Can Knowledge Correct for Partisan Bias in Political Perceptions
Introduction
The local political arena represents something of a paradox: although residents interact with and are directly affected by actions of local government to a much greater extent than either state or national government, interest and engagement in local political affairs is fairly limited (Caren 2007; Hajnal and Lewis 2003; Holbrook and Weinschenk 2014a). One important related issue that is beginning to attract serious attention is the extent to which local electorates are positioned to hold local political officials accountable for local conditions (Arnold and Carnes 2012; Berry and Howell 2007; Burnett and Kogan 2017; Holbrook and Weinschenk 2014b; Hopkins and Pettingill 2018; Kaufmann 2004; Lay and Tyburski 2017; Oliver and Ha 2007). Of particular importance is the issue of what factors structure citizens' assessments of local conditions. Although the question of how people perceive social, political, and economic conditions has been explored by political scientists, the bulk of existing research has focused on state and national-level conditions (Duch, Palmer, and Anderson 2000; Evans and Andersen 2006; Funk and Garcia-Monet 1997; Hetherington 1996; Hopkins 2011; McDonald and Tolbert 2012; Niemi, Bremer, and Heel 1999), leaving us with little understanding of how people react to and assess economic and other conditions at the local level.
Given the growing body of research showing that local electorates integrate assessments about local conditions, such as the state of the local economy, taxes, schools, and crime, into their local electoral decisions (Arnold and Carnes 2012; Berry and Howell 2007; Kaufmann 2004; Oliver, Ha, and Callen 2012), it is important to evaluate the accuracy of peoples' perceptions of local conditions, even if only in broad strokes. If the assessments that people form and use to evaluate local elected officials are disconnected from reality, but instead reflect relative ignorance or partisan bias, there may be cause for concern about the quality of democratic accountability at the local level. Indeed, while voting on the basis of political, social, or economic conditions does not require that people have precise information about those conditions, it seems sensible to suggest that if people are going to use such conditions as a mechanism for political accountability, they should have broadly accurate perceptions about what is going on.
In this paper, we study perceptions of three different local conditions—the state of the local economy, the quality of local public schools,1 and levels of local crime—in cities across the United States. It is important to note that there is a tremendous amount of variation in the conditions that people experience in cities across the United States: some people live in places with a thriving local economy, while others live in places that consistently struggle; some live in communities where good schools abound and others live in places where the local schools get low marks; some live in places where they rarely think about their safety, while others live in places with exceptionally high crime rates. Do peoples' perceptions about such conditions connect to the reality of the cities where they live? We examine the extent to which objective indicators guide individual perceptions of local conditions, with a particular emphasis on the role of individual characteristics and predispositions in shaping the connection between objective conditions and perceptions. Thus, to understand individual-level differences in local perceptions, we make use of two different theoretical frameworks, one that emphasizes factors that limit information acquisition and may exacerbate political inequalities, and another that emphasizes political motivations for information processing.
Previous Research and Expectations
The question of whether people make accurate assessments about political, social, and economic conditions has long been of interest to scholars (Duch, Palmer, and Anderson 2000; Holbrook and Garand 1996; Niemi, Bremer, and Heel 1999), with a number of scholars finding that the public's perceptions of conditions are not always fully rooted in reality (Achen and Bartels 2016; Bartels 2002; Duch, Palmer, and Anderson 2000; Gramlich 2016; Shao and Goidel 2016). Other studies provide reason to be more optimistic about the capability of the public to connect their assessments of conditions to reality. For example, Niemi, Bremer, and Heel (1999, 188) find that "state economic perceptions are clearly grounded in economic reality, that is, in the actual conditions of the state." Similarly, in his research on public perceptions of income inequality, Franko (2017) finds that state-level perceptions of growing economic inequality, measured by aggregating survey questions, are significantly shaped by objective measures of inequality. Xu and Garand (2010) also study perceptions of income inequality and find that the level of income inequality in one's state has a direct impact on individual-level perceptions about income inequality. In their studies on perceptions of national economic conditions, both Erikson and Wlezien (2012) and Lewis-Beck, Martini, and Kiewiet (2013) find strong relationships between objective economic indicators and aggregated survey measures that capture perceptions of the state of the national economy. While many studies have focused on perceptions of economic conditions, we should note that scholars have also examined whether people accurately assess other conditions related to political life, including perceived levels of political competition and projections of election outcomes. For instance, some scholars (Guinjoan et al. 2014; McDonald and Tolbert 2012) have found that perceived levels of political competition correspond to actual levels of competition in congressional elections, and others have found that, in aggregate, survey respondents do a pretty good job predicting actual election outcomes (Holbrook 2010; Lewis-Beck and Skalaban 1989).
When it comes to local politics, we know virtually nothing about whether residents' perceptions of conditions in their city are grounded in reality. Initially, one might argue that because most people are fairly disconnected from local affairs, they will be ill-informed about conditions in their city. Consequently, perceptions might not match reality. On the other hand, the local context is "closest" to people and, even though most people do not pay a great deal of attention to local affairs, they may have a sense of what is generally happening where they live. Interestingly, Haller and Norpoth (1997, 567) find that although people who pay attention to national news about the economy are better at connecting objective economic conditions to their economic perceptions, even those with little exposure to news about the economy somehow manage "to get a glimpse of economic conditions." Our analysis allows for us to explore this possibility at the local level and also provides an opportunity to expand beyond the potentially low hanging fruit of national economic conditions.
While numerous studies indicate that individual perceptions, especially those related to the economy, are shaped by objective conditions (Lewis-Beck, Martini, and Kiewiet 2013; Niemi, Bremer, and Heel 1999; Xu and Garand 2010), some individual characteristics appear to influence perceptions either directly or by moderating the role of objective conditions. For example, perceptions about national economic conditions seem to be colored by partisanship, though there is debate about how pervasive this effect is (Achen and Bartels 2016; Bartels 2002; Duch, Palmer, and Anderson 2000; Gerber and Huber 2010; Lewis-Beck, Martini, and Kiewiet 2013; Niemi, Bremer, and Heel 1999). Of course, perceptions are necessarily based purely on predispositions or purely on objective information. Perceptions of conditions may be structured by a combination of objective information and predispositions (Franko 2017; Lewis-Beck, Martini, and Kiewiet 2013; Niemi, Bremer, and Heel 1999), and other individual characteristics may interact with information about conditions to structure perceptions. In one of the few studies to examine the interplay between objective conditions and individual-level attributes, Xu and Garand (2010) find that people with lower income levels are more likely than their counterparts to connect objective information on state income inequality to their perceptions about inequality.
Generally speaking, we expect that objective conditions will be related to individual perceptions of those conditions. There are numerous environmental signals, such as local media reports, discussions with friends or neighbors, or personal observations, that provide people with a general sense of how things are going where they live. Importantly, we are also interested in the role of individual predispositions in shaping the connection between objective conditions and perceptions. Indeed, people may respond differently to information about city conditions. To understand individual-level differences in perceptions about city conditions, we make use of two different frameworks, one that emphasizes factors that limit information acquisition and may exacerbate political inequalities, and another that emphasizes partisan motivations for information processing.
To begin, we examine whether the "knowledge gap" hypothesis, which has largely been studied in the context of national politics, holds in the context of perceptions about city conditions. The knowledge gap hypothesis suggests that when information enters a social system, it is likely to exacerbate underlying inequalities in previously held information (Gaziano 1997, 2013; Holbrook 2002; Jerit, Barabas, and Bolsen 2006; Kwak 1999; Moore 1987; Prior 2005; Tichenor, Donohue, and Olien 1970; Viswanath and Finnegan 1996). Specifically, while people from all strata may learn as information becomes available, those with higher levels of preexisting information (typically measured as level of education and/or political knowledge) are likely to learn more than those with low levels of preexisting information, and the gap between the two groups expands. In the context of perceptions about local conditions examined here, the knowledge gap hypothesis suggests that respondents with high levels of education or political knowledge are more likely than their counterparts to be aware of and respond to cues and information about local conditions. Research shows that people who are predisposed to information acquisition are indeed more interested in and informed about local affairs (Oliver 1999). Thus, relying on education and political knowledge as proxies for information acquisition, we expect to see a stronger connection between objective conditions and perceptions of those conditions among those with high levels of education and knowledge, compared with other respondents.
We are also interested in the motivations that people have when they process information. In particular, we want to learn about the role of partisanship in structuring assessments about local conditions. Decades of political science research has illustrated that partisanship influences how people react to, process, and use information about the political world and that it plays a role in shaping assessments of politicians, conditions, and events (Achen and Bartels 2016; Bartels 2000, 2002; Campbell et al. 1960; Citrin and Green 1986; Gerber and Huber 2010; Weinschenk 2013). In general, researchers have shown that people tend to view conditions more favorably when their partisanship matches the partisanship of the person or institution who presides over the conditions. For example, Gerber and Huber (2010) find that immediately following a Democratic takeover of Congress, Democrats became much more positive about the state of the economy and Republicans became much more negative even though the state of the economy did not change much during the change in power, a finding very much in keeping with Bartels' (2002) litany of examples of partisan-based evaluations of political and economic outcomes. At the local level, we know very little about the role of partisanship in structuring how people make assessments about local conditions (Trounstine 2009), although a few studies have indicated that partisanship does play a role in shaping vote choice and mayoral approval (Holbrook 2009; Kaufmann 2004; Oliver and Ha 2007). Building on the large body of research indicating "partisan bias" in assessments of conditions and politicians (Bartels 2002; Duch, Palmer, and Anderson 2000; Jessee 2010; Weinschenk 2012), we expect that respondents who share the mayor's party affiliation hold more positive assessments about city conditions than those who do not share the mayor's partisanship. Relatedly, we expect that this relationship is stronger in cities with partisan elections than in cities with nonpartisan contests. In short, in places where partisan cues are more pervasive, decision-making should be more heavily influenced by party identification. This fits with previous work by Schaffner, Streb, and Wright (2001), who show that when partisan labels are removed, voters tend to place less weight on party when casting their ballots.
It is worth noting that voters may integrate other political leaders into their assessments of local conditions. Indeed, sometimes citizens pin responsibility or blame for conditions on numerous political actors or on a political actor whose impact on conditions is more indirect (Brown 2010; Stein 1990). Given the mixed and potentially much weaker nature of local partisan cues, coupled with the heightened level of partisanship in national politics (Achen and Bartels 2016), we also take into account the potential influence of presidential partisanship in shaping perceptions of local conditions. So, for instance, Democrats might express optimistic views of local conditions not as a means of cheerleading for an incumbent Democratic mayor but instead as a means of signaling support for an incumbent Democratic president (vice versa for Republicans).2 To the extent that presidential partisanship plays a role in assessments of local conditions, we expect that people who share the president's party affiliation at the time of the survey will hold more positive assessments about local conditions than those who do not.
Before turning to our data and measures, it is worth noting that we do not see the two ideas described above as competing theories, but instead as two plausible processes that can happen at the same time. Rather than seeing this as a contest of some sort, we see both theories as worthy of examination in the context of local politics and believe that doing so can enhance our understanding of how people make assessments about local conditions, regardless of which approach finds the strongest empirical support.
Data and Measures
We use two unique datasets for our analyses. The first dataset is the Urban Mayoral Election Study (UMES), a public opinion survey administered prior to forty separate mayoral elections in thirty-nine cities from 2007 to 2011.3 The overall sample size is 6,365 respondents, with an average of 159 respondents from each city. The survey was administered via telephone interviews utilizing separate random-digit-dialing samples from each city and included approximately 90 questions.4 One of the advantages of the UMES design is the ability to capture and use considerable variation in the political, social, and economic contexts of cities. Indeed, the survey sample was designed to capture the range of experience in urban political life specifically to enhance the generalizability of the findings.
The second dataset we use is the Knight Foundation's Soul of the Community Study. In 2008, 2009, and 2010, the Knight Foundation conducted surveys in each of twenty-six Knight Foundation cities.5 The surveys were administered by Gallup and a randomly identified adult eighteen years of age or older completed a fifteen-minute telephone interview. In each of the three years, the survey instrument was very similar, which means that we have a very large sample when we pool the three-survey years. In each year, there are at least 400 interviews in each of the twenty-six cities. Thus, across the 3 years we have between 1,200 and 3,670 respondents per city. In total, there are over 47,000 respondents when we combine the 2008, 2009, and 2010 surveys. Data are weighted in each community to reflect the U.S. adult population by age, gender, race and ethnicity. The data were originally collected to understand the causes and consequences of community attachment, but the surveys also included questions about how people perceive conditions in the city where they live.
Measuring Perceptions of City Conditions
To measure perceptions of local conditions, we make use of three different questions that were included in the UMES. Capturing perceptions of the local economy, we use the following question: "Would you say that over the past year economic conditions in your city have gotten better, stayed about the same, or gotten worse?" To measure perceptions of crime, respondents were asked "Overall, how would you describe the problem of crime in your city? Is it very serious, somewhat serious, not very serious, or not serious at all?" Finally, to gauge perceptions about local public schools, we make use of an item that asks "Overall, how would you rate the quality of education students receive in kindergarten through grade twelve in your city's public schools? Would you rate it as very good, somewhat good, somewhat poor, or very poor?"
The Soul of the Community Study also included questions about city conditions. To assess perceptions about the economy, we use a question that asked "On the same rating scale, where 5 means very good and 1 means very bad, how would you rate economic conditions in (local geography) today?" We measure perceptions of crime by using a question that asks "On a five-point rating scale, where 5 means extremely low and 1 means extremely high, how would you rate the level of crime in your community?" Lastly, to measure perceptions about public schools, we make use of the following questions: "On a five-point rating scale, where 5 means very good and 1 means very bad, how would you rate the following in (local geography)? The overall quality of public schools in your community."
Objective Measures of City Conditions
Since we are interested in the link between perception and reality, we also need measures of objective conditions for each of the dimensions described above. As objective measures of the state of a city's economy, we make use of two indicators: the unemployment rate in the city in the month preceding the survey and the change in the unemployment rate over the last year.6 One important reason to use change in unemployment alongside the actual unemployment question is that the UMES survey item for the local economy is the only question in our data set with an explicitly dynamic emphasis ("Would you say that over the past year . . . "). We gathered data on city unemployment rates from the Bureau of Labor Statistics website on Local Area Unemployment Statistics. As an objective measure of crime, for each city we calculated the crime rate using data from the FBI's Uniform Crime Reports. Thus, we summed the number of violent and property crimes for each city and divided the total number of crimes by the city's population. Due to substantially skewed distributions, we use the logged (base 10) values for crime rate and unemployment rate.7 To assess how local public schools are doing, we use the averaged freshman graduation rate, which is calculated by dividing the number of graduates with regular diplomas by the size of the incoming freshman class four years earlier and expressed as a percentage. This provides a measure of the percent of students who graduate on time. The measure is available for U.S. schools from the Local Education Agency (School District) Universe Survey Dropout and Completion Data from the National Center for Education Statistics. There is considerable variation in the averaged freshman graduation rate across cities. In some places, the rate is in the low 40s (e.g., Gary, IN; Detroit, MI) and in other places the rate is as high as 98 percent (e.g., State College, PA).8
Before moving on to our analyses, we need to be clear about what it means to have perceptions reflect actual conditions accurately. We have no expectation that respondents know exactly what the local unemployment rate, crime rate, or graduation rate is, only that when unemployment and crime rates are relatively high (low), or the high school graduation rate is low (high), they report relatively negative (positive) perceptions of conditions in those domains. In other words, we are focusing on whether respondents generally understand when times are good and when times are bad. Importantly, this means that, on average, respondents in cities where local conditions are good are expected to report relatively positive evaluations, compared with respondents in cities in which conditions are not as good.
Perceptions of Local Conditions
We turn now to the individual-level models, where we incorporate objective measures as contextual variables. Our primary interest in these models is in evaluating the extent to which public evaluations are responsive to objective conditions and the extent to which differences in those evaluations are shaped by information inequalities and partisan cheerleading. Before proceeding, we should also note that there are some differences between the two sets of data that limit the extent to which we can provide completely comparable analyses. Specifically, since the Knight Soul of the Community surveys were not constructed as "political" surveys, they do not include many of the standard political questions that could be used to evaluate political (i.e., partisan) bases to evaluations of local conditions, or to evaluate the extent to which the impact of local conditions on perceptions is conditioned preexisting levels of political knowledge, which is likely to be strongly related to information acquisition (Zaller 1991). Given these differences, we proceed by examining the conditioning influence of level of education in both the Knight Foundation and the UMES data, but explore the partisan bias hypothesis only using the UMES data. In both datasets, we measure education as a four-point variable ranging from high school degree or less, to some college, to college degree, to postgraduate degree. In the analyses using political knowledge (UMES data only), we use a four-point knowledge variable, ranging from 0 (no correct answers) to 3 (three correct answers).9
Given the ordered nature of the dependent variable, along with the structure of the data set (individuals sampled within cities), a mixed-effects model (random intercepts for the cities) with a cumulative probability (ordered logit) link is used to estimate the model parameters.10 Tables 1 and 2 present the initial individual-level models, which, in addition to our key variables of interest, include some basic demographic controls for respondent sex (1 = female, 0 = male), income (dummy variables indicating high, middle, and low income levels, with "don't know/refused" as the omitted group),11 and race (dummy variables for white and black respondents, with all other racial groups as the omitted category), dummy variables capturing the year of the survey,12 and a measure of one's general level of satisfaction toward their city. The control for general satisfaction helps guard against the possibility that perceptions of specific conditions could be influenced by positive or negative developments in other domains, or reflect respondents' generalized support or opposition to local incumbents not captured by the simple matched partisanship analysis.13 Consequently, the inclusion of this measure should mean that we have tougher, more conservative empirical tests when it comes to the variables of interest.14 The models also include two other city-level measures that may be related both actual and perceived conditions: the proportion of the local population who are white, and the log (base 10) of the population density of each city.
| | Table 1. Summary of Results for the Direct Influence of Aggregate Indicators on Individual-Level Evaluations of Local Conditions, UMES and Knight Foundation Data (Control Variables Not Shown). |
| | Table 2. Impact of Objective Indicators on Perceptions of Local Domain-Specific Outcomes, Conditioned by Level of Education (Knight Foundation Data). |
Before getting to the main results, it is instructive to consider the baseline effects of aggregate indicators on evaluations of local conditions. The results in Table 1 summarize these effects by reporting the slopes, standard errors, and significance levels for all four indicators across the two survey data sets, with the results for the control variables not reported for the sake of tidiness.15 Here, we see that when aggregated conditions are considered as additive, contextual variables, they generally have the anticipated impact individual-level evaluations of local conditions. In six of eight cases, when conditions are favorable, evaluations tend to be positive, and when conditions are not favorable, evaluations tend to be negative. The key questions for the remainder of this analysis are whether the connection between conditions and evaluations is stronger for some groups than for others, and whether evaluations are also driven by partisan considerations.
For the most part, our expectations regarding information inequalities are confirmed in Tables 2 and 3 as the interaction effects are significant and in the anticipated direction in seven of the eight models. Across both data sets, there is a significant information interaction for the impact of unemployment and change in unemployment on perceptions of the economy, as well as a significant interaction effect for the freshman graduation rate, and a significant interaction effect for the crime rate in the Knight Foundation data in Table 2. It should be noted that a nonsignificant interaction means that information about the specific indicators is broadly accessible and connected to perceptions. In the case of the crime rate in the UMES cities, for instance, there is a significant relationship between local crime rate and individual perceptions of local crime among those with a high school degree or less education, and the relationship is not stronger or weaker for those with higher levels of education. In other words, crime has roughly equal accessibility to all respondents. The results for political expertise are presented in Table 4 where we see many similarities to the results for education: significant interaction effects for the unemployment rate, change in unemployment, and the freshman graduation rate, but no significant interaction effect for the crime rate. So, across two separate data sets using similar but not identical questions from survey data gathered from largely different sets of cities across partially overlapping time periods, we get very similar results that point to important knowledge gaps in perceptions of local conditions in U.S. cities.
| | Table 3. Impact of Objective Indicators on Perceptions of Local Domain-Specific Outcomes, Conditioned by Level of Education (UMES Data). |
| | Table 4. Impact of Objective Indicators on Perceptions of Local Domain-Specific Outcomes, Conditioned by Political Knowledge (UMES Data). |
To provide a clearer understanding of these effects, some of the relationships are plotted in Figure 1 . Here, we see visualizations of the relationships between objective indicators and perceptions, conditioned by education, using results from the UMES data from Table 3.16 The strongest evidence of interaction effects is found in the results for perceptions of the state of the local economy. In the graph using the unemployment rate as the measure of economic conditions (upper left), respondents with a high school degree or less education are not very responsive to the objective indicator whereas respondents in all three of the other education categories react increasingly strongly and connect the unemployment rate to their perceptions of the state of the local economy. The results for change in the local unemployment rate (upper right) show even greater differentiation according to level of education: respondents with no more than a high school degree are unmoved by objective conditions, those with some college experience respond positively to changes in the unemployment rate, and those with four years or more college experience react very strongly to changes in the unemployment rate. The patterns for both unemployment and change in unemployment fit the expectations of the knowledge gap hypothesis quite nicely. The findings for perceptions of local education (lower right) are similar in nature: there is a negative relationship between the freshman graduation rate and perceptions of the quality of local public schools that is somewhat weak among those with the lowest level of education and grows increasingly strong as education increases.17 The results for perceptions of local crime present completely null findings for the conditional information effects. While the four groups differ in an additive sense, the differences in their slopes are not large enough to constitute real differences. This is not to say that perceptions of crime are not grounded in reality, just that all respondents react similarly to the local crime rate, regardless of level of information. One possible explanation for this null finding is that the impact of crime is not typically evenly distributed within cities, usually hitting poor neighborhoods the hardest. Since people living in these neighborhoods are also likely to have lower than average levels of education, their more direct experience with crime might offset the impact of educational differences and lead to null effects for the interaction term. At the same time, it is important to recall that there is a significant interaction for the crime model using the Knight Foundation data (Table 2).
Figure 1. Illustration of the conditioning impact of expertise on the relationship between objective indicators and perceptions of local conditions with education as a conditioning influence (Urban mayoral election study data).
All probability estimates were calculated based on the results from Table 3 using the R "effects" package, Version 4.1–2.
One other pattern we note in Figure 1 is the differential effects of education at varying levels of local conditions. When times are bad—high or increasing unemployment, high levels of crime, and poorly-performing schools—perceptions of those conditions depend quite a bit on level of education, but when times are good, education level has virtually no effect on perceptions of local conditions—all respondents, regardless of education, get it. One implication of this, from a reward-punishment perspective, is that those with high levels of information are best positioned to hold local officials (mayors, council members, school board members) accountable for negative outcomes, but are no more likely than others to reward for positive outcomes. On one hand, since those with high levels of information are among the most likely to turn out to vote (Leighley and Nagler 2013), the prospects for accountability may benefit from this disparity. However, this also highlights the overlap between informational and political inequalities.
The evidence in Tables 2 through 4, along with the visualizations in Figure 1 , drive home a couple of points. First, broadly speaking, perceptions of local conditions are responsive to objective indicators of those conditions. We take this as a positive indication of the potential for local electorates to reward or pushing local elected officials on the basis of local conditions. Second, there are information-related effects that create inequalities in the connection between indicators and perceptions: generally, the connection is stronger for those with higher levels of education, resulting in the already-information-rich being best positioned to react to changes in local conditions. Finally, information inequalities tend to be least apparent for perceptions of crime. As we pointed out above, this may be related to how the effects of crime are distributed within cities, but another possibility is that the effects are weak here because crime is a policy area that very clearly falls under the domain of local politics. In addition to null findings for the crime interaction, the impact of the education interaction terms, while statistically significant, is substantively weaker than the interaction effects in the unemployment models. We take this as evidence that information regarding conditions related to typically "local" policy areas is generally more broadly accessible than information in policy areas with a more clearly national or state connection. We see this pattern as an indication that the accountability playing field is leveled a bit more when it comes to traditionally "local" policy domains, though it is important to acknowledge that we did find some information effects in these areas. This "local issues" connection is also bolstered by the model-fit statistics, which show that the collection of individual and city-level variables we use here consistently do a better job of explaining variation in evaluations of schools and crime than in evaluations of the local economy.
Incorporating Political Influences
We explore potential partisan biases in two different ways. First, we consider whether respondents who share the party affiliation of the incumbent mayor view conditions more positively than their counterparts, controlling for whether respondents live in a partisan or nonpartisan city.18 The expectations for party identification are clear: respondents should be more likely to view local conditions positively if they share the mayor's party affiliation than if they do not. The extent to which there is shared partisan identity with the mayor is determined by using a three-category party identification measure ("Generally speaking, do you think of yourself as a [ROTATE: Republican, a Democrat] an Independent, or what?") alongside data on mayoral party affiliation.19 Same-party pairs are coded 1 on a party "match" dummy variable, with the rest coded 0; and opposite-party pairs are coded 1 on a party "mismatch" variable, and the rest coded 0. Of course, this leaves independents as the reference category. To account for partisan influence from national politics, we interact respondent party identification with a dichotomous variable for presidential administration (0 = Bush, 1 = Obama) at the time of the survey. The expectation is that presidential co-partisans are motivated by ingroup bias and have perceptions of local conditions that are appreciably more positive than those reported by out-party partisans, who are also motivated by ingroup bias. Although there is abundant evidence of partisan bias in evaluations of national politics based on presidential partisanship (Bartels 2002; Gerber and Huber 2010; Weinschenk 2013), we know very little about how presidential partisanship shapes local politics.
The results of the partisan bias analysis are found in Tables 5 and 6. The first thing to note is that adding the partisan controls leaves the other coefficients relatively unperturbed from their original values—the knowledge gap findings remain intact in both tables—but does produce some interesting, if somewhat mixed, effects regarding the role of party. Curiously, there is very little evidence that a partisan match with the incumbent mayor introduces any bias into evaluations of specific local conditions, at least in the anticipated manner.20 In both of the local economy models in Table 5, there is a significant negative coefficient for out-party status, but only in nonpartisan cities. However, while those slopes are significantly different from zero, they are not statistically different from the negative slopes for in-party status. So, at best, it looks like in nonpartisan cities there may be a slight tendency for partisans to be somewhat less pessimistic about the local economy. None of the local partisan match coefficients are statistically significant in Table 6, where political knowledge is used as the conditioning information-based variable.
| | Table 5. Impact of Objective Indicators on Perceptions of Local Domain-Specific Outcomes, Conditioned by Level of Education and Controlling for Partisan Bias (UMES Data). |
| | Table 6. Impact of Objective Indicators on Perceptions of Local Domain-Specific Outcomes, Conditioned by Level of Political Knowledge and Controlling for Partisan Bias (UMES Data). |
On the other hand, in both tables, there are strong effects from shared presidential partisanship on perceptions of the local economy but no effects on evaluations of local crime and local schools. The impact of presidential partisanship on perceptions of the local economy is summarized in Figure 2 , where the slope for party identification is plotted separately for respondents interviewed during the (Democratic) Obama and (Republican) Bush presidencies. What we see here is virtually opposite effects from party identification, depending on the party of the president. Under President Bush, the probability of strong Republicans rating their local economy as "worse" than the year before was about .42, compared with .56 for strong Democrats; while under President Obama, the probability of strong Democrats rating the local economy as "worse" was .34, compared with .50 for strong Republicans. These effects are all the more impressive given that the model controls for actual economic indicators as well as general levels of satisfaction with local conditions. It is interesting that presidential partisanship exerts this influence on perceptions of the local economy and not on perceptions of local crime or local schools. In retrospect, we think this makes some sense, given the prominent role the economy plays in national politics, influencing both presidential approval (MacKuen, Erikson, and Stimson 1992; Norpoth 1996) and presidential election outcomes (Alvarez and Nagler 1995, 1998; Nadeau and Lewis-Beck 2001). In addition, this pattern fits nicely with the earlier discussion of somewhat weaker information effects in these areas because crime and public schools are more easily seen as local issue areas. Also, in some sense, respondents probably treat questions about the state of the economy as an extension of their evaluation of their general sense of well-being, which may be buoyed when their own party is in control. By contrast, crime rates and educational outcomes are more likely to be seen as the outcomes of public services that are provided by local governments. Overall, our finding regarding the influence of national politics on evaluations at the local level is suggestive of an important set of considerations in the study of local politics. In short, future studies on local politics should take seriously the possible influence of national-level politics, such as partisanship or whether certain local issues are more likely to be connected to national politics than others.
Figure 2. Impact of party identification on evaluations of the local economy, conditioned by presidential party.
All probability estimates were calculated based on the results in Table 5 using the R "effects" package, Version 4.1–2.
Conclusion and Future Research
In this paper, we examined the extent to which peoples' perceptions of local conditions are connected to reality. This is an important question given the growing body of research showing that local electorates make use of a variety of local conditions when evaluating local political officials. If people have inaccurate perceptions about local conditions, and then use those assessments when voting in local elections, there may be cause for concern about the quality of democratic accountability at the local level.
Overall, we found that objective conditions influence individuals' perceptions of those conditions, but the perceptions of those with access to more information were the most responsive to objective conditions, while those with the lowest levels of information were the least responsive. This finding is important as it indicates that the "knowledge gap" hypothesis holds in the context local politics. Among political scientists, the knowledge gap hypothesis has largely been examined in the context of national politics, and this research represents an important contribution by extending these ideas to the local political arena. In addition to examining the role of education and knowledge, we examined the role of partisanship in shaping local perceptions. Although local politics is often discussed as a place where partisanship and partisan divides are less pronounced, we found that partisanship does play a role in shaping perceptions of local conditions. Interestingly, while the partisan match between a respondent and the mayor of their city had little effect on local perceptions, we found that presidential partisanship makes its way into the local political arena—the match between a respondent's partisanship and the president's party has a strong effect on perceptions of the local economy.
One particularly impressive aspect of these findings stems from their consistency. As described earlier, although the two survey data sets do not use the exact same question wordings for most items used in these models (though both datasets are appropriate to the task), and the survey data were gathered from largely different sets of cities across partially overlapping time periods, we get very similar results that point to important knowledge gaps in perceptions of local conditions in U.S. cities. Whether using the Knight Foundation or UMES data, relying on education or political knowledge as the conditioning variable, controlling for political influences or not, the same general picture emerges. Popular evaluations of local conditions generally reflect objective indicators of those conditions, though the connection between conditions and evaluations is moderated by propensity to gather information, and some evaluations are clouded by (presidential) partisan bias.
Still, there is a lot of room to expand research in this area. We examined three local conditions in this paper, but there are certainly other conditions that could be studied at the local level (e.g., taxes, infrastructure, the environment, corruption, etc.). Thus, the framework used here could be extended to other domains. And, of course, although we were able to make use of two fairly large surveys data sets, it would be worth examining our findings in other datasets and contexts. Finally, we encourage additional work on the role of partisanship in local politics. Although local politics is often thought to be less partisan than national politics, our results provide evidence that even factors like presidential partisanship shape local political assessments. We need to learn more about the role of partisanship and national politics in local political affairs.
Declaration of Conflicting Interests Funding ORCID iD Supplemental Material
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported with funding from University of Wisconsin-Milwaukee Research Growth Initiative (Study #101X074) and the National Science Foundation (Study #0921343).
Aaron C. Weinschenk 
https://orcid.org/0000-0002-2278-0349
Data and code necessary to replicate the analyses in this paper can be found here: https://www.acweinschenk.net/infobiasperceptionsreplication.html.
Notes
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.
It is important to include a measure of school performance here even though the connection to mayoral (or perhaps city council) elections is less direct than is the case for the local economy or the crime rate. Beyond such municipal elections, voters across the country, and across the cities we study here, are also asked to make judgments about the performance of elected school board members, and there is some evidence that local electorates are capable of making those judgments (Berry and Howell 2007). In other words, local electoral politics can extend beyond the confines of City Hall.
Interestingly, Holbrook (2009) investigates the role of presidential partisanship in mayoral elections and finds that views of the president have an independent effect on mayoral vote choice, especially in cities with partisan elections.
The cases (Atlanta, GA, 2009; Baltimore, MD, 2007; Boise, ID, 2007; Boston, MA, 2009; Charlotte, NC, 2007, 2009; Cincinnati, OH, 2009; Cleveland, OH, 2009; Columbus, GA, 2010; Columbus, OH, 2007; Dallas, TX, 2011; Denver, CO, 2011; Detroit, MI, 2009; Durham, NC, 2007; Ft. Wayne, IN, 2007; Fresno, CA, 2008; Garden Grove, CA, 2010; Greensboro, NC, 2007; Houston, TX, 2009; Indianapolis, IN, 2007; Jacksonville, FL, 2011; Laredo, TX, 2010; Mesa, AZ, 2008; Miami, FL, 2009; Philadelphia, PA, 2007; Pittsburgh, PA, 2007; Reno, NV, 2010; Riverside, CA, 2009; Sacramento, CA, 2008; Salt Lake City, UT, 2007; Santa Ana, CA, 2010; Seattle, WA, 2009; Shreveport, LA, 2010; Spokane, WA, 2007; St. Petersburg, FL, 2009; Tacoma, WA, 2009; Toledo OH, 2009; and Yonkers, NY, 2007) selected for this study are drawn from among the 125 largest cities in the United States. While other large cities held elections during the same time period, the cities selected for this study were chosen, in part, based on competitiveness and with the purpose of maximizing variance in demographic and candidate diversity.
The target population is the citizen voting-age population. Because there is a slight tendency to overrepresent the non-Hispanic White population, poststratification weights are used to bring the composition of the local samples into line with existing Census estimates of local racial and ethnic composition, based on the adult citizen population.
The 26 cities are as follows: Detroit, MI; Philadelphia, PA; Miami, FL; St. Paul, MN; San Jose, CA; Palm Beach, FL; Charlotte, NC; Bradenton, FL; Akron, OH; Gary, IN; Long Beach, CA; Boulder, CO; Columbia, SC; Wichita, KS; Lexington, KY; Tallahassee, FL; Columbus, GA; Fort Wayne, IN; Duluth, MN; Macon, GA; Biloxi, MS; Grand Forks, ND; Myrtle Beach, SC; State College, PA; Milledgeville, GA; Aberdeen, SD.
We calculated the change measure by using data on the unemployment rate in the month preceding the survey and then data on the unemployment rate twelve months before that. For example, if respondents in a city were surveyed in November of 2007, the change in unemployment would be based on the change between October 2007 (the month immediately before the survey) to October 2006 (one year before that). Thus, if unemployment was 5 percent in October 2006 and 7 percent in October 2007, the change value would be +2 percent.
Similar results are obtained when we use the raw crime rate and unemployment rate, but the mixed-effects models had fewer problems (errors and warnings) when using the logged versions.
More recent evaluations of school performance complement the Average Freshman Graduation Rate (AFGR) with the Average Cohort Graduation Rate (ACGR), which tracks the graduation rates of the new freshman cohort from a given year, excluding those students who transfer out during the next four years and including those who transfer in. Unfortunately, the ACGR data were not gathered until the 2010–2011 school year, so we cannot use it here, since the vast majority of our survey data were gathered from 2007 to 2010. In a few cases, due to missing data, the AFGR from the most proximate year was substituted for the missing information.
The measure of political knowledge (only available in the UMES) is a three-item scale that captures the number of correct responses given to questions about U.S. politics: which party controlled the U.S. House of Representatives and the U.S. Senate at the time the respondent was interviewed, and whether they knew what job or political office Nancy Pelosi (before 2011) or John Boehner (after 2010) held. On this measure of expertise, 22 percent were unable to answer any questions correctly, 16 percent could answer one question correctly, 24 percent answered two questions correctly, and 38 percent answered all three correctly. Although this is a measure of political expertise at the national level, it is positively related to measures of local engagement (how closely respondents said they followed mayoral elections and self-reported familiarity with mayoral candidates).
All models are estimated in R, using the clmm function from the Ordinal package (version 2019.4-25).
The codes for income categories are somewhat different in the two data sets, leading us to use slightly different cut points. The three income categories in the UMES data are less than $31,000, $31,000 to $75,000, and greater than $75,000 per year. In the Knight data, they are less than $44,999, $45,000 to $99,999, and $100,000 or more.
The mixed-effects models did not always tolerate all of the year dummy variables when using the UMES data, so this group of controls varies a bit across models. To be clear, the results for the other variables in the model are unaffected by the combination of year dummy variables.
The wording for this item from the UMES survey is "On the whole, are you very satisfied, somewhat satisfied, somewhat dissatisfied, or very dissatisfied with the way things are going in your city?" and for the Knight Foundation it is "Taking everything into account, how satisfied are you with (local geography) as a place to live? Please use a five-point scale, where 5 means you are extremely satisfied and 1 means you are not at all satisfied. You may use any of the numbers 1, 2, 3, 4, or 5 for your rating."
We should also note that beyond providing conservative tests, the inclusion of this measure may also help account for common-method bias (Podsakoff et al. 2003), whereby respondents answering similarly formatted questions might generally provide positive, negative, or neutral responses, due to the question format rather than to underlying attitudes. We are indebted to an anonymous reviewer for reminding us of this point.
The effects for the control variables are very similar to the effects reported in other tables throughout this paper.
The effects of unemployment and crime rate are presented using the raw versions of these variables.
The negative slope between the average freshman graduation rate (AFGR) and evaluations of local schools makes sense, since high values of AFGR represent a positive outcome (high graduation rates) and high values of the dependent variables represent negative evaluations.
Cities in which partisan labels appeared on the ballot are Baltimore, Charlotte (2007 and 2009), Fort Wayne, Indianapolis, Jacksonville, New York City, Philadelphia, Pittsburgh, Shreveport, and Yonkers. Shreveport uses a "nonpartisan primary" and runoff system, but one in which the candidates run on party labels and the ballots designate the party of the candidate. This is referred to as "nonpartisan" because the candidates from all parties appear on the ballot. In this particular election, the general election featured a Democrat and a Republican.
Although more than two-thirds of respondents live in nonpartisan cities, we were able to make educated guesses about the party affiliation of the mayors, based on prior office holding, prior campaigns, other public political experience, or average respondent party placement of the candidates. When possible, we gleaned partisanship information from local news articles and web searches. In many officially nonpartisan cities, local media coverage of the candidates will mention candidate partisanship directly or indirectly (e.g., the candidate has the backing of the county Republican Party). If we could not find data on partisanship through news articles or searches, we relied on marginal responses from survey questions about perceived candidate partisanship that were included in the UMES to identify how local residents perceived the party affiliation of the mayor. Our approach is certainly in line with other recent studies on local politics, including de Benedictis-Kessner and Warshaw (2016), who "coded candidates' partisanship based on any clear indicators that candidates' leaned toward one of the two parties. These indicators included past or future partisan elected offices that a candidate held, mentions in historical newspaper articles of their partisanship, and campaign-donation-based data."
The null findings are not surprising in the case of evaluations of local schools, since mayors are generally not responsible for local schools and their partisanship should not be relevant to evaluations. However, since mayors might be the focus of "blind retrospection," where voters hold them responsible for conditions generally, regardless of their level of control, it is important to test for partisan effects in the schools model as well.
References
| Achen, Christopher H., Bartels, Larry M. 2016. Democracy for Realists. Princeton: Princeton University Press. Google Scholar | Crossref | |
| Alvarez, R. Michael, Nagler, Jonathon. 1995. "Economics, Issues, and the Perot Candidacy." American Journal of Political Science 39:714–44. Google Scholar | Crossref | |
| Alvarez, R. Michael, Nagler, Jonathan. 1998. "Economics, Entitlements, and Social Issues: Voter Choice in the 1996 Presidential Election." American Journal of Political Science 42:1349–63. Google Scholar | Crossref | |
| Arnold, R. Douglas, Carnes, Nicholas. 2012. "Holding Mayors Accountable: New York's Executives from Koch to Bloomberg." American Journal of Political Science 56 (4): 949–63. Google Scholar | Crossref | |
| Bartels, Larry . 2000. "Partisanship and Voting Behavior, 1952-1996." American Journal of Political Science 44 (1):35–50. Google Scholar | Crossref | ISI | |
| Bartels, Larry . 2002. "Beyond the Running Tally: Partisan Bias in Political Perceptions." Political Behavior 24 (2): 117–50. Google Scholar | Crossref | |
| Berry, Christopher R., Howell, William G. 2007. "Accountability and Local Elections: Rethinking Retrospective Voting." Journal of Politics 69 (3): 844–58. Google Scholar | Crossref | |
| Brown, Adam . 2010. "Are Governors Responsible for the State Economy? Partisanship, Blame, and Divided Government." Journal of Politics 72 (3): 605–15. Google Scholar | Crossref | Medline | |
| Burnett, Craig M., Kogan, Vladimir. 2017. "The Politics of Potholes: Service Quality and Retrospective Voting in Local Elections." Journal of Politics 79 (1): 302–14. Google Scholar | Crossref | Medline | |
| Campbell, Angus A., Converse, Philip E., Miller, Warren E., Stokes, Donald E. 1960. The American Voter. New York: John Wiley. Google Scholar | |
| Caren, Neal L. 2007. "Big City, Big Turnout? Electoral Participation in American Cities." Journal of Urban Affairs 29 (1): 31–46. Google Scholar | Crossref | ISI | |
| Citrin, Jack, Green, Donald. 1986. "Presidential Leadership and the Resurgence of Trust in Government." British Journal of Political Science 16(4): 431–453. Google Scholar | Crossref | ISI | |
| de Benedictis-Kessner, Justin, Warshaw, Christopher. 2016. "Mayoral Partisanship and Municipal Fiscal Policy." Journal of Politics 78 (4): 1124–38. Google Scholar | Crossref | |
| Duch, Raymond, Palmer, Harvey, Anderson, Robert. 2000. "Heterogeneity in Perceptions of National Conditions." American Journal of Political Science 44 (4): 635–52. Google Scholar | Crossref | |
| Erikson, Robert, Wlezien, Christopher. 2012. "The Objective and Subjective Economy and the Presidential Vote." PS: Political Science & Politics 45 (4): 620–24. Google Scholar | Crossref | |
| Evans, Geoffrey, Andersen, Robert. 2006. "The Political Conditioning of Economic Perceptions." Journal of Politics 68 (1): 194–207. Google Scholar | Crossref | ISI | |
| Franko, William . 2017. "Understanding Public Perceptions of Growing Economic Inequality." State Politics & Policy Quarterly 17 (3): 319–48. Google Scholar | SAGE Journals | |
| Funk, Carolyn L., Garcia-Monet, Patricia. 1997. "The Relationship between Personal and National Concerns in Public Perceptions about the Economy." Political Research Quarterly 50 (2): 317–42. Google Scholar | SAGE Journals | |
| Gaziano, Cecilie. 1997. "Forecast 2000: Widening Knowledge Gaps." Journalism & Mass Communication Quarterly 74 (2): 237–64. Google Scholar | SAGE Journals | |
| Gaziano, Cecilie . 2013. "Knowledge Gaps, Belief Gaps, Ideology, and Culture Wars." Open Journal of Political Science 3 (4): 116–30. Google Scholar | Crossref | Medline | |
| Gerber, Alan S., Huber, Gregory A. 2010. "Partisanship, Political Control, and Economic Assessments." American Journal of Political Science 54 (1): 153–73. Google Scholar | Crossref | |
| Gramlich, John . 2016. "Voters' Perceptions of Crime Continue to Conflict with Reality." Pew Research Center. Accessed January 6, 2019. http://www.pewresearch.org/fact-tank/2016/11/16/voters-perceptions-of-crime-continue-to-conflict-with-reality/. Google Scholar | |
| Guinjoan, Marc, Simon, Pablo, Bermudez, Sandra, Lago, Ignacio. 2014. "Expectations in Mass Elections: Back to the Future?" Social Science Quarterly 95 (5): 1346–59. Google Scholar | Crossref | |
| Hajnal, Zoltan, Lewis, Paul. 2003. "Municipal Institutions and Voter Turnout in Local Elections." Urban Affairs Review 38:645–68. Google Scholar | SAGE Journals | |
| Haller, H. Brandon, Norpoth, Helmut. 1997. "Reality Bites: News Exposure and Economic Opinion." Public Opinion Quarterly 61:555–75. Google Scholar | Crossref | |
| Hetherington, Marc . 1996. "The Media's Role in Forming Voters' National Economic Evaluations in 1992." American Journal of Political Science 40:372–95. Google Scholar | Crossref | |
| Holbrook, Thomas M. 2002. "Presidential Campaigns and the Knowledge Gap." Political Communication 19 (4): 437–54. Google Scholar | Crossref | |
| Holbrook, Thomas M. 2009. "Local and National Conditions in Mayoral Elections." Paper presented at Annual Meeting of Midwest Political Science Association, Chicago, IL, April 2–5. Google Scholar | |
| Holbrook, Thomas M. 2010. "Forecasting US Presidential Elections." In The Oxford Handbook of American Elections and Political Behavior, edited by Leighley, Jan E. , 346–71. Oxford: Oxford University Press. Google Scholar | |
| Holbrook, Thomas M., Garand, James. 1996. "Homo Economus? Economic Information and Economic Voting." Political Research Quarterly 49:351–75. Google Scholar | SAGE Journals | |
| Holbrook, Thomas M., Weinschenk, Aaron C. 2014a. "Campaigns, Mobilization, and Turnout in Mayoral Elections." Political Research Quarterly 67 (1): 42–55. Google Scholar | SAGE Journals | ISI | |
| Holbrook, Thomas M., Weinschenk, Aaron C. 2014b. "Money, Candidates, and Mayoral Elections." Electoral Studies 35:292–302. Google Scholar | Crossref | ISI | |
| Hopkins, Daniel J. 2011. "Whose Economy? Perceptions of National Economic Performance during Unequal Growth." Public Opinion Quarterly 76 (1): 50–71. Google Scholar | Crossref | ISI | |
| Hopkins, Daniel J., Pettingill, Lindsay M. 2018. "Retrospective Voting in Big-City US Mayoral Elections." Political Science Research and Methods 6:697–714. Google Scholar | Crossref | |
| Leighley, Jan E., Nagler, Jonathan. 2014. Who Votes Now? Demographics, Issues, Inequality, and Turnout in the United States. Princeton: Princeton University Press. Google Scholar | |
| Jerit, Jennifer, Barabas, Jason, Bolsen, Toby. 2006. "Citizens, Knowledge, and the Information Environment." American Journal of Political Science 50 (2): 266–82. Google Scholar | Crossref | |
| Jessee, Stephen . 2010. "Partisan Bias, Political Information, and Spatial Voting in the 2010 Presidential Election." Journal of Politics 72 (2): 327–40. Google Scholar | Crossref | Medline | |
| Kaufmann, Karen . 2004. The Urban Voter: Group Conflict and Mayoral Voting Behavior in American Cities. Ann Arbor: University of Michigan Press. Google Scholar | Crossref | |
| Kwak, Nojin . 1999. "Revisiting the Knowledge Gap Hypothesis: Education, Motivation, and Media Use." Communication Research 26 (4): 385–413. Google Scholar | SAGE Journals | ISI | |
| Lay, J. Celeste, Tyburski, Michael. 2017. "The Buck Stops with the Education Mayor: Mayoral Control and Local Test Scores in Mayoral Elections." Politics & Policy 45 (6): 964–1002. Google Scholar | Crossref | |
| Lewis-Beck, Michael S., Martini, Nicholas F., Kiewiet, D. Roderick. 2013. "The Nature of Economic Perceptions in Mass Publics." Electoral Studies 32 (3): 524–28. Google Scholar | Crossref | |
| Lewis-Beck, Michael S., Skalaban, Andrew. 1989. "Citizen Forecasting: Can Voters See into the Future?" British Journal of Political Science 19 (1): 146–53. Google Scholar | Crossref | |
| MacKuen, Michael B., Erikson, Robert S., Stimson, James A. 1992. "Peasants or Bankers? The American Electorate and the U.S. Economy." American Political Science Review 86:597–611. Google Scholar | Crossref | ISI | |
| McDonald, Michael P., Tolbert, Caroline J. 2012. "Perceptions vs. Actual Exposure to Electoral Competition and Effects on Political Participation." Public Opinion Quarterly 76 (3): 538–54. Google Scholar | Crossref | |
| Moore, David W. 1987. "Political Campaigns and the Knowledge-Gap Hypothesis." Public Opinion Quarterly 51 (2): 186–200. Google Scholar | Crossref | ISI | |
| Nadeau, Richard, Lewis-Beck, Michael. 2001. "National Economic Voting in U.S. Presidential Elections." Journal of Politics 63:159–81. Google Scholar | Crossref | |
| Niemi, Richard, Bremer, John, Heel, Michael. 1999. "Determinants of State Economic Perceptions." Political Behavior 21 (2): 175–93. Google Scholar | Crossref | |
| Norpoth, Helmut . 1996. "Presidents and the Prospective Voter." Journal of Politics 58:776–92. Google Scholar | |
| Oliver, J. Eric . 1999. "The Effects of Metropolitan Segregation on Local Civic Participation." American Journal of Political Science 43 (1): 186–212. Google Scholar | Crossref | ISI | |
| Oliver, J. Eric, Ha, Shang E. 2007. "Vote Choice in Suburban Elections." American Political Science Review 101 (3): 393–408. Google Scholar | Crossref | ISI | |
| Oliver, J. Eric, Ha, Shang E., Callen, Zachary. 2012. Local Elections and the Politics of Small-Scale Democracy. Princeton: Princeton University Press. Google Scholar | Crossref | |
| Podsakoff, Philip, MacKenzie, Scott, Lee, Jeong-Yeon, Podsakoff, Nathan. 2003. "Common Method Biases in Behavioral Research: A Critical Review of the Literature and Recommended Remedies." Journal of Applied Psychology 88 (5): 879–903. Google Scholar | Crossref | Medline | ISI | |
| Prior, Markus . 2005. "News vs. Entertainment: How Increasing Media Choice Widens Gaps in Political Knowledge and Turnout." American Journal of Political Science 49 (3): 577–92. Google Scholar | Crossref | |
| Schaffner, Brian, Streb, Matthew, Wright, Gerald. 2001. "Teams without Uniforms: The Nonpartisan Ballot in State and Local Elections." Political Research Quarterly 54 (1): 7–30. Google Scholar | SAGE Journals | ISI | |
| Shao, Wanyun, Goidel, Kirby. 2016. "Seeing Is Believing? An Examination of Perceptions of Local Weather Conditions and Climate Change among Residents in the U.S. Gulf Coast." Risk Analysis 36 (11): 2136–57. Google Scholar | Crossref | Medline | |
| Stein, Robert M. 1990. "Economic Voting for Governor and U.S. Senator: The Electoral Consequences of Federalism." Journal of Politics 52:29–53. Google Scholar | Crossref | ISI | |
| Tichenor, P. J., Donohue, G. A., Olien, C. N. 1970. "Mass Media Flow and Differential Growth in Knowledge." Public Opinion Quarterly 34 (2): 159–70. Google Scholar | Crossref | |
| Trounstine, Jessica . 2009. "All Politics Is Local: The Reemergence of the Study of City Politics." Perspectives on Politics 7 (3): 611–18. Google Scholar | Crossref | |
| Viswanath, K., Finnegan, John R. 1996. "The Knowledge Gap Hypothesis: Twenty-Five Years Later." Annals of the International Communication Association 19 (1): 187–228. Google Scholar | Crossref | |
| Weinschenk, Aaron C. 2012. "Partisan Pocketbooks: The Politics of Personal Financial Evaluations." Social Science Quarterly 93 (4): 968–87. Google Scholar | Crossref | |
| Weinschenk, Aaron C. 2013. "Partisanship and Voting Behavior: An Update." Presidential Studies Quarterly 43 (3): 607–17. Google Scholar | Crossref | |
| Xu, Ping, Garand, James. 2010. "Economic Context and Americans' Perceptions of Income Inequality." Social Science Quarterly 91 (5): 1220–41. Google Scholar | Crossref | |
| Zaller, John . 1991. "Information, Values, and Opinion." American Political Science Review 85 (4): 1215–37. Google Scholar | Crossref |
Can Knowledge Correct for Partisan Bias in Political Perceptions
Source: https://journals.sagepub.com/doi/10.1177/1065912919892627