Is It Us? Is It Them? Or Is It This Place? Predicting Civility in State Legislatures

Abstract

While many scholars and analysts have observed a decline in civility in recent years, there have been few examinations of how political, economic, and institutional structures may partially explain inter-state differences in these trends. We suggest three potential explanations: (1) institutional structures, such as legislative professionalism and gubernatorial power, have created different contexts in which legislators build and maintain inter-personal relationships; (2) partisan competition has led to less bipartisan cooperation and contributed to strained relationships between members of different parties; and, (3) economic inequity and change has contributed to economic anxiety among citizens, contributing to conflict in legislative bodies as elected officials attempt to navigate emerging policy challenges. To test these explanations, we develop an innovative measure of civility using a national survey of lobbyists and a partial Multilevel Regression and Poststratification (MRP) design. Findings suggest that there is some validity to all three explanations, and signifying that civility is at least partially a result of structural issues.

Introduction

The outbreak of COVID-19 has spurred a vigorous political debate in America concerning the balance between public health and economic freedom as national, state, and local leaders decide when and how to reopen economies. This has tested the limits of civility in American politics and public life, as elected officials shift blame and throw accusations at one another and average citizens have vitriolic interactions in public places. While these events and the strain they have put on our economic, social, and political institutions are unprecedented, scholars and political analysts have observed that civility has been on the decline for years. For instance, polling in 2017 indicated that Americans are increasingly concerned about civility in all aspects of American life. Sixty-nine percent of Americans believe that the United States has a major civility problem (Andrews 2017). That number rises to 93% when including those who think civility is a minor problem. Further, the number of Americans who think incivility is a problem is increasing over time (Shandwick and Tate 2018). Many scholars and observers have linked the decline in civility to the increase in partisan polarization and identity politics, but there have been few examinations of how political and economic structures affect civility.

In order to address this, we examine how political institutions, partisan conflict, and economics explain perceived civility in state legislatures. We suggest three potential explanations for inter-state differences: (1) institutional structures, such as legislative professionalism and gubernatorial power, have created different contexts in which legislators build and maintain inter-personal relationships; (2) partisan competition has led to less bipartisan cooperation and contributed to strained relationships between members of different parties; and (3) economic inequity and change have contributed to economic anxiety among citizens, contributing to conflict in legislative bodies as elected officials attempt to navigate emerging policy challenges. To test these explanations, we develop an innovative measure of civility using a national survey of lobbyists and a partial Multilevel Regression and Poststratification (MRP) design. Findings suggest that there is some validity to all three explanations, signifying that civility is only partially a result of structural issues.

Literature Review

Civility and civil discourse, or the lack thereof, have become rather common points of discussion concerning the changing nature of American politics. However, it is not always clear “what precisely ‘counts’ as civil or uncivil behavior…[or] how do we differentiate between useful information relayed in a negative tone and the type of rudeness capable of undermining democracy altogether” (Strachan and Wolf 2012, 401). The Institute for Civility in Government (ICG) defines civility as “more than just politeness, though politeness is a necessary first step. It is about disagreeing without disrespect, seeking (and finding) common ground as a starting point for dialogue about differences, listening past one’s preconceptions, and teaching others to do the same” (Institute for Civility in Government 2020). A key challenge in examining civility and/or incivility is that they are grounded in perceptions of sociocultural norms of acceptable behavior. Thus, in order to determine what is civil or uncivil, one must interpret what is normal and appropriate behavior and how far one may deviate from these behaviors before crossing the line into abnormal or inappropriate (i.e., what is civil to one may be perceived as uncivil to others). Furthermore, civility is more than just words exchanged, and includes tone, inflection, and body language, and there are different standards imposed across settings so that what may be considered hostile in an office place may be normal on a sports field.

Additionally, there is the underlying issue of performative acts of civility where one may act cordially out of courtesy, but not faithfully engage in deliberation (e.g., pretending to listen politely externally, but ignoring the other person’s argument internally). As such, this requires others to perceive and interpret the meaning of inter-personal exchanges and whether it conforms to commonly accepted behavioral norms. Consequently, most definitions of civility and incivility have an implicit subjectivity that relies on the perceptions of participants in exchanges to determine whether they conform to commonly held standards for appropriate discourse within context (Boatright, Shaffer, and Sobieraj 2019; Jamieson et al. 2017; Strachan and Wolf 2012). Thus, in order to examine civility, we must recognize that it is rooted in perceptions of what are appropriate and normal ways to treat others, and these perceptions may differ across groups and geographies. To this end, we define civility: as the perception that one is complying with commonly accepted behavioral norms for how to interact with others in an appropriate manner given the context. As a corollary to this definition, incivility then is: the perception that one is deviating from commonly accepted behavioral norms for inter-personal interactions to the extent that their behavior is considered inappropriate given the context.

Despite the fluid nature of its definition, civility plays a key role in cultivating inter-personal relationships and may provide a strategic political advantage as elected officials work together over the long-term, so while civility is a component in everyday political life for average citizens, the potential negative externalities (e.g., lack of productive communication) are most apparent among elected officials who must maintain productive working relationships in order to conduct the business of democracy. In general, legislatures are designed to foster civility and manage conflict in policy deliberations in order for adversarial political processes to produce consensus (Uslaner 1993, 2015). For instance, the use of parliamentary procedures, committee structures, the hierarchy of leadership, and other rules help establish order and norms of respect and civility. Furthermore, informal behaviors emphasize process, protect the right to speak, foster respectful interactions among members, discourage personal attacks, and encourage deference to seniority and leadership (Dodd and Schraufnagel 2012; Rosenthal 2004). Normatively, civil debate is the normal mode of operation in legislative bodies, so incivility involves a departure from or ignorance of civility norms in favor of behaviors that may include personal attacks like insults, name-calling, or even physical alterations (Dodd and Schraufnagel 2012). Although civility and civil discourse are national issues that permeate politics at federal, state, and local levels, state legislatures have been a particular focal point for advocates and nonprofits, such as the ICG, seeking to improve civil discourse in our public institutions.

Examples of incivility in recent years include shouting matches that occurred in Pennsylvania’s Senate over ending a cash assistance benefits program (Law 2019) and in Maine’s House over a bill to ban conversion therapy (McCammon 2019), as well as false smear campaigns in Kansas (Shorman 2020). Former New Jersey Governor Chris Christie provides several colorful examples including calling a state legislator “numb nuts,” asking a reporter “are you stupid?,” and demanding a protester “sit down and shut up” (Churchill 2019). Along with the National Institute for Civil Discourse (NICD), the ICG has sponsored civility workshops for state legislatures across the country, including one for the entire Idaho state legislature in 2016 (Russell 2016). As legislatures are the most prominent policy-making venue in most states, the discourse between legislators likely establishes norms of political discourse for the rest of the subnational government. Given the number of state and local officials, as well as lobbyists or average citizens, who engage with legislators during a session this sets a tone for what is acceptable behavior. Additionally, many federal officials gain initial experience through state and local government, so behavioral norms from state legislatures may also trickle up to affect national government. With more than 7,000 legislators serving across 50 states, state legislatures are an adept unit-of-analysis for us to isolate some of the institutional, political, or economic factors that may explain civility.

Institutions

Institutions create the setting by which behavioral norms are established for individual policy actors, so it is likely that institutional features contribute to civility. To this end, observers of state legislatures have long noted changes to institutions have increased the level of conflict and incivility among members (e.g., Meinke and Hasecke 2003). Theoretically, this is described by scholars of institutional rational choice, who argue that institutional rules shape individual valuations of rational behavior. Key mechanisms by which this occurs is through formal or informal incentives (or disincentives) for compliance with rules or norms. From this perspective, choice (and by extension, behavior) is based on a quasi-cost–benefit analysis, so that rational policy actors will choose to follow the path that produces the most benefit at the least cost (Ostrom 1999). As such, if failure to comply with civility norms is met with high costs (e.g., censure and suspension of committee assignments), then policy actors are more likely to comply. While there are rudimentary examples of this (e.g., Joseph McCarthy’s censure for verbal abuse of colleagues), institutional rules of legislative bodies likely create incentives for civility in indirect ways with a complex causal pathway (Weingast 1979).

On one hand, institutional designs of state governments in general and legislatures in specific likely shape the strategic political advantage to maintaining productive working relationships with other government officials (Berkman 1994). For instance, in states where policy communities are smaller, more stable over time, and/or based on community ties, alienating actors within a community is likely to make it difficult for legislators to achieve their goals. Conversely, in states with larger, more fluid policy communities tied to professional networks, legislators may be able to be uncivil to some, and still find other professionals who are willing to set aside personal feelings to work with on policy issues. On the other hand, institutional designs also likely shape the electoral or political pressure on individual legislators, which may increase the probability of emotional, rather than logical, responses to political debate. Particularly in states with powerful legislatures, legislators are likely under more pressure to prove themselves to both party leadership and their constituencies, as there is more competition for their position and media scrutiny. As such, legislators may be more likely to see adversarial debate as a threat and respond with emotion, which reflects the high stakes game they are playing.

Among these institutional differences is the professionalism of legislative bodies, with its resultant increase in careerism among members and reliance upon staff. A key difference between professional and citizen legislatures is legislators serving as full-time professional politicians, as opposed to citizens who serve part-time as policymakers (Moncrief and Squire 2020; Moncrief, Thompson, and Kurtz 1996; Rosenthal 2005). Most notably, increased professionalism weakens the traditions and norms of legislative bodies, and leads to a marked decrease in socializing among legislators that further undermines the mutual trust that results from inter-personal relationships. In contrast, citizen legislatures tend to be tied to more traditional, closer communities that include bipartisan alliances, and stronger inter-personal relationships (Andrews 2017; Rosenthal 2005; Thompson, Kurtz, and Moncrief 1996). Furthermore, legislative professionalism is also linked to the increased polarization of political parties, which reduces the avenues for bipartisan consensus and increases the electoral incentives for conflict (Hinchliffe and Lee 2016). Thus, legislators serving in professional legislatures likely perceive the cost–benefits of civility differently than those serving in citizen legislatures, so we expect professional legislatures to correlate with lower levels of civility.

While not directly a feature of legislative bodies, gubernatorial power may also shape civility in legislatures. Although gubernatorial powers include both formal and informal powers, institutional powers of governors in particular may structure the state policymaking process in a manner that may influence interactions between legislators. Specifically, stronger governors have more control in the budget process, longer tenure potential, more appointment powers, and more veto powers (Dometrius 1979). Additionally, stronger governors have more capacity to define policy priorities and drive policy in the direction of collective goods, which reduces particularistic policies and disagreements over zero-sum benefits (Barrilleaux and Berkman 2003; Lewis, Schneider, and Jacoby 2015). By extension, stronger governors and executive branches tend to also weaken legislative branches. As such, legislatures are likely held less accountable for policy outcomes, which reduces the pressure on individual legislators, leading to more civil and less emotional deliberations. In contrast, a weaker governor places more policymaking power and responsibility on the legislature, which may increase the stakes for legislators, leading to more conflict as policy proposals are debated. Thus, legislators serving in states with strong governors are under less pressure to break from civility norms, so we expect stronger governors to be correlated with lower levels of civility.

Finally, the respective size of state governments may be a factor in legislative incivility as well. In a similar vein to gubernatorial power, there are likely higher stakes to the legislative process in states where government plays a larger functional role in the economy. Most directly, implications of policy choices for social or economic well-being increase with the relative size of government, where citizens are more reliant on government expenditures in one form or another (Aidt 2012; Malhotra 2006). For instance, in states with relatively large governments, budget cuts will likely be felt widely by the public as state employees, contractors, and those receiving aid will be directly impacted. Of course, this likely puts legislators under more pressure, leading to less civility when conflict emerges. Additionally, states with relatively larger governments also likely have larger policy communities, which provides more opportunities for legislators to find policy actors willing to overlook uncivil treatment. In contrast, legislators in states with relatively smaller budgets are likely dealing with smaller policy communities where their reputation precedes them, so it is important to retain impressions of civility. Thus, legislators serving in states with relatively large governments are playing higher stakes games with less consequence of uncivil treatment, so we expect larger state governments to be correlated with lower levels of civility.

Partisan Conflict

One of the most obvious and often discussed antecedents of incivility is increasing partisan conflict. In recent years, as the parties have become more polarized, there are fewer opportunities for bipartisan cooperation on key issues, which has also reduced the need to maintain working relationships with members of the opposing party. Previous scholarship on Congressional deliberations finds more incivility when polarization is higher, and that partisanship and polarization in state legislatures now mirror Congress with some state legislatures even more polarized than Congress (Dodd and Schraufnagel 2012; Rosenthal 2005; Shor and McCarty 2011; Uslaner 1993, 2015). Connected to these developments is the heightened partisanship among citizens, especially increased affective polarization (Iyengar et al. 2019). Americans strongly dislike opposing partisans, partially driven by the alignment of partisanship with other social identities, including race and religion (Mason 2015). Affective polarization reduces citizens’ interactions with and trust of opposing partisans. Moreover, it can motivate discrimination against members of the opposing party (Iyengar et al. 2019). This is largely built on an “us versus them” mentality, which some scholars have described as a form of tribalism (Van Bavel and Pereira 2018). In many cases, this results in stereotypes (e.g., Republicans are close-minded, Democrats are socialists) being used as the basis for personal attacks aimed at undermining the character of opposing party members.

We believe there are at least two avenues by which partisan conflict affects legislative civility. First, at a psychological-level, scholars argue that as partisan identities become a bigger part of political identities, individuals see members of other political parties as threats and respond accordingly (Huddy, Mason, and Aaroe 2015). In other words, when facts or opinions that challenge one’s existing cognitive biases towards certain ideological preferences are presented, they are perceived as attacks on one’s own identity, which are then met with an emotional response that is common when faced with threats. Furthermore, some scholars contend that ideologically motivated information processing is used as a mechanism to “promote individuals’ interests in forming and maintaining beliefs that signify their loyalty to important affinity groups” such as political party (Kahan 2013, 407). Being presented with facts or opinions that challenge one’s beliefs can also trigger cognitive dissonance (i.e., state of having inconsistent, thoughts, beliefs, or attitudes), which requires individuals to rectify their underlying existing beliefs with new information (Acharya, Blackwell, and Sen 2018). While this can lead to more deliberative opinions, it can also arouse negative emotions, and previous scholars have connected disagreement directly to incivility in political debate (Masullo Chen and Lu 2017).

While threat responses to attacks on one’s identity occur regardless of party affiliation or political ideology, previous scholars contend that there are cognitive differences between liberals and conservatives that contribute to the way that they respond to their political environment. Most notably, conservatives tend to be more sensitive to threats and risk-averse, which typically manifests as less “open-minded” to new ideas, especially when those ideas may challenge their underlying political identity. The core ideological components of conservativism prioritize resistance to change and seek to manage uncertainty and threats as they vary by situation, which would generally indicate that challenges to accepted orthodoxy are likely to be met with negative responses (Jost et al. 2003). Furthermore, De Zavala, Cislak, and Wesolowska (2010) find that “conservative political outlook inspires coercive and aggressive behavior but only when the intergroup situation is portrayed or perceived as a threat to the ingroup or an intergroup conflict.” Consequently, Republican legislators may be more likely to perceive opposing viewpoints as threats and respond with uncivil remarks as compared to Democratic legislators, especially when Republicans hold the majority and political debate is perceived as a challenge to cohesion in the dominant group.

Second, electoral competition is a key avenue by which legislators are presented with threats to their positions, with previous scholars tying this to increases in state legislative polarization. In recent years, many more legislative seats have become competitive, and the need to raise campaign funds and spend time on campaign-related concerns has increased as well (Abbe and Herrnson 2003; Rosenthal 2005; Thompson, Kurtz, and Moncrief 1996). More frequent shifts in the partisan control of legislative chambers also increase the competition and conflict between parties as they seek to distinguish themselves (Hinchliffe and Lee 2016). Political rhetoric, particularly in campaigns, also contributes to political conflict and incivility, and increased scrutiny of legislators by the media has amplified the feeling of incivility, especially with the development of social media to monitor every legislative move (Andrews 2017; Moncrief, Thompson, and Kurtz 1996; Rosenthal 2005). As a consequence, legislators are likely under more threat of losing their position, and parties have less capacity to help broker or peacefully manage conflict within legislatures (Dahl 1967; Dodd and Schraufnagel 2012). Thus, we expect there to be a correlation between electoral competition and civility, where states with more partisan competition in elections will be more uncivil.

Economic Inequality and Change

Historically, economic struggles tend to underline political conflict, with the linkage being observed in partisan polarization, democratization, voting patterns, and policy choices. In general, scholars argue that economic heterogeneity and/or instability tend to cause ideological positions or policy preferences to be concentrated within citizen groups, leading to competition and conflict over resources and benefits controlled by governing institutions (McCarty, Poole, and Rosenthal 2016). In contrast, when economic systems create homogeneity and stability, policy preferences tend to be more equally distributed across groups, which produces more avenues for consensus. For instance, in states with high income inequality and a changing marketplace that is displacing low-income earners, there is likely to be a high degree of conflict over economic policies. Specifically, low-income earners are likely to advocate for policies that reduce their economic hardships, while high-income earners are likely to advocate for a continuation of policy regimes that benefit their caste.

To this end, the wealth gap in America has steadily increased over the last 50 years, and recent polling data indicates that 61% of Americans say there is too much inequality (Schaeffer 2020). Further, while economic growth tends to be viewed as a positive, its benefits are not always equally distributed and can lead to displacement and competition over resources (Batabyal and Nijkamp 2014). On one hand, the U.S. has shifted towards a service-based economy and a strong technology sector, and away from more traditional industries, such as manufacturing and agriculture. As such, some areas are experiencing economic retraction, which may lead to more conflict as citizens and legislators are faced with tough choices about their community linkages, economic realities, and policy preferences (Harrison 2017). On the other hand, Americans have also become more mobile in recent decades, seeking out more desirable locations with better amenities, including job availability and housing affordability, and political cultures that match their preferences (Bishop 2008). Consequently, these areas are seeing the character of communities change by the influx of newcomers and the evitable displacement of long-term residents. Additionally, economic change also leads to questions about how to manage growth and how benefits will be distributed across groups, which dovetails with broader concerns about inequality.

At the micro-level, economic identities are firmly intertwined with political identities, so policy positions that may be undermining one’s economic position within a community may be perceived as an attack on one’s political identity, leading to emotive responses (Dalby and Mackenzie 1997; Fadjukoff, Kokko, and Pulkkinen 2010). Psychologically, this is grounded in information processing where people tend to rely on cognitive biases in order to cope with analyzing complex decisions. In essence, people regress to preprogrammed emotive responses in order to make quick judgments based on previous experiences, with negative experiences (i.e., threats) tied closer to emotional responses than positive experiences. Thus, when faced with circumstances that threaten their economic identity, people are likely to respond with negative emotions, rather than with logical, objective discourse (Arceneaux 2012; Bargh and Chartrand 1999; Rozin and Royzman 2001). For instance, most coal workers in West Virginia are following an inter-generational legacy from which their personal identities are based, and local community customs and traditions tend to be tied to the coal industry. Thus, for coal workers, debates about the future of fossil fuels would likely trigger anxieties about joblessness and economic struggle, as well as the end of family legacies and the erosion of community ties (Bell and York 2010). Therefore, when faced with this complex situation, they are likely to respond emotionally, which may be perceived as incivility.

Given that legislators spend a notable portion of their time interacting with citizens, they are likely to pick up on the underlying emotive aspects of their responses to policy or ideological questions. While neuroscientists and psychologists argue that there is an evolutionary component, sociocultural norms, as well as cognitive biases, are learned behaviors, so exposure to how others are responding to the political environment is a mechanism by which citizens may communicate with legislators about their policy preferences, and vice versa (Arceneaux 2012; Petty and Cacioppo 1986). If citizens have emotionally charged responses to certain policy issues, then legislators may see that level of emotion as appropriate, and as a tool in communicating the seriousness by which the issue should be taken. For instance, a legislator that attends a town hall in which people respond to policy issues with high emotions is likely to internalize a cognitive bias learned from that experience, and respond to the same policy issue presented during the legislative process with a similar emotionally charged response. This may occur passively via the unconscious acquisition of a bias, or actively by legislators seeking to represent their constituents. In either case, the emotional responses to economic threats by citizens are likely to influence legislator behavior when dealing with the same issues during legislative sessions.

This is likely exacerbated by partisan polarization and reduced opportunities for relationship building with the opposition party. McCarty, Poole, and Rosenthal (2016) contend that when income inequality is high, interests of high- and low-income earners are more likely to be in direct competition, which pushes the respective ideological positions of their parties further apart. In contrast, when there is relatively low inequality, parties are more likely to include a heterogeneous mixture of income-levels, leading to a convergence of policy positions. Within legislative bodies, Republicans and Democrats in states with high economic inequality are likely to find far fewer issues on which bipartisan cooperation is possible, as compared to their counterparts in low economic inequality states. Consequently, it may be easier for legislators to dehumanize the opposing party and rely on cognitive biases in responding to people who they perceive as a threat, as opposed to legislatures that have a more bipartisan character which enables legislators to build and maintain relationships across party lines and replace cognitive biases based on negative emotional responses with those tied to positive experiences.

Methods

Measuring Civility

Measuring civility (or incivility) can be difficult due to the subjective nature of the topic. As discussed above, definitions of civility often focus on politeness and respect in discussions and behavior, including in disagreements; although, civility is not just politeness. Furthermore, civility is often based on perceptions and context, so that it may be difficult to identify from the outside looking in. While civility norms guide how people relate and interact in their workplaces and daily lives, differentiating civil and uncivil behavior can be difficult (Strachan and Wolf 2012). As such, scholars often measure civility (or incivility) with survey questions asking about individuals’ perceptions (e.g., Clark, Sattler, and Barbosa-Leiker 2018; Donovan et al., 2016; Ferriss 2002), which can then be combined into a scale or index (e.g., Ferriss 2002; Walsh et al. 2012). On one hand, this creates limitations in that any measure of civility becomes biased by the individual’s perception of what is appropriate or inappropriate behavior for the context. On the other hand, it is likely implausible to create an “objective” measurement of civility that is free from perceptional biases and the social construction of behavioral norms.

In keeping with our definition that embraces the perceptional nature of civility, using the survey approach helps encompass different individual-level perceptions or definitions of civility and use them to measure civility at the group-level in order to come to an understanding of a collective interpretation of how well actual behavior aligns with expectations, given the context. To this end, we construct a state civility index using a three-step process. First, we use data from the National Survey of State Legislative Lobbyists (2018–2019) to measure individual perceptions of state legislative behaviors. Given that lobbyists serve as key political actors in legislatures but are independent of institutional responsibilities, their perceptions provide us with a relatively objective point of comparison across states. The survey was developed by a consortium of researchers located at 11 universities across the United States, and supported by the National Institute for Civil Discourse at the University of Arizona, the Thomas S. Foley Institute for Public Policy and Public Service, the William Ruckleshaus Center, and the Division of Governmental Studies and Services at Washington State University. The sample was selected randomly from the official state registries of lobbyists in each state, usually obtained through the Secretary of State’s office.

The survey was done in three phases. The first phase was an online Qualtrics survey using a randomly selected sample of registered lobbyists. The second phase was a mail survey follow-up conducted in two waves to the same sample, financed by the NICD. The third phase was done in 12 oversampled states ¹ where an attempt was made to reach a minimum of 30 responses by adding some additional returns for known experienced lobbyists. There were 1,257 total surveys returned across 50 states with an average of 25 respondents per state. Response rates varied by state, with an overall response rate of 25%. After reviewing the survey, we identified three items (see Table A1 for descriptions) that capture lobbyist perceptions of state legislative activities as it relates to civility.

The first item asks respondents to rate the quality of legislative deliberation (rated on a scale from (1) uniformly poor to (5) uniformly good), and the second two items asks respondents to rate the general civility in their state legislature (rated on a scale from (1) very uncivil to (9) very civil) and the general civility of behavior for legislators in their state (rated from (1) very uncivil to (5) very civil). In order to establish some consistency in item interpretation, the survey defines deliberation in the question as “fairness of process of hearing differing views, due consideration of evidence and testimony and commitment to doing the right thing for the people of your state.” Additionally, for the civility items, respondents are instructed to respond “based on your understanding of norms of civility.” This is largely in keeping with our definition that suggests that while fairness is a key component, civility largely boils down to individual interpretation of appropriate norms. At the individual-level, items were moderately to strongly correlated (Spearman’s rank correlation ranged from 0.27 to 0.57); at the state-level, items were very strongly correlated (Pearson’s r ranged from 0.44 to 0.86). This would indicate that these items are generally measuring perceptions of a similar dynamic across respondents and states.

Second, we use an adapted multilevel regression and poststratification (MRP) technique to create state-level estimates for each survey item. Although the aggregation method of constructing state-level estimates from individual survey responses is most common, recent scholarship has called into question the validity of that technique and argued for a more sophisticated approach (e.g., Pacheco 2011; Warshaw and Rodden 2012). Specifically, the aggregation approach relies on calculating the mean of responses for respondents from the same state, but this method tends to require large n-sizes and is sensitive to bias if samples are unweighted or there are low response rates for some states. In comparison, scholars argue that MRP produces purer estimations of state-level public opinion that are more accurate and that outperform the aggregation technique along with many indicators (Fowler 2016; Lax and Phillips 2009a, 2009b; Pacheco 2011; Warshaw and Rodden 2012). MRP involves a three-step process. Step one is to estimate a multilevel model with fixed effects for individual-level predictors at level 1 and random effects for states at level 2. Based on this model, an intercept for each state is calculated which represents the comparative differences in the dependent variable when all the individual-level predictors are controlled for. Then, estimates are made for values of the dependent variable for each “person-type” in the dataset. Poststratification weights are applied to those “person-types” based on actual state demographic characteristics.

Given that our interest is in the comparative perceptions of civility across states and not public opinion of a specific (i.e., lobbyists) or general population, we use an adapted form of this technique, and do not estimate values for person-types or apply post-stratification weights. By doing so, our multilevel model controls for individual-level characteristics that may create bias in perceptions of civility (e.g., white males are more likely to perceive high levels of incivility), but we do not take the additional step of extrapolating these relationships to entire state populations. Thus, this approach overcomes one of the primary shortcomings of the aggregation technique, while creating a functional point of comparison that is unbiased by state-level demographic characteristics. The real value here though is that the random effects portion of the multilevel model produces state-level intercepts which represent the distinct differences between states. In other words, once we control for a series of demographic and political characteristics of individual respondents, the state-level intercepts represent how a hypothetic lobbyist would likely perceive civility within any given state, which provides us with a point of comparison that is relatively unbiased.

In order to execute this step, we estimate multilevel ordered logistic regression models of our three survey items, which all use ordered response categories. ² At level 1, we used a series of variables to control for individual-level characteristics. For demographics, we include age, gender, and race. ³ Additionally, we include variables for partisan affiliation and political preferences on social and fiscal issues. ⁴ We also include four variables to account for how often the lobbyists represent business/business associations, professional trade associations, labor unions, and public interest nonprofits. ⁵ Results indicated our models were relatively good fits for the data (see Table A2). Demographic factors proved to be the most consistent predictors, with older, white males most likely to perceive high levels of civility. We then calculated the random effects portions of the model for each dependent variable, which produced state-level intercepts for all 50 states.

Third, we calculate the mean for each state across the three survey items to create a single measure of state-level civility. Given that this measure could potentially be affected by respondents across states interpreting the scale and incivility in different ways, we use another survey item asking respondents to rate the civility of state legislators in the United States as a common point of comparison to correct for these biases. Specifically, we estimate state-level intercepts using the same process as outlined above and subtract this estimate from the average of the scale item. ⁶ This final estimate, then, represents how an average lobbyist is likely to perceive civility of state legislators from behind a veil of ignorance of both their own personal and professional experiences as well as interpretational differences that may occur in responding to the survey. The civility index has a mean value near zero (4.00e-06) and ranges from −0.971 (Oklahoma) to 0.763 (Maryland). The index functions like a standardized variable in that positive values represent values above the mean (i.e., more civility), while negative, below the mean (i.e., less civility). Figure 1 illustrates the distribution of the civility index across states.

Figure 1. Civility Index by State.

Note. Positive values indicate more civility, negative values indicate less civility.

Testing Explanations

We use ordinary least squares (OLS) regression to test our hypotheses. Diagnostic tests indicate no OLS assumptions were violated. Specifically, variance inflation factor scores indicate no multicollinearity between predictor variables, and White’s test for heteroskedasticity fell within acceptable ranges. Although we have 50 states in our dataset, Nebraska is not included in our analytical models due to its non-partisan legislature, reducing our analytical dataset to 49 observations. We operationalized our potential explanations in state civility with seven predictor variables (see Table 1 for descriptions). For time-variant predictors, we used the most recently available data.

Table 1. Variable descriptions for OLS models

Abbreviation: GDP, gross domestic product; OLS, ordinary least squares.

First, to test the effects of institutions on lobbyists’ perceptions of state legislative civility, we use citizen legislature to account for the degree to which legislators engage in politics at a professional-level, gubernatorial power to account for the balance of power with the executive branch of government, and government size to measure the scope and breadth of subnational government across states. Citizen legislature is measured as an ordinal variable ranging from full-time professional legislatures (0) to part-time, citizen legislatures (4). We use the National Conference of State Legislature’s classification system, which is based on the amount of time legislators spend on legislative work, compensation, and the size of professional staff. In part-time citizen legislatures, legislators spend less than half of a full-time job doing legislative work, with low compensation, and little professional staff, as compared to full-time professional legislators whose full-time job is legislative work, are compensated accordingly, and supported by large staffs. We measure gubernatorial power using Beyle (2007)’s index by taking the numerical average across the six areas of institutional powers identified: budgeting, veto power, separately elected officials, tenure potential, gubernatorial party control, and appointment power. ⁷ Each area is rated on a scale from one to five with higher scores indicating more gubernatorial power. Government size is measured as the percentage of state gross domestic product produced by state and local governments, using data from the US Bureau of Economic Analysis (BEA).

Second, to test the effects of partisan politics, we use partisan control of state legislatures, and electoral competition to control for effects derived from contested election cycles. For partisan control, we used data from the NCSL to create an ordinal variable ranging from Democratic control (0) to split control (1) and Republican control (2). For electoral competition, using data from Ballotpedia, we measure the portion of elections from the 2018–2019 cycle in which candidates from both parties appeared in the general elections.

Third, to test the effects of economics, we use economic inequality to account for inequitable distributions of resources within states, and growth in the gross domestic product (GDP) to account for economic change. For inequality, we use GINI coefficients based on estimations for the US Census Bureau. For economic growth, we use the percentage change in GDP between 2010 and 2018, using data from the BEA. ⁸

Results

Table 2 displays results for the OLS model and indicates support for our explanations. In general, perceptions of civility in state legislatures is higher for states with citizen legislatures, Democratic-controlled legislatures, low electoral competition, low economic inequality, and a growing economy. Although gubernatorial power and government size are not statistically significant, coefficients would indicate that civility is higher in states with strong governors, and smaller governments, but these findings may not be generalizable. Standardized coefficients further indicate that electoral competition is by far the strongest indicator of civility, with partisan control also standing out as a particularly important predictor. On the other hand, citizen legislatures, economic inequality, and economic growth are relatively comparative in their substantive power in predicting civility. This may suggest that partisanship is a root cause of incivility, which is not particularly surprising given the prevailing anecdotal evidence and media accounts. On the other hand, it is particularly notable that institutional and economic structures are also contributing factors, and these explanations are likely deserving of further inquiry. Finally, the R ² and adjusted R ² indicate our model is a strong predictor of the level of perceived civility across states.

Table 2. OLS results

Note. Standard errors in parentheses.

Abbreviation: OLS, ordinary least squares.

^* p < 0.1.

^** p < 0.05.

^*** p < 0.01.

Discussion and Conclusions

Our findings provide some evidence to support our institutional, partisan, and economic explanations of rising incivility in state legislatures. Given the substantive strength of partisan control and electoral competition, a root cause of incivility is certainly tied to polarization’s impacts on campaigning and bipartisan coalitions in legislatures. On this front, our findings would also suggest that incivility is more common among Republican-controlled legislatures, which is likely tied to group loyalty among conservatives and their responses when under attack. This dovetails with our findings on economic inequality, where states with higher inequality are likely to have more political strife surrounding policy decisions. Consequently, it is likely that legislators in these states find fewer opportunities for bipartisan cooperation, so maintaining decorum with members of the opposing party is less of a concern and conflict is more likely to deteriorate into incivility. However, our findings are limited on this front, so it is difficult to draw detailed conclusions about the specific causal pathway at the group, inter-personal, or neuro-cognitive-level.

While our findings in relation to partisanship and economic inequality are unsurprising, the institutional explanations suggest additional nuances to the context in which incivility arises. Incivility appears to be most common where legislators are full-time professional politicians and rely less on their inter-personal connections. Specifically, in states with citizen legislatures, legislators are playing a less active role in state politics and are operating among a discrete group of policy actors. Thus, legislators are both under less pressure and have more incentives to maintain working relationships with others. In contrast, legislators in states with professional legislatures are on the forefront of political decision-making, so they are under more scrutiny from media, peers, and constituents. Additionally, they may have more opportunities to behave badly and still find policy actors willing to engage with them, as compared to their counterparts in states where a bad reputation likely creates barriers to achieving policy goals. Although partisan polarization and economic inequality may be broader issues that are unresolvable in the short-term, our findings suggest that some institutional reforms may create an environment more conducive to civility.

Possibly the most important contribution here to the scholarship on incivility is our methodological approach. While civility has been the topic de jour for some time, researchers have struggled with how best to measure it in a way that allows for empirical analysis. Although our approach does not allow us to make specific inferences about civility at the individual-level or how micro-aggressions occur, it has allowed us to make inter-state comparisons and test hypotheses concerning how structural and contextual characteristics of legislatures contribute to incivility, which represents a novel contribution to this scholarship. Further, the MRP approach is a sophisticated technique for estimating state-level public opinion, but our adapted version provides a unique method in that it estimates how a hypothetical lobbyist is likely to perceive legislative civility within any given state. On one hand, using a multilevel ordinal logit model allows us to remove bias created from individual-level characteristics (i.e., demographics, lobbying experience) when estimating state-level intercepts. On the other hand, by not extending these estimates to a specific or general population within states, we remove biases that come along with population heterogeneity across states. The result is a relatively unbiased look at how an “average” lobbyist (when controlling for demographics, state, etc.) would view civility within a specific legislature.

Our findings are still limited to perceptions of lobbyists though, which constrains the inferences that we can make from this data. Most notably, this is based on perceptions so that it relies on an individual’s evaluations of both what is civil behavior and whether or not state legislators are deviating from it. From a research design perspective, this allows our measure to capture the different ways in which civil or uncivil behavior may manifest across states and in the minds of respondents. However, it also means that there is likely a lack of consistency in how civil behavior is judged. Although they have a unique perspective of legislatures as outside policy actors, lobbyists serve a specific function in influencing legislative processes and may have conflicting interests at times. Consequently, their perceptions may materially differ from those of other policy actors, such as citizens, executive branch officials, reporters, or legislators themselves. In sum, we must be circumspect in inferring too much from these findings as other groups may come to different conclusions about the degree of civility present in any given state legislator. After all, we would have to imagine that someone directing uncivil behavior at a colleague likely sees their behavior as warranted and appropriate under the circumstances, even if others do not. Consequently, additional research should compare and contrast perceptions across other groups of policy actors to determine how accurate and/or valid our estimates are as a broad measure of legislative civility, and whether it is possible to come to an unbiased measure of civility across state legislatures.

Additionally, our survey considers state legislatures as a single body and does not differentiate between upper and lower houses. It is certainly plausible for civility to fluctuate between the two, as chambers have historically been designed to represent different interests. For instance, at the national level, this has led to different norms of civility across chambers, where verbal abuse has been more aggressively condemned in the Senate as compared to the House (e.g., Joseph McCarthy censure). This may be particularly important as scholars consider the institutional mechanisms that impact civility norms, so that inter-chamber rule or leadership differences may play an important role in maintaining civility. Furthermore, more nuanced data at the individual level is necessary to identify the specific causal pathway that leads from structural, economic, and/or political factors that exist at the aggregate state-level to behaviors by individual legislatures. Finally, and possibly most importantly, future scholarship should also explore how these contextual elements can be used to support civil interactions and the norms of decorum that existed in legislative bodies for much of US history.