a candidate can find the solution for one’s problems. We can see, as addressed above, that racial match increases the likelihood that a respondent will say a candidate is qualified, has solutions, or understands however the influence of class-match varies in distinct ways. Perception of candidates’ general qualifications and ability to understand problems do not vary by respondent income. Qualifications and understanding are dependent on the candidates’ class, in this case occupation. Respondents report that retail-working candidates are more likely to understand their problems, even though lawyers are perceived as being more qualified. FIGURE 3.3 PERCEPTIONS OF CANDIDATES ABILITY TO SOLVE ONE’S PROBLEMS IS INFLUENCED BY CLASS REPRESENTATION Note: Values represent the respondents’ average estimated perception of a hypothetical candidate based on the interaction between a candidate and respondent’s class (occupation x median income) for women candidates with school board political experience, and no candidate training controlling for candidate respondent racial match. Error bars represent the 95% confidence interval estimated using standard errors clustered by unique election. Estimates are based on ordinary least squares regression reported in Appendix Table B.3 N= 5,008. 30 Interestingly, respondents’ faith that a candidate can find a solution for problems facing people like themselves is influenced by the extent to which candidate’s class matches their own. When a candidate’s class matches the respondent’s class—when the candidate is a lawyer and the respondent has income over the median income, or the candidate is a retail worker and the respondent has income below the median—respondents are more likely to say that candidate is more capable of finding a solution for the problems facing people like themselves. When the classes are unaligned respondents are less likely to say candidates can find solutions. Returning to the theory that expectations of local government are influenced by candidate perceptions, I examine the results for expected satisfaction with local government services. Figure 3.4 panels A and B show the influence of constituent- candidate racial and class match on expected satisfaction with local services by perceived candidate qualities (See Appendix Table B.4). Here we see that racial representation positively influences expected satisfaction in similar ways to its influence on perception of candidates – suggesting that the importance of racial and ethnic representation extends beyond the influence of candidate perceptions to expected satisfaction with government. However, the influence of class congruency does not influence expected satisfaction with local services beyond its influence on candidate perceptions. Figure 3.4 highlights that respondents’ perception of candidates’ ability to understand and solve problems for people like themselves along with racial representation are what differentiates their expectations for local services based on who is in office. 31 FIGURE 3.4. PERCEPTIONS OF UNDERSTANDING AND SOLVING PROBLEMS AND RACIAL REPRESENTATION INFLUENCE EXPECTED SATISFACTION WITH LOCAL SERVICES Note: Values represent the respondents’ average estimated expected satisfaction with local services based on (Panel A) the interaction between a candidate and respondent’s race for women candidates with school board political experience, no candidate training, and lawyer occupation; and (Panel B) the interaction between a candidate and respondent’s class (occupation x median income) for women candidates with school board political experience, and no candidate training controlling for candidate respondent racial match. Error bars represent the 95% confidence interval estimated using standard errors clustered by unique election. Estimates are based on ordinary least squares regression reported in Appendix Table B.4 Panel A: N=2,503; Panel B: N= 2,500. 32 Discussion This study is not without limitations; the results draw from one survey experiment about hypothetical candidates. While these pretend elections deviate from the real-world election environment, which would include media messaging from candidates, supporters, and opponents that alter candidate perceptions in real and important ways, they do allow us to answer questions about perceptions of candidates we are not always privy to seeing in actual elections for reasons far beyond lack of political ambition. The results presented in this paper make an important contribution to our understanding of what it means to be a qualified candidate and how perceptions of candidates influence citizens’ expectations of government performance. I show that perceptions of qualification for elected office is not entirely dependent on traditional markers of education or political experience, but that citizens also consider their personal identity alongside a candidate’s identity when evaluating how well a prepared a candidate is to effectively govern. While traditional members of government, like lawyers, might be perceived as being educationally qualified for government, many may believe they miss the mark on other important aspects of governing. These findings show the importance of a more nuanced understanding of citizen needs of politicians and what it means to be fit for elected office. I also empirically demonstrate that expected satisfaction with local public services can be influenced by perceptions of who might be in power. Expected satisfaction with local services is positively related to being descriptively represented (for nonwhite citizens) and perceptions of how well a candidate understands a citizen’s problems and if they are perceived to be able to solve the problem. As described at the beginning of this dissertation, 33 these expectations have influence of experienced satisfaction and furthermore, the way people choose to participate politically. 34 CHAPTER 4: PARTICIPATION AS A FUNCTION OF EXPECTATIONS AND SATISFACTION The preceding chapters of this dissertation have explored the ways in which different people might develop different expectations of government performance. This knowledge begs the question: how do expectations of policy outcomes influence political participation? Evidence from Policy Feedback literature shows that policy experiences influence perceptions of government and perceptions of citizens’ roles, and hence participation, within government (Campbell, 2003; Mettler, 2002; Mettler and Soss, 2004; Michener, 2018; Pierson, 1993; Sharp, 2012; Soss, 1999). Public administration scholars have found that citizens hold both predictive and normative expectations (Hjortskov, 2020). Citizens have ideas about what government is capable and willing to provide and they also have needs or goals of government. Although, it has generally been understood that citizen satisfaction is a function of both the citizen’s experience with the policy, outcome, or office being evaluated and the citizen’s existing ideas of what should happen, it is less understood how predictive expectations—expectations of what government is going to provide—fit in the model of citizen satisfaction (Favero & Kim 2021; Jacobsen et. al 2015; James 2007, 2009, 2011). Understanding how citizens form their evaluations of government would not only inform strategies to improve perceptions of government but it could create a more responsive and democratic political system. In this study, I aim to understand how pre-existing expectations of how government will perform and post-experience satisfaction with policy influence political participation. Specifically, is satisfaction with services independent of pre-existing expectations? And how does the relationship between satisfaction and expectations of government performance influence intended political participation? 35 Incorporating Expectations into the Feedback Model While existing policy feedback studies provide a great deal of knowledge about the relationship between policies and participation, they often begin to explore this relationship in the aftermath of a policy experience, forgoing consideration for a citizen’s expectations for a policy experience. Figure 4.1 presents my proposed additions to the existing policy feedback framework. Beginning with the existing framework discussed above, for each new policy experience citizens hold a certain level of satisfaction influenced by a combination of resource and interpretive effects from the policy experience that can induce different forms of political participation (link B). My model provides a new contribution-- showing that not only do perceptions of policy experiences influence the ways citizens participate politically, but the experience also influences future expectations (link C). The study of expectations and satisfaction by public administration scholars supports this argument as well as the relationship suggested by link A—expectations influence satisfaction (Favero and Kim ,2021; Hjortskov 2019; Van Ryzin, 2004). Taking these relationships all together I argue that expectations also have a mediated relationship with political participation—as expectations shape satisfaction levels with each new policy experience. Naturally, the most direct effect of expectations on participation is absorbed by the effect of satisfaction on participation. However, as Hjortskov (2019) finds, predictive expectations (and hence satisfaction) are mutable. I argue that it is important to understand the relationship between the change in satisfaction—that is satisfaction minus expected satisfaction—and political participation. Understanding this relationship may help clarify some of the motivations behind political participation—in particular how experiences with new policies confirm or refute prior expectations of government. 36 FIGURE 4.1 NEW THEORETICAL MODEL FOR POLICY FEEDBACK Hypotheses Political science scholars have long noted the declining trust in government within the United States and corresponding decreases in participation (Lerman, 2019). I expect that citizens with lower expected satisfaction will have lower levels of experienced satisfaction compared with those who express high expectations of satisfaction. Additionally, I expect that, increased external efficacy, or more positive expected satisfaction, will be associated with political engagement like voting or contributing to public services monetarily (Nelson, 2021; Verba, Schlozman, and Brady, 1995). Decreasing efficacy or negative expectations are associated with increased use of public voice, like protests (Nelson, 2021). I expect that, accounting for existing expectations of what will happen, a negative level of experienced satisfaction will lead to use of public voice while a positive level of experienced satisfaction will lead to increased political engagement like voting. Accordingly, I expect that confirmation of low expectations and low external efficacy will be met with increasing usage of public voice, while having a better experience than expected will lower the usage of public voice. I expect a greater likelihood of 37 positive participation as the difference between expectations and satisfaction moves in a positive direction. Data and Experimental Design To answer these questions, I fielded a survey experiment through Lucid Theorem, an online survey vendor, during February and March of 2022. The survey has 1477 respondents from a national representative sample.9 To understand how pre-experience expectations and post-experience satisfaction influence participation choice, I conduct a randomized experiment such that I randomize the valence of policy experience that each respondent receives during a city initiative to renovate a local park and recreation facility. This allows me to explore how satisfaction can be influenced by pre- existing ideas of performance as well as policy experience. Respondents were prompted to “Suppose that your community has just released plans to renovate the city's parks and recreation facilities.” They were then asked about their goals for the project as well as their predicted expectations of the outcome of the project within their city in regard to five general aspects (timeliness, safety, quality, community input, and budget). I then randomly assign them to treatment based on their response to the question “How satisfied do you think you will be with this project?” Respondents were able to answer this question on a 5-point Likert scale, such that 1 corresponds to expecting to be extremely dissatisfied, 3 somewhat satisfied, and 5 extremely satisfied. While this paper does not discuss the findings regarding these goals and predicted expectations, I argue that they prime the respondents to think objectively about satisfaction with the local project. By asking respondents about specific aspects of the project, I am asking them to isolate these tangible outcomes as they think about their level of overall level of satisfaction. 9 Lucid Theorem relies on demographic quotas to create more nationally representative samples. Please see their FAQs for more details on how they select survey audiences. https://lucidtheorem.com/faq 38 Based on overall expected level of satisfaction respondents were then evenly and randomly placed into one of two treatment conditions or the control group based on if they expected to be satisfied, dissatisfied, or somewhat satisfied.10 There were 491 respondents in the control condition, 490 in the positive condition, and 496 in the negative condition. Due to random sampling and random assignment, the respondents in each experimental condition are representative of the national population and do not differ substantially in their demographics or political attitudes across treatments. The treatments for the experiment were a positive and negative experience, while the control group received no treatment and were instead immediately asked about how they would like to participate. Below are the vignettes for each treatment condition. Positive Vignette: Your city just reopened the parks and recreation facilities after completing renovation within its proposed 6-month timeline and just within budget. The city was able to include your community’s requested pool, though not to the Olympic standard size requested. The parks have well-manicured lawns, ADA compliant pathways and new rubber tiled surface to accompany the brand-new play structure. Negative Vignette: Your city has halted renovations of the parks and recreation facilities after a 6-month extension to the originally proposed timeline. The project has run out of funding and exceeded the budget. The pool your community requested will not be built, the play structure failed to meet basic safety requirements, and the grounds have failed to meet ADA compliance. Following the treatment prompts I asked those that received the positive and negative treatment how accurate it was that each of the previous five elements of the project occurred and then I asked “How satisfied are you with this project?”. Respondents answered these questions in the same way they answered the pre-treatment questions. The “how satisfied” question is intended to be a direct comparison to the expected satisfaction question, enabling me to 10 Please see appendix for Balance Tables and the full survey instrument 39 understand the relationship between expected satisfaction and experienced satisfaction. To understand how influential the expected experience and the experience with government are to the choice to participate, I ask all respondents “Based on the described city project, how would you like to proceed?” Respondents were asked to choose options related to political engagement, public voice, cognitive engagement, and civic engagement (Cohen, 2010; Nelson, 2021; Zukin et al. 2006). Respondents were allowed to select all of the options that applied to them. I also allowed the respondents to choose not to participate. I presented one choice from each category specifically related to the prompt scenario. In this paper, I focus on political engagement and public voice specifically. If a respondent chose that they would like to “give money to support local parks and recreation” I consider that choosing to engage politically as it is most proximal to a campaign contribution. And I consider choosing to “Publicly express discontent” to be an expression of public voice. Each of these variables is binary. In order to understand how pre-existing expectations influence satisfaction with services, I regress experienced satisfaction on expected satisfaction controlling for political party, race, income, education. I also estimate the same model including a treatment interaction to understand if there is a moderation effect by the type of policy experience. In order to understand how pre-existing expectations and post-experience satisfaction with government influence political participation, I rely on the control group to understand how citizens may choose to participate in the absence of having an experience, in this way I am able to focus on the way that initial expectations influence participation. I compare that with participation choices of those in the treatment groups to understand how mutable these choices are and in what ways. 40 Results The Effects of Expected Satisfaction and Treatment on Experienced Satisfaction This experiment allows for the investigation of the effects of different policy experiences on satisfaction. Along an increasing Likert scale, where a value of 1 is equal to Extremely Dissatisfied and 5 equals Extremely Satisfied, the average level of expected satisfaction prior to treatment is about 3.8 (Satisfied) in all treatments. The average level of expressed satisfaction among those in the positive condition is 4.08 (Satisfied) and 2.71 (Somewhat satisfied) in the negative condition. While it is clear the experiment produces differences in satisfaction of policy experience (as intended), research question one aims to disentangle the extent to which a person’s expected satisfaction interacts with the policy experience to influence their evaluation of a given policy experience (experienced satisfaction). Table 4.1 reports the multivariate results, examining the impact of the interactions between treatment and expected satisfaction. Model 1 shows that expecting to be satisfied has a significant increasing positive relationship with experienced satisfaction. For each increasing level of expected satisfaction, the respondent could be expected to increase their level of experienced satisfaction by 0.614. These effects hold when accounting for treatment (Model 3). To better understand how expected satisfaction may have a varied influence depending on the valence of an experience, I interact the treatment with expected satisfaction in model four (model 5 includes additional control variables). 5 For ease of interpretation, I turn to Figure 4.2 which visualizes the marginal effect of each treatment condition on experienced satisfaction at varying levels of predicted satisfaction with all other covariates held at their means. I find support that predicted expectations have a positive relationship with experienced satisfaction. 41 Here we see that regardless of treatment as predicted satisfaction increases, satisfaction with the policy experience also increases. The policy condition received in the treatment moves the starting point for experienced satisfaction, as expected. 42 Expected Satisfaction Positive Treatment Negative Treatment Negative x Expected Satisfaction Republican Nonwhite Household Income Education TABLE 4.1 EXPECTED SATISFACTION AND TREATMENT EFFECTS ON EXPERIENCED SATISFACTION (1) 0.614*** (0.049) (2) (3) 0.589*** (0.042) (4) 0.460*** (0.061) (5) 0.439*** (0.063) ref. ref. ref. ref. -1.335*** (0.068) -2.262*** (0.330) -2.324*** (0.339) - 1.364** * (0.075) 0.240** (0.083) 0.256** (0.086) -0.015 (0.011) 0.270*** (0.080) -0.015* (0.006) -0.020 (0.025) 1.790*** (0.169) 2.290*** (0.242) 2.557*** (0.276) _cons 1.019*** (0.194) 4.078** * (0.053) 983 0.254 983 N 0.380 r2 Note:Standard errors in parentheses + p<0.10, * p<0.05, ** p<0.01, *** p<0.001 983 0.137 983 0.385 934 0.398 43 FIGURE 4.2 MARGINAL EFFECT OF EXPECTED SATISFACTION ON EXPERIENCED SATISFACTION BY TREATMENT Note: The figure depicts the average marginal effect on experienced satisfaction based on model 5 in Table 2. Expected Satisfaction and Experienced Satisfaction are both coded as a five point scale from extremely dissatisfied to extremely satisfied. Non-Participation Figure 4.3 shows the effect of treatment and expected and experienced satisfaction on the likelihood that a respondent chooses to not engage with the renovation project in any of the ways listed (See Appendix Table C.5).11 Here we see that treatment does not significantly change the likelihood of choosing not to engage. The influence of experienced satisfaction is negative and consistent across treatment groups (Figure 4.3B). Having a more satisfying policy experience than expected increases the likelihood of choosing not to participate. Figure 4.3 shows the effect of treatment and expected and experienced satisfaction on the likelihood that a respondent chooses to not engage with the renovation project at all (See Appendix Table C.5) . Here we see that treatment does not significantly change the likelihood of choosing not to engage. The 11 To clarify the phrasing of the question allows respondents to pick ways to participate with the renovation, and this section refers to those who chose to do “none of the above” options. While I argue that the options given cover a wide range of political engagement activities—and therefore interpret these results as choosing not to participate, it is possible to interpret that option as simply not choosing to express discontent, donate, discuss with friends and family, or volunteer. 44 influence of experienced satisfaction is negative and consistent across treatment groups (Figure 4.3B). Having a more satisfying policy experience than expected increases the likelihood of choosing not to participate. FIGURE 4.3 MARGINAL EFFECTS ON NON-PARTICIPATION LIKELIHOOD Note: The figure depicts the average marginal effect on likelihood of choosing not to politically participate based on the linear probability models Appendix Table C.5. Non-Participation is a binary indicator for choosing not to participate. Positive Engagement- Donating to the Cause Figure 4.4 shows that politically engaging (providing monetary donation to the local parks and recreation board) is largely dependent on the change in satisfaction one experiences due to the policy (Appendix Table C.6). Figure 4.3 Panel A shows the marginal effect of treatment on likelihood of participating with control variables at the means. The effect on likelihood of choosing to make a monetary donation to their parks and recreation board is marginally lower for the negative treatment group than both the positive treatment and control groups. Figure 4.3 Panel B shows the effects by treatment across experienced satisfaction level. Here we see that experienced satisfaction has a significant positive relationship with the likelihood of participating politically. This relationship is consistent across treatment groups. All together we see that the change in satisfaction after treatment has a significant positive relationship with the likelihood of 45 participating politically; this does not vary by treatment. A one level change in satisfaction—for example, from Somewhat Satisfied to Satisfied—is associated with a 3.5 percentage point increase in likelihood of providing a monetary donation. FIGURE 4.4 MARGINAL EFFECTS ON DONATION LIKELIHOOD Note: The figure depicts the average marginal effect on likelihood of donating to local parks and recreation based on linear probability models in Appendix Table C.6. Political Engagement in a binary indicator for choosing to donate to local parks and recreation. Critical Engagement- Expressing Discontent Figure 4.5 shows the effect of treatment and experienced satisfaction on the likelihood that a respondent chooses to express discontent with the project (Appendix Table C.7). Here the negative treatment significantly increases the likelihood of using public voice. Those who received the negative treatment were 14 percentage points more likely to say they would use their public voice compared to the control group and 9.6 percentage points more likely than those who received the positive treatment. The relationship between experienced satisfaction and policy experience varies by treatment. For those who received the negative policy condition, those who were more satisfied were less likely to want to publicly express discontent; while experienced satisfaction did not matter for those who received the positive condition—they were generally unlikely to want to express discontent. The findings also support the theory that changes in satisfaction matter, regardless of treatment. 46 FIGURE 4.5 MARGINAL EFFECTS ON PUBLIC VOICE LIKELIHOOD Note: The figure depicts the average marginal effect on likelihood of publicly expressing discontent based on linear probability models in Table C.7. Public Voice in a binary indicator for choosing to publicly express discontent. Discussion This study demonstrates that a policy experience itself is not the sole determinant of constituent satisfaction with a government provided good or service. Constituents expected level of satisfaction with government is related to their future evaluations of government. While the effects of the treatments on experienced satisfaction are unsurprising and intended, the influence of policy perceptions follow the common theories on policy feedback and political participation. In fact, only considering the influence of satisfaction without expectations of the policy experience may overstate the influence of the policy experience. There are some limitations with survey experiments that should be considered. Methodologists have raised concerns that respondents have little reason to realistically respond to a hypothetical scenario. However, by asking them to rely on their actual local context and asking about objective measures of performance, I argue that respondents have very little reason to respond differently than they would in the real world. Prior to asking about satisfaction based on the scenario vignettes, I ask respondents the level of which they agree that five objective measures (safety, time, quality, budget, and community input) existed within the project. In a 47 sense this primes them to think objectively about tangible policy outcomes rather than the additional political context in which the project may be occurring locally. I argue this is a divergence from typical political science evaluations that are often more focused on the perceptions of government officials and institutions as opposed to evaluations of policy outcomes. In this way this study aides the understanding of how policy outcomes can influence political participation and the democratic process. This should be considered in conjunction with policy feedback as it relates to constituent’s experience with policy institutions. 48 CHAPTER 5: CONCLUSION In this dissertation, I aimed to understand specific determinants of expectations of government performance and how expectations, more generally, influence political participation. I also argued that the discipline should pay more attention to expectations of policy output, not just expectations of candidate behavior or how well candidates ideology matches with constituents. The second chapter explores possible implications for real world policy initiative in Detroit, Michigan. The nature of the study may provide useful information to policymakers as they work to implement the I-375 replacement project. For this dissertation, it provides evidence that citizens do not forget what has happened in the past, they still consider it when they when evaluate current policies, and those who have experienced policies differently will evaluate accordingly. Contextual information provided by policymakers may shape perceptions of the project for individuals without prior historical knowledge of direct experiences; however, for individuals who are informed and rooted in local knowledge, policy-maker framing may not be enough to shift attitudes. The conjoint experiment extended in Chapter 3 presents further insight into citizens evaluation of candidates, providing a more nuanced discussion of what it means to be qualified. Candidates need to be able to do more than look or act the part of a politician. Constituents need to feel that representatives understand and can solve their problems to have satisfactory local services. These perceptions of candidates are shaped by race and ethnicity—particularly when respondents who are Black or Latinx share the racial/ethnic identity of the candidate. Chapter 4 highlights the implications of citizens expectations of government performance. Expectations influence final evaluations and mediate how people choose to politically engage. 49 While citizens may respond to differing outcomes of a project, their prior expectations will often set them up to be more or less satisfied, regardless of the reported outcomes. These findings contribute to our understanding of policy feedback theories, which have not traditionally incorporated measures of expectations. The unique and underexplored perspectives presented here marry several different bodies of scholarship. I have applied a theory most often used in public administration studies (the expectancy disconfirmation theory of satisfaction) to understand policy feedback and political participation. I've extended the discussion of candidate perceptions beyond understanding who citizens vote for to also understanding how satisfied they expect to be with policies produced (at least at the local level) based on the interaction of their own and candidates’ identities. I have also drawn on literatures dealing with racial representation in politics as well as policy feedback scholarship that addresses race and unequal experiences with policies. Perspectives on Expectations of Government Performance and Policy Scholarship The studies conducted in this dissertation inspire questions about how resistant citizens are to change in their perspective. Despite the findings shown in Chapter 4, I think it would be unfair to believe that citizens are entirely resistant to changing their mindset about government. Public administration scholars have shown that citizens do incorporate information into the development of their expectations (Favero and Kim, 2021), In this dissertation I explore the influence citizens’ information about the identity of government leaders, themselves, and identity-based history on expectations. Naturally, introducing new positive information might shift growing negative expectations. Shifting the status quo of how things are done and providing that information might improve citizens’ expectations. My study of perceptions in southeastern Michigan in Chapter 2 provides interesting future 50 directions of study. The current study and policy context explores one condition of policy information, but what happens to expectations when the key information is different? That is— what if the leaders supporting the policy or the goals of the policy were different? Citizens might not be so resistant to changing perceptions about government if the way it produced policy also changed. These questions present the opportunity for political science, public administration, and policy scholars to explore what it is citizens really want from government? What kind of new information do citizens responsive to positively? We need not wait for the end of a policy experience to understand if citizens were happy with the experience. Public opinion and qualitative studies could provide useful information to understand what is and is not working about the way that government functions. It is my hope that this dissertation speaks to a larger discussion of how to make government and leaders more effective with the goal of a more equal democracy. To understand the answer to that quandary more fully would require the interaction of political science, public administration, and policy evaluation scholars. 51 BIBLIOGRAPHY “All Things Considered.” 2021. In Tuskegee, Painful History Shadows Efforts To Vaccinate African Americans. https://www.npr.org/2021/02/16/967011614/in-tuskegee-painful- history-shadows-efforts-to-vaccinate-african-americans. Arnesen, Sveinung, Dominik Duell, and Mikael Poul Johannesson. 2019. “Do Citizens Make Inferences from Political Candidate Characteristics When Aiming for Substantive Representation?” Electoral Studies 57: 46–60. Avery, James M. 2006. "The Sources and Consequences of Political Mistrust Among African Americans." American Politics Research 34(5). ———. 2009." Political Mistrust Among African Americans and Support for the Political System." Political Research Quarterly 62(1). Beland, Daniel. 2010. “Reconsidering Policy Feedback: How Policies Affect Politics.” Administration & Society 42(5): 568–90. Bobo, Lawrence, and Franklin D. Gilliam. 1990. “Race, Sociopolitical Participation, and Black Empowerment.” The American Political Science Review 84(2): 377. Brady, Henry E., Sidney Verba, and Kay Lehman Schlozman. 1995. “Beyond SES: A Resource Model of Political Participation.” American Political Science Review 89(02): 271–94. Broockman, David. 2013. “Black Politicians Are More Intrinsically Motivated to Advance Blacks’ Interests: A Field Experiment Manipulating Political Incentives.” American Journal of Political Science 57 (3): 521–36. Burden, Barry C. 2007. Personal Roots of Representation. Princeton, NJ: Princeton University Press. Butler, Daniel, and David Broockman. 2011. “Do Politicians Racially Dis- criminate against Constituents? A Field Experiment on State Legisla- tors.” American Journal of Political Science 55 (3): 463–77. Campbell, Andrea Louise. 2002. “Self-Interest, Social Security, and the Distinctive Participation Patterns of Senior Citizens.” American Political Science Review 96(3): 565–74. ———. 2003. “Participatory Reactions to Policy Threats: Senior Citizens and the Defense of Social Security and Medicare.” Political Behavior: 21. Carmichael, Stokely. 1966. “Who Is Qualified?” The New Republic. 52 Carnes, Nicholas, and Noam Lupu. 2016. “Do Voters Dislike Working-Class Candidates? Voter Biases and the Descriptive Underrepresentation of the Working Class.” American Political Science Review 110(4): 832–44. Carnes, Nicholas. 2012. “Does the Numerical Underrepresentation of the Working Class in Congress Matter? Evidence from Roll Call Voting in the House of Representatives.” Legislative Studies Quarterly 37(1): 5–34. ———. 2016. “Why Are There so Few Working-Class People in Political Office? Evidence from State Legislatures.” Politics, Groups, and Identities 4(1): 84–109. Cohen, Cathy J. 2010. “Democracy Remixed: Black Youth and the Future of American Politics.” New York, NY: Oxford University Press. Davidson, Roger H. 1970. "Public Perceptions for the Job of the Congressman." Midwest Journal of Political Science. Dembosky, April. 2021. “It’s Not Tuskegee. Current Medical Racism Fuels Black Americans’ Vaccine Hesitancy.” Los Angeles Times. https://www.latimes.com/science/story/2021- 03-25/current-medical-racism-not-tuskegee-expls-vaccine-hesitancy-among-black- americans. Edelman, Murray Jacob. 1985. The Symbolic Uses of Politics. University of Illinois Press. Favero, Nathan, and Minjung Kim. 2021. “Everything Is Relative: How Citizens Form and Use Expectations in Evaluating Services.” Journal of Public Administration Research and Theory 31(3):561–77. Fenno, Richard F. 1978. Home Style: House Members in Their Districts. Boston: Little, Brown. Funk, Carolyn. 1999. “Bringing the candidate into models of candidate evaluation.” Journal of Politics 61:700-20. Grant, Emma C. 2022. "Black Bottom and Paradise Valley: The Intersection of Race, Class, and Memory in Twentieth Century Detroit," The Great Lakes Journal of Undergraduate History.8(1). https://scholar.uwindsor.ca/gljuh/vol8/iss1/6 Grant, J Tobin and Thomas J. Rudolph. 2004. "The Job of Representation in Congress: Public Expectations and Representative Approval." Legislative Studies Quarterly 29(3). Graves, Scott, and Jongho Lee. 2000. “Ethnic underpinnings of voter preference: Latinos and the 1996 Senate elections in Texas.” Social Science Quarterly 87:227-36. Griffin, John and Patrick Flavin. 2007. "Racial Differences in Information, Expectations, and Accountability." The Journal of Politics. 53 Hainmueller, Jens, Daniel J. Hopkins, and Teppei Yamamoto. 2014. “Causal Inference in Conjoint Analysis: Understanding Multidimensional Choices via Stated Preference Experiments.” Political Analysis 22(1): 1–30. Hainmueller, Jens, Dominik Hangartner, and Teppei Yamamoto. 2015. “Validating Vignette and Conjoint Survey Experiments against Real-World Behavior.” Proceedings of the National Academy of Sciences 112(8): 2395–2400. Herrnson, P. S., Claassen, R. L., Niemi, R. G., & Patterson, K. D.2013."Exceeding Expectations? Determinants of Satisfaction with the Voting Process in the 2008 U.S. Presidential Election." The Journal of Politics, 75(2), 451–463. Hjortskov, Morten. 2020. “Interpreting expectations: Normative and predictive expectations from the citizens’ viewpoint.” Journal of Behavioral Public Administration 3 (1): 11. doi:10.30636/ jbpa.31.72 Huddy, L., & Feldman, S. 2006. “Worlds apart: Blacks and whites react to Hurricane Katrina.” Du Bois Review: Social Science Research on Race, 3(1), 97-113. 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(3):831–48. James, Oliver. 2007. “Evaluating the Expectations Disconfirmation and Expectations Anchoring Approaches to Citizen Satisfaction with Local Public Services.” Journal of Public Administration Research and Theory 19(1): 107–23 ———.2009. “Evaluating the expectations disconfirmation and expectations anchoring approaches to citizen satisfaction with local public services.” Journal of Public Administration Research and Theory 19 (1): 107–23. ———. 2011. “Managing citizens’ expectations of public service performance: Evidence from observation and experimentation in local government.” Public Administration 89 (4): 1419–35. Kimball, David C. and Samuel C. Patterson. 1997. "Living Up to Expectations: Public Attitudes Toward Congress." The Journal of Politics 59(3):701-28 Lawerence, Eric. 2022. “I-375 Replacement Project in Detroit Moves Closer to Reality, Gets OK from Feds.” Detroit Free Press. https://www.freep.com/story/news/local/michigan/detroit/2022/03/17/interstate-375- replacement-detroit-black-bottom-paradise-valley/7073405001/. Lawless, Jennifer L., and Richard L. Fox. 2001. “Political Participation of the Urban Poor.” Social Problems 48(3): 362–85. 54 Lerman, A. E. 2019. Good enough for government work: The public reputation crisis in America (and what we can do to fix it).University of Chicago Press. Lowi, Theodore J. 1972. “Four Systems of Policy, Politics, and Choice.” Public administration review 32(4): 298–310. Mansbridge, Jane. 1999. “Should Blacks Represent Blacks and Women Represent Women? A Contingent ‘Yes’.” Journal of Politics 61(3): 628–57. Manzano, Sylvia and Gabriel R. Sanchez. 2010. “Take One for the Team? Limits of Shared Ethnicity and Candidate Preferences.” Political Research Quarterly 63(3): 568-580. Mead, Lawrence. 1997. The New Paternalism: Supervisory Approaches to Poverty. Brookings Institute. Mettler, Suzanne and Joe Soss. 2004. “The Consequences of Public Policy for Demo- cratic Citizenship: Bridging Policy Studies and Mass Politics.” Perspectives on Politics 2(1):55– 73. Mettler, Suzanne. 2002. “Bringing the State Back In to Civic Engagement: Policy Feedback Effects of the G.I. Bill for World War II Veterans.” American Political Science Review 96(02): 351–65. Michener, Jamila. 2018. Fragmented Democracy: Medicaid, Federalism, and Unequal Politics. Cambridge University Press. Morgeson, Forrest V. 2013. “Expectations, disconfirmation, and citizen satisfaction with the US federal government: Testing and expanding the model.” Journal of Public Administration Research and Theory 23 (2): 289–305. Nelsen, Matthew D. 2021. “Teaching Citizenship: Race and the Behavioral Effects of American CivicEducation.” The Journal of Race, Ethnicity, and Politics 6(1): 157–86. Oliver, Richard L. 2010. Satisfaction: A behavioral perspective on the consumer, 2nd ed. Armonk, NY: ME Sharpe. Pew Research. “The Growing Diversity of Black America.” Accessed March 14, 2022. https://www.pewresearch.org/social-trends/2021/03/25/the-growing-diversity-of-black- america/ Pierson, Paul. 1993. “When Effect Becomes Cause: Policy Feedback and Political Change.” World Politics 45(04): 595–628. Pitkin, Hanna F. 1967. 75 The Concept of Representation. Univ of California Press. 55 Rosenberg, Shawn W., Lisa Bohan, Patrick McCafferty, and Kevin Harris. 1986. “The image and the vote: The effect of candidate presentation on voter preference.” American Journal of Political Science 30:108-27. Rosenstone, Steven J, and John M. Hansen. 1993. Mobilization, Participation, and Democracy in America. Macmillan. Rosenthal, Aaron.2021."Submerged for Some? Government Visibility, Race, and American Political Trust". Perspectives on Politics. Sanchez, Michael W. and Hye Young You. 2017. “Who Pays for Government? Descriptive Representation and Exploitative Revenue Sources.” Journal of Politics 79(3). Schnaz, Jenn. 2021. “Past and Future: Gov. Whitmer Bids to Revamp I-375 Back into an Urban Boulevard.” WXYZDetroit. https://www.wxyz.com/news/governor-whitmer-bids-to- transform-i-37-back-into-an-urban-boulevard. Schneider, Anne, and Helen Ingram. 1993. “Social Construction of Target Populations: Implications for Politics and Policy.” American political science review 87(2): 334–47. Sharp, Elaine B. 2012. 19 Does Local Government Matter?: How Urban Policies Shape Civic Engagement. University of Minnesota Press. Shklar, Judith N. 1995. American Citizenship: The Quest for Inclusion. Harvard University Press. Skocpol, Theda. 1991. “Targeting within Universalism: Politically Viable Policies to Combat Poverty in the United States.” The urban underclass 411(411): 437–59. Soss, Joe. 1999. “Lessons of Welfare: Policy Design, Political Learning, and Political Action.” American Political Science Review 93(2): 363–80. Syed, Ben. 2015. "How Do Citizens Evaluate Public Officials? The Role of Performance and Expectations of Political Trust." Political Studies 63(S1):73-90 Tate, Katherine. 2001. “The political representation of blacks in Congress: Does race matter?.” Legislative Studies Quarterly 26:623-38. ———. 2003. Black Faces in the Mirror: African Americans and the Perspectives in the U.S. Congress. Princeton University Press Van Ryzin, Gregg G. 2004. “Expectations, Performance, and Citizen Satisfaction with Urban Services.” Journal of Policy Analysis and Management 23(3): 433–48. Verba, Sidney, Kay Lehman Schlozman, and Henry Brady. 1995. “Voice and Equality: Civic Voluntarism in American Politics.” Cambridge, MA: Harvard University Press. 56 Waterman, Richard w., Hank C. Jenkins-Smith and Carol L. Silva. 1999. "The Expectations Gap Thesis: Public Attitudes Toward an Incumbent President."The Journal of Politics 61(4): 944-966 Whitby, Kenny J. 1997. The Color of Representation: Congressional Behavior and Black Interests. Ann Arbor: University of Michigan Press. Whitmer Sec Buttigieg and Local Leaders Celebrate Historic I-375 Project. 2022. Retrieved 22 April 2023, from https://www.michigan.gov/whitmer/news/press- releases/2022/09/15/whitmer-sec-buttigieg-and-local-leaders-celebrate-historic-i-375- project Wilson, William Julius. 1991. “Public Policy Research and the Truly Disadvantaged.” The urban underclass: 460–81. Zukin, Cliff, Scott Keeter, Molly Adolina, and Krista Jenkins. 2006. “A New Engagement? Political Participation, Civic Life, and the Changing American Citizen.” New York, NY: Oxford University Press. 57 APPENDIX A: SUPPLEMENTAL MATERIAL FOR CHAPTER 2 FIGURE A.1 MAP OF MICHIGAN Map of the state of Michigan. Macomb, Wayne, and Oakland Counties combined make up Metro-Detroit. Detroit, MI is within Wayne County. The SOSS surveys the entire state. Source: Social Explorer. 58 FIGURE A.2 PREDICTED SUPPORT FOR I-375 REPLACEMENT Support for I-375 replacement is coded as a three-category 1 to 3 scale from oppose to support. Support is predicted on the interaction between living in Wayne County, being Black, historical knowledge (answering correctly former Mayor Young’s first name), and vignette treatment. Confidence intervals are at the 0.95 confidence level. 59 Survey Experiment Questionnaire Vignettes: (randomize- split sample between two vignettes below- for questions 12-16) Examples: [Low information(control)] ½ sample gets this information You may have heard that state and local leaders in southeast Michigan are developing plans to remove and replace Interstate 375 in Detroit with a boulevard and business corridor. [Historical Context (treatment)] ½ sample gets this information You may have heard that state and local leaders in southeast Michigan are developing plans to remove and replace Interstate 375 in Detroit with a boulevard and business corridor. The proposed business corridor is intended to support Black Businesses in what was a prominent historically Black area—Black Bottom and Paradise Valley—that was demolished in the early 1960s to make space for Interstate 375, displacing 130,000 residents. 12. How strongly do you support or oppose this project? • Strongly support • Support • Neither support nor oppose • Oppose • Strongly oppose 13. How satisfied do you think you will be with this project? • Extremely Satisfied • Satisfied • Neither satisfied nor dissatisfied • Dissatisfied • Extremely Dissatisfied 14. How beneficial do you think the I-375 replacement project is for residents of Metro-Detroit? 15. How beneficial do you think the I-375 replacement project is for residents of the City of Detroit? 16. How beneficial do you think the I-375 replacement project is for the families and descendants of those that were displaced during the development of I-375? • Extremely Beneficial • Beneficial • Neither Beneficial nor Harmful • Harmful • Extremely Harmful 17. What is the first name of Mayor Young of Detroit? • Andrew • Coleman • Dennis • Mike • I Don’t Know 60 APPENDIX B: SUPPLEMENTAL MATERIAL FOR CHAPTER 3 TABLE B.1 CANDIDATE TRAITS INFLUENCE ON CANDIDATE PERCEPTIONS Woman Latinx Black Lawyer Six Month Training Three Week Training Volunteer Constant (1) Qualified -0.0033 (0.0170) 0.0360. (0.0211) 0.0624** (0.0215) 0.1487*** (0.0176) 0.3162*** (0.0207) 0.1698*** (0.0204) -0.0069 (0.0170) 0.2370*** (0.0203) 5,012 0.08942 0.09069 (2) (3) Understanding Has Solution 0.0387* (0.0175) 0.0152 (0.0218) 0.0580** (0.0220) -0.0563** (0.0182) 0.1980*** (0.0218) 0.1117*** (0.0213) 0.0563** (0.0172) 0.3554*** (0.0218) 5,016 0.03410 0.03545 0.0374* (0.0175) 0.0426* (0.0216) 0.0859*** (0.0220) 0.0206 (0.0182) 0.2137*** (0.0218) 0.1361*** (0.0211) 0.0377* (0.0173) 0.2952*** (0.0213) 5,014 0.03735 0.03869 N Adjusted R2 R2 Standard errors in parentheses. Standard Errors Clustered by Election + p<0.10, * p<0.05, ** p<0.01, *** p<0.001 FIGURE B.1 TASK CARRYOVER EFFECTS PERCEIVED CANDIDATE UNDERSTANDING Note: Predicted perception of candidate understanding of “problems facing people like you” by election task. 61 TABLE B.2 CANDIDATE TRAITS INFLUENCE ON CANDIDATE PERCEPTIONS, BY RACIAL MATCH (1) Qualified (2) Understanding (3) Has Solution Candidate Race: Latinx Black Respondent Race: White Black Hispanic White Respondent x Latinx Candidate White Respondent x Black Candidate Black Respondent x Latinx Candidate Black Respondent x Black Candidate Hispanic Respondent x Latinx Candidate Hispanic Respondent x Black Candidate Woman Lawyer Six Month Training Three Week Training Volunteer Constant N Adjusted R2 R2 0.1474 (0.0917) 0.1178 (0.0988) 0.0835 (0.0551) 0.0223 (0.0647) -0.0833 (0.0715) -0.1620. (0.0946) -0.0944 (0.1016) -0.0932 (0.1123) 0.0484 (0.1198) 0.1622 (0.1193) 0.0859 (0.1278) -0.0060 (0.0167) 0.1477*** (0.0175) 0.3251*** (0.0204) 0.1748*** (0.0205) -0.0097 (0.0169) 0.1805*** (0.0540) 5,012 0.09523 0.09811 0.1749. (0.0925) 0.2300* (0.0913) 0.1509** (0.0536) -0.0040 (0.0664) 0.0299 ( 0.0713) -0.2190* (0.0956) -0.2395* (0.0944) -0.0377 (0.1144) 0.0741 (0.1155) 0.0099 (0.1251) -0.0792 (0.1240) 0.0376* (0.0174) -0.0567** (0.0184) 0.2076*** (0.0218) 0.1150*** (0.0214) 0.0533** (0.0172) 0.2432*** (0.0532) 5,016 0.04327 0.04632 0.1183 (0.0945) 0.1132 (0.0945) 0.0742 (0.0547) -0.0716 (0.0671) -0.1068 (0.0698) -0.1351 (0.0975) -0.0859 (0.0976) 0.0248 (0.1156) 0.2202. (0.1154) 0.2080. (0.1218) 0.1300 (0.1242) 0.0346* (0.0173) 0.0195 (0.0181) 0.2236*** (0.0214) 0.1407*** (0.0212) 0.0342* (0.0173) 0.2576*** (0.0541) 5,014 0.04743 0.05047 Standard errors in parentheses. Standard Errors Clustered by Election + p<0.10, * p<0.05, ** p<0.01, *** p<0.001 62 TABLE B.3 CANDIDATE TRAITS INFLUENCE ON CANDIDATE PERCEPTIONS, BY CLASS MATCH Lawyer Respondent Inc. Below Median Lawyer Candidate x Resp. Inc. Below Median Racial Match Woman Six Month Training Three Week Training Volunteer Constant N Adjusted R2 R2 (1) Qualified 0.1597*** (0.0195) 0.0250 (0.0227) -0.0384 (0.0434) 0.0279 (0.0179) -0.0068 (0.0169) 0.3163*** (0.0208) 0.1701*** (0.0204) -0.0057 (0.0170) 0.2548*** (0.0196) 5,006 0.08856 0.09002 (2) Understanding -0.0468* (0.0207) 0.0163 (0.0222) -0.0348 (0.0442) 0.0548** (0.0181) 0.0357* (0.0175) 0.1994*** (0.0219) 0.1130*** (0.0213) 0.0577*** (0.0172) 0.3575*** (0.0205) 5,010 0.03351 0.03505 (3) Has Solution 0.0399. (0.0204) 0.0417. (0.0221) -0.0842. (0.0444) 0.0412* (0.0181) 0.0341* (0.0174) 0.2137*** (0.0218) 0.1372*** (0.0210) 0.0402* (0.0173) 0.3143*** (0.0199) 5,008 0.03455 0.03609 Standard errors in parentheses. Standard Errors Clustered by Election + p<0.10, * p<0.05, ** p<0.01, *** p<0.001 63 TABLE B.4 EXPECTED SATISFACTION, BY RACIAL MATCH AND CLASS MATCH Qualified Understands Has Solutions Candidate Race: Latinx Black Respondent Race: White Black Hispanic Respondent Inc. Below Median Lawyer Candidate x Resp. Inc. Below Median Racial Match White Respondent x Latinx Candidate White Respondent x Black Candidate Black Respondent x Latinx Candidate Black Respondent x Black Candidate Hispanic Respondent x Latinx Candidate Hispanic Respondent x Black Candidate Woman Lawyer Six Month Training Three Week Training Volunteer Constant N Adjusted R2 R2 (1) Expected Satisfaction 0.0386 (0.0434) 0.1635*** (0.0432) 0.1499** (0.0502) (2) Expected Satisfaction 0.0437 (0.0446) 0.1633*** (0.0448) 0.1544** (0.0519) 0.0480 (0.1434) 0.3279* (0.1338) 0.2542* (0.1002) 0.0798 (0.1259) 0.1610 (0.1542) -0.0596 (0.1504) -0.3079* (0.1401) -0.1143 (0.1943) -0.0672 (0.1710) 0.1032 (0.2041) -0.1815 (0.2122) 0.0175 (0.0311) 0.0086 (0.0315) 0.0205 (0.0396) 0.1183** (0.0376) -0.0216 (0.0305) 1.954*** (0.1039) 2,503 0.08120 0.08818 0.0265 (0.0508) -0.0451 (0.0774) 0.0574. (0.0322) 0.0213 (0.0314) 0.0225 (0.0353) 0.0093 (0.0397) 0.1164** (0.0379) -0.0142 (0.0307) 2.155*** (0.0458) 2,500 0.07073 0.07482 Standard errors in parentheses. Standard Errors Clustered by Election + p<0.10, * p<0.05, ** p<0.01, *** p<0.001 64 Survey Experiment Questionnaire In this candidate conjoint study, respondents should evaluate three sets of candidate pairs. Randomize each candidate trait (by items indicated within each row) per candidate per respondent with equal assignment probabilities. Row 1: Row 2: Row 3: Row 4: Row 5: Man, Woman White, Black, Latinx Lawyer, Retail worker None, Participated in a three week training program for people interested in running for office, Participated in a six month training program for people interested in running for office Served on the local school board, Volunteers for local charities and nonprofits Gender Race/ethnicity Job experience Candidate training experience Consider the following two [RESPONDENT’S PARTY] hypothetical candidates for city council. Candidate A Man Black Lawyer Participated in a three week training program for people interested in running for office Local councilmember Candidate B Woman Latinx Retail worker Participated in a six month training program for people interested in running for office Volunteers for local charities Political experience Please answer the prompt for the following items. Which candidate would you guess that… Rows: You would be more likely to vote for? Your neighbors would be more likely to vote for? Better understands the problems facing people like you? Is more qualified for elected office? Is more capable of finding the solution for the problems facing people like you? Columns: 1 Candidate A 2 Candidate B Suppose [Candidate A/Candidate B] has won the election. If you had to guess, how satisfied would you expect be to be local public services? 65 1 2 3 4 5 Extremely satisfied Satisfied Neither satisfied nor dissatisfied Dissatisfied Extremely dissatisfied 66 APPENDIX C: SUPPLEMENTAL MATERIAL FOR CHAPTER 4 TABLE C.1 DESCRIPTIVE STATISTICS AND BALANCE TABLE, CONTROL VS POSITIVE TREATMENT TABLE C.2 DESCRIPTIVE STATISTICS AND BALANCE TABLE, CONTROL VS NEGATIVE TREATMENT Control mean 0.49 8.56 3.55 4.71 3.91 0.75 0.12 0.01 0.06 0.05 0.25 3.83 n 491 480 488 491 491 484 484 484 484 484 484 491 sd 0.50 6.65 1.47 3.31 1.72 0.43 0.33 0.11 0.25 0.22 0.43 0.84 Male Household Income Education Republican Conservative White Black Native American AAPI Other Nonwhite Pre Treatment Expected Satisfaction Control mean 0.49 8.56 3.55 4.71 3.91 0.75 0.12 0.01 0.06 0.05 0.25 3.83 n 491 480 488 491 491 484 484 484 484 484 484 491 sd 0.50 6.65 1.47 3.31 1.72 0.43 0.33 0.11 0.25 0.22 0.43 0.84 Male Household Income Education Republican Conservative White Black Native American AAPI Other Nonwhite Pre Treatment Expected Satisfaction Positive Treatment n mean 0.48 8.66 490 475 489 490 489 480 480 480 480 480 480 490 Negative Treatment n mean 0.50 8.88 3.66 4.49 3.89 0.73 0.12 0.02 0.07 0.06 0.27 3.88 3.70 4.67 3.91 0.74 0.13 0.01 0.07 0.06 0.26 3.84 sd 0.50 6.49 1.51 3.11 1.70 0.45 0.32 0.14 0.26 0.24 0.45 0.80 Diff -0.015 0.101 0.113 -0.213 -0.025 -0.027 -0.001 0.006 0.011 0.011 0.027 0.053 sd 0.50 6.93 1.51 3.24 1.68 0.44 0.34 0.09 0.25 0.23 0.44 0.84 Diff 0.007 0.320 0.150 -0.039 -0.005 -0.016 0.010 -0.004 0.002 0.008 0.016 0.008 496 478 496 496 496 485 485 485 485 485 485 496 67 TABLE C.3 DESCRIPTIVE STATISTICS AND BALANCE TABLE, POSITIVE VS NEGATIVE TREATMENT Positive Treatment n mean 0.48 8.66 Negative Treatment n mean 0.50 8.88 Male Household Income Education Republican Conservative White Black Native American AAPI Other Nonwhite Pre Treatment Expected Satisfaction 490 475 489 490 489 480 480 480 480 480 480 490 sd 0.50 6.49 1.51 3.11 1.70 0.45 0.32 0.14 0.26 0.24 0.45 0.80 3.66 4.49 3.89 0.73 0.12 0.02 0.07 0.06 0.27 3.88 496 478 496 496 496 485 485 485 485 485 485 496 3.70 4.67 3.91 0.74 0.13 0.01 0.07 0.06 0.26 3.84 sd 0.50 6.93 1.51 3.24 1.68 0.44 0.34 0.09 0.25 0.23 0.44 0.84 Diff 0.022 0.220 0.037 0.173 0.020 0.011 0.011 -0.011 -0.009 -0.003 -0.011 -0.045 TABLE C.4 DESCRIPTIVE STATISTICS AND BALANCE TABLE, POST-TREATMENT VARIABLES Post Treatment Satisfaction Post Sat - Expected Sat Political Eng.- Monetary Support Public Voice - Express Discontent Positive Treatment n mean 4.08 sd 0.78 487 Negative Treatment n mean 2.71 sd 1.45 Diff -1.364*** 496 487 0.19 0.82 496 -1.12 1.35 -1.314*** 490 0.27 0.44 496 0.21 0.41 -0.020 490 0.06 0.23 496 0.23 0.42 0.072*** 68 TABLE C.5 TREATMENT AND SATISFACTION EFFECTS ON CHOOSING NOT TO ENGAGE (1) Treatment (2) Experienced Satisfaction (3) Experienced Satisfaction (4) Change in Satisfaction Control Group ref. ref. ref. ref. 0.015 (0.031) -0.031 (0.031) 0.007+ (0.004) -0.052+ (0.030) -0.004+ (0.002) -0.015 (0.009) Positive Treatment Negative Treatment Republican Nonwhite Household Income Education Experienced Satisfaction Negative x Experienced Satisfaction Change in Sat. Negative x Change in Sat. Expected Satisfaction -0.457* (0.125) 0.007 (0.005) -0.024 (0.036) -0.005* (0.003) -0.018 (0.011) -0.127 (0.028) 0.089** (0.032) -0.313* (0.125) 0.006 (0.005) -0.024 (0.035) -0.004+ (0.002) -0.021* (0.011) -0.071* (0.029) 0.062* (0.031) -0.125*** (0.020) -0.029 (0.036) 0.008+ (0.005) -0.027 (0.036) -0.005* (0.003) -0.016 (0.011) 0.030 (0.027) -0.024 (0.032) 0.423*** (0.053) 934.000 0.019 _cons 1.222*** (0.130) 934.000 0.075 Note:Standard errors in parentheses. + p<0.10, * p<0.05, ** p<0.01, *** p<0.001 0.957*** (0.125) 934.000 0.038 0.415*** (0.045) 1410.000 0.010 N Adj. r2 69 TABLE C.6 TREATMENT AND SATISFACTION EFFECTS ON POSITIVE POLITICAL ENGAGEMENT (DONATION) (1) Treatment (2) Experienced Satisfaction (3) Experienced Satisfaction (4) Change in Satisfaction ref. ref. ref. Control Group ref. 0.017 (0.028) -0.046 (0.028) -0.009* (0.004) 0.023 (0.027) 0.003+ (0.002) 0.003 (0.008) Positive Treatment Negative Treatment Republican Nonwhite Household Income Education Experienced Satisfaction Negative x Experienced Satisfaction Change in Sat. Negative x Change in Sat. Expected Satisfaction 0.245* (0.113) -0.011* (0.004) 0.029 (0.032) 0.002 (0.002) 0.004 (0.010) 0.211+ (0.115) -0.011* (0.004) 0.029 (0.032) 0.002 (0.002) 0.004 (0.010) 0.110*** (0.025) 0.097*** (0.026) -0.059* (0.029) -0.052+ (0.029) 0.029 (0.019) -0.017 (0.033) -0.012** (0.004) 0.030 (0.033) 0.002 (0.002) 0.001 (0.010) 0.034 (0.024) 0.002 (0.028) 0.287*** (0.048) 934.000 0.026 _cons -0.228+ (0.120) 934.000 0.047 Note: Standard errors in parentheses + p<0.10, * p<0.05, ** p<0.01, *** p<0.001 0.248*** (0.041) 1410.000 0.012 -0.228+ (0.120) 934.000 0.055 N Adj. r2 70 TABLE C.7 TREATMENT AND SATISFACTION EFFECTS ON CRITICAL POLITICAL ENGAGEMENT (EXPRESS DISCONTENT) (1) Treatment (2) Experienced Satisfaction (3) Experienced Satisfaction (4) Change in Satisfaction Control Group ref. ref. ref. ref. 0.236** (0.091) 0.100*** (0.025) 0.274** (0.089) -0.001 (0.003) -0.008 (0.025) 0.003+ (0.002) 0.022** (0.008) -0.011 (0.020) -0.044+ (0.023) -0.031 (0.021) 0.140*** (0.021) -0.000 (0.003) -0.005 (0.020) 0.002+ (0.001) 0.017** (0.006) Positive Treatment Negative Treatment Republican Nonwhite Household Income Education Experienced Satisfaction Negative x Experienced Satisfaction Change in Sat. Negative x Change in Sat. Expected Satisfaction -0.001 (0.003) -0.008 (0.025) 0.003+ (0.002) 0.023** (0.008) -0.025 (0.021) -0.037 (0.023) 0.033* (0.015) -0.001 (0.003) -0.014 (0.025) 0.003 (0.002) 0.024** (0.008) -0.039* (0.019) -0.015 (0.022) -0.041 (0.037) 934.000 0.111 _cons -0.070 (0.095) 934.000 0.114 Note: Standard errors in parentheses. + p<0.10, * p<0.05, ** p<0.01, *** p<0.001 0.006 (0.030) 1410.000 0.0640 -0.001 (0.089) 934.000 0.103 N Adj.r2 71 Survey Experiment Questionnaire Prompt Suppose that your community has just released plans to renovate the city's parks and recreation facilities. Please consider what you know about your community in your responses to the following questions. How important is it to you that the following elements or outcomes occur for the project? Extremely Important Extremely Important Extremely Important Extremely Important Extremely Important On Time Completion Safety and Compliance Quality Construction Remaining on Budget Include Community Input How likely do you think is it that the following elements or outcomes for the project will occur? Extremely Likely Extremely Likely Extremely Likely Extremely Likely Extremely Likely On Time Completion Safety and Compliance Quality Construction Remaining on Budget Include Community Input 72 How satisfied do you think you will be with this project? • Extremely Satisfied • Satisfied Somewhat Satisfied • Dissatisfied • Extremely Dissatisfied Positive Treatment Your city just reopened the parks and recreation facilities after completing renovation within its proposed 6-month timeline and just within budget. The city was able to include your community’s requested pool, though not to the Olympic standard size requested. The parks have well-manicured lawns, ADA compliant pathways and new rubber tiled surface to accompany the brand-new play structure. Negative Treatment Your city has halted renovations of the parks and recreation facilities after a 6- month extension to the originally proposed timeline. The project has run out of funding and exceeded the budget. The pool your community requested will not be built, the play structure failed to meet basic safety requirements, and the grounds have failed to meet ADA compliance. Post Treatment How satisfied are you with this project? • Extremely Satisfied • Satisfied • Somewhat Satisfied • Dissatisfied • Extremely Dissatisfied 73 How accurate is it that the following elements or outcomes for the project occurred? Extremely Accurate Extremely Accurate Extremely Accurate Extremely Accurate Extremely Accurate On Time Completion Safety and Compliance Quality Construction Remaining on Budget Include Community Input Outcome Based on the described city project, how would you like to participate? • Engage in discourse about the renovationsGive money to support local parks and recreation • None of the above Work or volunteer with the Park and Recreation Board • Publicly express discontent 74