Bad apples or rotten orchards? Public attitudes of interactions with police and the role of political ideology

Abstract

Purpose

The purpose of this study is to explore the ideological gaps across a range of policing interactions with the public.

Design/methodology/approach

In a survey distributed via Mechanical Turk (MTurk) (n = 979), the authors explore the role that respondents' political ideology plays in the agreement of 13 aspects of policing services, their demeanor and decorum.

Findings

Attitudes toward policing interactions are slightly positive. Conservatives steadfastly hold positive attitudes about police. Liberals vacillate from negative to positive attitudes across the 13 policing interaction statements.

Social implications

Although small, there is an ideological consensus that police adequately protect citizens and are knowledgeable about the law.

Originality/value

Even at record lows of public confidence in the police, some subsections of the sample, such as conservatives, firmly hold positive attitudes about police. The unwavering support for police by conservatives continues across the multi-item measure of policing interactions, whereas liberals illustrated less uniformity in their attitudes.

Citation

Hansen, M.A. and Navarro, J.C. (2023), "Bad apples or rotten orchards? Public attitudes of interactions with police and the role of political ideology", Policing: An International Journal, Vol. 46 No. 5/6, pp. 861-877. https://doi.org/10.1108/PIJPSM-07-2023-0098

Introduction

North America reported the greatest decline in confidence in local police than other regions of the world between 2020 and 2021 (Gallup, 2022). The 2020 and 2022 Gallup polls of public confidence in police recorded a value of less than 50% for the first time since Gallup began recording public sentiment toward police (Gallup, 2023). The most recent year of data collection—2022—shows a record low of the public who had quite a lot/a great deal of confidence in police (45%) across the thirty years of polling data. These trends are significant as individuals with lower confidence in the police are less likely to cooperate with and tend to view the police with less legitimacy (Pyo, 2023; Ren et al., 2005; Sunshine and Tyler, 2003b; Tyler and Fagan, 2008). Evaluations of police services are important as the public are its consumers, evidenced by the 53.8 million interactions recorded in 2020 (Tapp and Davis, 2022). Given their status as gatekeepers to the criminal justice system, public sentiment toward police can also shape perspectives of legitimacy toward other criminal justice systems such as prosecutions, courts and corrections (Albrecht and Green, 1977; Foster et al., 2023; Pyo, 2023).

Attitudinal divisions on policing occur across a range of empirically and theoretically relevant variables (see Bolger et al., 2021 for review; see Brown and Benedict, 2002 for review; see Peck, 2015 for review) despite the public holding relatively positive attitudes toward the police (see Brown and Benedict, 2002 for review) with the police routinely being among the top three most trusted institutions year after year (Gallup, 2023). Scholars have consistently advocated investigating possible confounders between other socio-demographics like race and socioeconomic status on attitudes toward police (Brown and Benedict, 2002; Cao and Wu, 2019). For instance, polarization among public reactions to the police can be driven by philosophical or political perspectives (Albrecht and Green, 1977; Sunshine and Tyler, 2003b). Non-demographic factors such as political ideology can shape American attitudes toward policing (Cao et al., 1998; Ekins, 2016; Huang and Vaughn, 1996; Navarro and Hansen, 2023; Reynolds et al., 2018; Williams and Maxwell, 2022; Wozniak et al., 2021; Zamble and Annesley, 1987) and to a greater extent than political partisanship (Navarro and Hansen, 2023; Wozniak et al., 2021). Although attitudes toward policing may be broadly polarized along ideological lines (Hansen and Navarro, 2023), such polarization may vary across a range of policing interactions, resulting in no attitudinal differences toward policing between conservatives and liberals (Cao et al., 1998).

Our contributions in this line of policing scholarship are to identify if and how attitudes among conservatives and liberals vary across various aspects of policing. The current study assesses public responses to a national-level survey that measured 13 aspects of police services and officer demeanor and decorum. We first hypothesize that political ideology is an important predictor of attitudes toward a range of police interactions, with conservatives consistently reporting higher ratings than liberals. We next hypothesize that these gaps in agreement across the 13 aspects of police interactions will vary based on liberals' attitudes toward policing, not conservatives. Our findings support Albrecht and Greens' (1977) reasoning that public attitudes toward the police should consider a broader context that includes the individual's personality and fundamental values related to the larger political-legal system by incorporating political ideology when exploring attitudes toward policing.

Literature review

The relationship between police and the public in America has been tumultuous since their formalization in 1837 (Sunshine and Tyler, 2003b). Public perceptions of police competence, effectiveness and performance have garnered sustained scholarly interest (see Bolger et al., 2021 for review; see Brown and Benedict, 2002 for review; see Peck, 2015 for review). Polling data shows that while public confidence in the police (vacillates and) is typically among the top three of the institutions tested, recent polling data suggests that public confidence in the police is becoming less favorable. A record low occurred in 2020, which was surpassed in 2022 (Gallup, 2023, see Figure 1). Sometimes, these declines can be isolated to one particular group, suggesting variation across subsections of the population. For instance, favorable perceptions of police were roughly twice as high among older generations, specifically the Silent Generation, Baby Boomers and Generation X, than those of Generation Z (Williams and Maxwell, 2022). These nationwide trends of waning confidence in the police can be reflected by negative perceptions of crime and disorder control management by police (Foster et al., 2023; Tyler and Fagan, 2008).

Since attitudes toward police are composed of a complex interplay of beliefs, there should be a diverse set of questions that tap into the various duties and functions of police, including their police demeanor and decorum (Dai et al., 2018; Koper et al., 2022; Pyo, 2023). Panel survey data before and after the death of Michael Brown in Ferguson, Missouri, lends support for the argument that support for police varies across particular groups but also that attitudes toward police can be similar and dissimilar across social groups, depending on the policing aspect (Kochel, 2019). For instance, Black residents in St. Louis County reported reduced perceptions of police legitimacy and procedural justice (and trust), as well as an increased frequency of witnessing aggressive policing tactics. While non-Blacks demonstrated slight increases across these three items, they were relatively small, with non-Blacks and Blacks both reporting an increased willingness to cooperate. While one of the more consistent divisions in the public's attitudes toward police is by race (Brown and Benedict, 2002; Peck, 2015), meta-analyses have shown that other socio-demographics, including gender, age, education and residential environment, are important determinants in the public's opinion toward police. However, as policing becomes increasingly politicized, it is crucial to identify how relevant variables such as political ideology shape attitudes toward the police.

Public attitudes toward policing and political ideology

Public confidence in the police has become increasingly politicized (see Roper Center for Public Opinion Research, 2022). While conservatives consistently hold higher confidence levels in the police than liberals, Gallup's Confidence in Institutions poll in October 2014—following the August 2014 shooting of Michael Brown in a suburb of St. Louis, Missouri—showed a substantial ideological gap. Conservatives' confidence in police demonstrated a pronounced upward effect following the events surrounding the fatal shooting of Michael Brown by a Ferguson Police Department officer, whereas liberals demonstrated a downtrend in confidence. This ideological gap toward confidence in the police has remained large and unchanged since October 2014. Thus, political ideology is a valuable variable to identify points of contention and possibly identify mutually beneficial goals in policing for conservatives and liberals.

Prior work in this line of research demonstrates that conservatives and liberals held similar opinions toward police, suggesting that this ideological gap in attitudes toward policing is worth reexamination. One study using polling data from the 1995 National Opinion Survey on Crime and Justice identified that attitudes toward a range of police duties, demeanor and decorum, such as their ability to prevent and solve crime, protect people from crime and friendliness were similar between political ideologies (Huang and Vaughn, 1996). Likewise, Cao et al. (1996) did not support their hypothesis that the law-and-order rhetoric of the conservative crime ideology (as measured by support of five “get tough” policies) would be related to increased confidence in the police, suggesting that conservative canon then was not uniquely intertwined with their support of police.

However, subsequent research has identified that attitudes toward police vary by political ideology, with conservatism associated with greater support for the police (Cao et al., 1998; Ekins, 2016; Huang and Vaughn, 1996; Navarro and Hansen, 2023; Reynolds et al., 2018; Williams and Maxwell, 2022; Wozniak et al., 2021; Zamble and Annesley, 1987). Conservatives believed police cared about the people in their community, protected people from violent crimes, are honest and trustworthy and are held accountable for misconduct and treated minorities and disadvantaged groups equally without biases (Ekins, 2016; Wozniak et al., 2021). Conservatives tend to have higher perceptions of police competency than liberals, which is reflected in their higher evaluations of local (and national) police effectiveness in enforcing the law and solving crime. One study demonstrated how political ideology stood apart from other socio-demographics when participants were asked whether the fatal shooting of a hit-and-run suspect (who was holding a knife) by police was justified after watching a body-worn camera video of the investigation call (Reynolds et al., 2018). No other variable in the model predicted that the fatal encounter was justified except for political ideology, with conservatives being 1.5 times more likely to indicate that the shooting was justified. These findings suggest that conservatives hold different attitudes and expectations of how policing is to function in America.

Individuals who identify with the police may share moral values, which could lead them to perceive the police as legitimate defenders of their community's norms and values (Sunshine and Tyler, 2003a). This moral solidarity is shaped by perceptions of fairness in the discretionary decisions exercised by the police. Those who believe they share moral values with the police are more likely to comply, cooperate and empower police with discretionary authority. Police are respected as authority figures, in part as they are perceived to represent the group's values. Social psychological research has consistently demonstrated that conservatives scored high on a moral concern termed respect for authority, which is correlated with increased favorable views toward police (Ekins, 2016; Graham et al., 2009; Haidt, 2012). Individuals with high scores on respect for authority were more likely to support police practices such as stop-and-frisk policies, believe that police tactics are appropriate and the use of lethal force when necessary (Ekins, 2016); liberals, on the other hand, held less favorable attitudes toward police use of force practices than conservatives (Ekins, 2016; Huang and Vaughn, 1996; Navarro and Hansen, 2023). For instance, when rating scenarios depicting reasonable and excessive use of force, there was an ideological distinction in approval (Navarro and Hansen, 2023). Conservatives rated all 13 use of force scenarios similarly, whereas liberals demonstrated a marked difference between what they considered reasonable and excessive. Although conservatives and liberals maintain opposing political perspectives on policing, understanding the degree of differences across a range of policing interactions can identify where a common consensus can be reached among both ideologies.

Hypotheses

H1.

Conservative respondents will have a higher level of agreement with 13 police interaction statements reflecting positively on the police than liberal respondents.

H2.

Very conservative respondents will have a consistent, positive agreement with the 13 police interaction statements, but very liberal respondents will indicate negative, neutral and positive agreement with the police interaction statements.

Data

To study public attitudes toward police interactions with society in the United States, we conducted an original survey through Amazon's Mechanical Turk (MTurk) in conjunction with Qualtrics survey software. The survey was launched on 24 October 2022. Adults aged 18 and older residing in the United States were potential respondents who were paid $0.80 for participation, even if they did not answer all the questions. The average time it took respondents to complete the survey was 6 min and 12 s, which, if calculated at an hourly rate, would be slightly over the federal minimum wage. All individuals who opened the survey gave informed consent. Of the 1,002 respondents who opened the survey, 984 respondents completed all the questions utilized in the multiple regression analyses.

Despite nearly all samples in social science being convenient (Landers and Behrend, 2015), MTurk provides several advantages for social science scholars to conduct surveys and obtain high-quality and useful data. MTurk samples tend to be demographically more diverse than typical online samples or traditional surveys of college students (Buhrmester et al., 2011). Moreover, the attentiveness of MTurk online participants recruited for research is comparable to that of offline participants, thus suggesting that online participants yield reliable, valid data by comparison (Thomas and Clifford, 2017). Likewise, MTurk respondents do not differ from popular national population-based surveys in unmeasurable ways (Clifford et al., 2015; Levay et al., 2016). Regarding criminal justice research, Thompson and Pickett (2020) found that MTurk samples are useful for measuring the direction of coefficients when comparing results from these nonprobability samples to results from the General Social Survey (GSS). The personality traits and values of MTurk workers are virtually identical to high-quality national data like the American National Election Studies (ANES); MTurk liberals exhibited more typical liberal dispositions and conservatives are indistinguishable across samples (Clifford et al., 2015). Therefore, if researchers acknowledge and account for how their sample might differ from the population, researchers can advance science using MTurk (Baker et al., 2010; Landers and Behrend, 2015; Levay et al., 2016). More yet, opt-in panels (like MTurk) “have proven to be a valuable resource for methodological research of all kinds … [but] researchers should … consider any biases … and qualify their conclusions appropriately” (Baker et al., 2010, p. 759). Further reading of the costs and benefits of opt-in panels and suggested best practices are provided by Baker and their colleagues (2010; 2013).

Our sample does not greatly deviate from the population when exploring important socio-demographic and attitudinal trends, with two exceptions. The sample had a substantially higher proportion of male respondents when compared to women. Additionally, after coding partisan leaners as partisans, the sample had a noticeably higher proportion of Democratic identifiers. To account for these two trends in the data, we estimate post-stratification survey weights based on US Census data and aggregated polling data. All multiple OLS regression models estimated in the analysis are calculated with the survey weights incorporated using the “survey” packages in R statistical software so that the sample better approximates the population so to be more representative.

Dependent variables and method

When measuring a broad latent variable, such as overall attitudes of police interactions, it is necessary to have several measures phrased in different ways to accurately capture the respondent's sentiment. Respondents were asked about their level of agreement with 13 statements that assess police officers' interactions with the public in the forms of the services provided and their general demeanor and decorum. The order in which respondents received the statements were randomized. Each of the 13 statements are presented in Table 1. The statements are phrased in a positive manner regarding police interactions with society for survey module consistency. The statements were also phrased in the same manner since mixing positively and negatively phrased statements in the same survey question module could have adverse effects on the internal consistency of the study. We also avoided the use of negatively phrased statements since these statements would be more likely to induce variance that would bias the results in favor of confirming the hypotheses. For each statement, respondents were asked to place their level of agreement on a sliding scale that ranges from “0 – not at all” to “10 – a great degree” [1].

The descriptive statistics in Table 1 indicate a substantial amount of variance (standard deviations >2.34 for variables on a 0–10 scale) in agreement with each of the police interaction statements [2]. In addition, the mean level of the agreement across the 13 statements demonstrates variance. The “[p]olice officers do not let their emotions impact their ability to do their job” statement reported the lowest mean level of agreement (M = 5.37), which was closer to “neither agree nor disagree” than a level of affirmative agreement. The statement with the highest mean level of agreement (M = 6.65) was, “[p]olice officers are knowledgeable about the law.” Again, even for this statement, the mean level of agreement leans closer to “neither agree nor disagree” than it does to “a great degree”. An initial glance at these descriptive statistics demonstrates a concerning assessment from respondents on police interactions with society.

The histograms in Figure 2 provide a clearer picture of the variance in agreement with the police interaction statements. First, a sizable proportion of the sample selected “0 = not at all”. For example, between 50–100 respondents selected “not at all” for all but four of the 13 statements, with few respondents giving low ratings for police as doing an adequate job in policing neighborhoods, protecting citizens, demonstrating courtesy to citizens and being knowledgeable about the law. Similarly, around 50 respondents who selected 10 “a great degree” agreed with 11 of the statements, with roughly 100 respondents indicating that police officers adequately protect citizens and are knowledgeable about the law. For 11 out of 13 statements, the modal level of agreement was a seven on the 0–10 scale. There were no instances where the modal level of agreement was “10 = to a great degree”. Thus far, the results point to a sample that recognizes considerable variation in police officers' interactions with the public.

Independent variables

We utilize several socio-demographic and attitudinal variables to predict agreement with the police interaction statements. First, we include age, gender, race, education and income in our Ordinary Least Squares regression models since the dependent variable is continuous. Second, we utilize a measure for the residential environment of the respondent: rural, suburban and urban (reference category). Third, we include partisan identification in the models as an attitudinal predictor. Finally, as the primary focus of this study is the role that political ideology plays in predicting attitudes of police interactions with the public, we measure political ideology as respondent self-placement on a scale from “0 = very liberal” to “10 = very conservative”. This is a useful approach for measuring political ideology across contexts and is comparable to large-scale, reputable cross-national surveys (e.g. European Social Survey, World Values Survey). The distribution of responses for political ideology mirrors the distributions of nationally representative surveys. Variable coding and descriptive statistics for all variables used in the study's analyses can be found in Appendix A.

Results

Table 2 displays 13 regression models predicting agreement with each of the police interaction statements. There are a few notable trends regarding the effect of the control variables on agreement with the interactions that support the MTurk sample is operating as empirically expected. First, older respondents were consistently more likely to indicate agreement that police officers do an adequate job, are knowledgeable about the law, do not bully civilians, respect the rights of citizens and are courteous to citizens, with no statistical differences between younger and older respondents for the other eight interaction statements. Second, there were only two instances where a gender difference occurred across these police interactions. Women were statistically more likely than men to indicate agreement that police officers adequately protect citizens and are knowledgeable about the law. Third, a greater number of differences in attitudes of police interactions occurred across race, with white respondents more likely than nonwhite respondents to indicate agreement that police officers do not let personal bias influence them, that they treat everyone with dignity, respect individual citizen's rights and provide the same quality of service to everyone. Fourth, a higher education level was consistently associated with a statistically significant higher level of agreement (but varying in strength) across all 13 interactions. Finally, respondents residing in rural areas have a statistically higher level of agreement with 9 out of 13 statements compared to suburban and urban residents. There was no difference between suburban and urban respondents' attitudes of police interactions.

Results for two controls, income and partisan identification, provide mixed insight (but merit discussion) for predicting agreement with the police interaction statements. Income had a negative relationship with predicting two out of the 13 interactions: Respondents of higher income categories were more likely to statistically indicate greater disagreement that police officers treat everyone with dignity and provide the same quality service to everyone. Independents had a statistically lower level of agreement than Democrats when predicting nine out of the 13 statements. However, there was not a clear trend between Democratic and Republican partisans as they were only statistically different in levels of agreement for two statements: Republicans were less likely to agree that police officers do not allow their biases to impact their work but more likely to agree that police officers are knowledgeable about the law.

We now shift to the study's focus on the relationship between the key independent variable (political ideology) and agreement with the 13 police interaction statements. The output in Table 2 indicates a consistent, statistically significant relationship between political ideology and levels of agreement. Conservative respondents have a statistically higher level of agreement across all 13 statements when compared to liberal respondents. The finding provides support for H1. Conservatives are more likely to view the police uncritically and to think that police officers' interactions with the public are mostly positive.

To illustrate the impact of political ideology, we plot predicted scores for the effect of political ideology on agreement with each of the 13 police interaction statements in Figure 3. Overall, political ideology has a positive and substantively notable impact on predicted levels of agreement. The effect of political ideology on levels of agreement is the smallest when predicting whether police officers do an adequate job, adequately protect citizens and are knowledgeable about the law. However, even though the effect of political ideology is comparably smaller when exploring these three statements, its overall effect is still substantive, approximately ranging from 18 to 25% of the variance in the levels of agreement. For example, with regard to the statement that police officers are knowledgeable about the law, in comparing the lowest value on political ideology to the highest value, there is a 1.83-point increase in the level of agreement on the 0–10-point agreement scale.

A much more substantial effect of political ideology is demonstrated on the levels of agreement for the other ten police interaction statements. In Figure 3, the predicted level of the agreement line sharply increases when moving from the lower values of political ideology to the higher values. For example, when comparing the lowest value of political ideology to the highest value, there is an increase of around four points on the 0–10-point agreement scale when predicting whether police officers provide the same quality of service to all citizens. In other words, approximately 40% of the variance in agreement that police officers provide the same quality of service to all citizens could be explained by the political ideology variable.

In Figure 4, we plot the predicted level of agreement with the 13 statements for individuals who identify as very liberal and very conservative to visually emphasize these striking differences in their attitudes about police interactions. As Figure 4 illustrates, the gaps between very liberal and very conservative respondents are large in terms of their level of agreement. Very liberal respondents are predicted to lean towards disagreeing with 11 out of the 13 statements; however, there is variance in the predicted level of agreement among these very liberal respondents. There are only two instances where very liberal respondents expressed a high level of agreement with the statements that police officers adequately protect citizens and are knowledgeable about the law, narrowing the gap with very conservative respondents.

In comparison, there is almost no variance in levels of agreement with the 13 statements among very conservative respondents. For almost every statement, very conservative respondents are predicted to have a level of agreement between 7.2 and 7.5 on the 0–10-point agreement scale. When accounting for the 95% confidence bounds, very conservative respondents statistically have the same level of agreement across all 13 statements. The results provide support for H2.

Discussion

Our findings demonstrate that although attitudes toward policing lean more positive than negative, these attitudes are closer to neutral than being pro-police. We also show an attitudinal division by political ideology across 13 aspects of police interactions, suggesting that there are two competing ideological opinions about policing. Whereas conservatives consistently illustrated unwavering strong support in all measured aspects of policing, liberals were not anchored to a singular concept of policing, as their attitudinal ratings of police interactions ranged from negative to positive. The concern of these negative attitudes toward police is political disengagement and alienation that can generate cynicism about their ability to engage with the political system when it comes to policing (Albrecht and Green, 1977; Easton, 1965; Williams and Maxwell, 2022).

One interpretation of this ideological pattern of ratings of police interactions is that liberals do not share a comparable reservoir of good will for the police as conservatives (Easton, 1965). Easton's (1965) postulation of diffuse support as a response to institutions, conceived as a stock of trust and favorable attitudes that people have towards the political system and community, helps explain the consistently high ratings of police interactions among conservatives. Diffuse support is a stable source of support for institutions and the political system and community, based on citizen loyalty and patriotism as well as the legitimacy they grant to the system. Easton (1965) regards one way to strengthen the bond between the system and its members is through indoctrination, a process of socialization that emphasizes ideology, which involves instilling legitimacy and a shared sense of welfare and belonging to the (political) community. These appeals to ingroup loyalty (and group solidarity) and patriotism, a moral concern that social psychology research has identified resonates more with conservatives than liberals, are responses to forming and maintaining coalitions (Easton, 1965; Graham et al., 2009; Haidt, 2012). Another response to a decline in support for a political system identified by Easton (1965, 1975) is specific support, which deals with people's evaluation of the actions and performance of authoritative institutions like the police; for this reason, it tends to fluctuate more than diffuse support. Thus, the (lower and) inconsistent ratings of police interactions among liberals indicating disapproval can undermine the level of diffuse support, leading to a disconnection from police and ultimately adverse effects on their legitimacy due to consistent frustration.

While political polarization exists on how policing services and officer demeanor and decorum were rated, conservatives and liberals do closely align on some aspects of policing. Specifically, there is an ideological agreement that police are knowledgeable about the law and adequately protect citizens from harm. These findings of political consensus are in tandem with research that has demonstrated that higher levels of (perceived) patrol and (positive) community engagement can improve relationships with police, including other dimensions, such as trust, satisfaction and procedural justice (Kochel, 2019; Koper et al., 2022; Ren et al., 2005; Schafer et al., 2003).

However, the similarities in attitudes toward policing among conservatives and liberals stop there, and future work should explore what policing efforts may shift these attitudes. For instance, from a policy standpoint, Ekins (2016) and Kochel (2019) state that confidence and trust in police can improve through transparency, such as external investigations of alleged police misconduct, publicly sharing information via social media and sharing granular statistics of stops and arrests. Even increasing the amount of information on police websites, which are notoriously intransparent and lacking basic information (Hansen et al., 2022), might increase public support for the police. Can good will be generated among liberals by implementing measures of data transparency while minimally affecting the generally favorable attitudes toward police among conservatives?

Another line of inquiry to identify is the durability of pro-police attitudes, especially among conservatives. Would attitudes toward policing continue its ideological division or shift with gender-specific statements? For instance, interactions with women officers expose drivers to fewer negative interactions while being as efficient in confiscating contraband as their male counterparts (Shoub et al., 2021). If attitudes shift for persons of either ideology, it would suggest that conservatives may carry firmly established attitudes that policing is a male-dominant role, which is not likely a shared attitude among liberals.

Our results also point to furthering research on residents from smaller locales. While there are mixed findings on whether attitudes toward police vary by community type, with rural residents tending to have greater positive attitudes about police performance than urban persons (Albrecht and Green, 1977; Zamble and Annesley, 1987), other research demonstrated relatively minimal to no rural-suburban-urban differences in their opinion of police (Brown and Benedict, 2002; Ekins, 2016; Huang and Vaughn, 1996; Weitzer and Tuch, 2002). As our results demonstrate, one plausible explanation for these mixed relationships is that rural residents held higher support for certain police interactions. Rural residents held greater favorable attitudes toward police use of force used in their communities than suburban and urban residents and the belief of a racial bias against Whites being treated less fairly by police than Blacks (Huang and Vaughn, 1996; Weitzer and Tuch, 2002). In a similar vein, sample results from a telephone survey conducted in 2010 and 2012 demonstrated that Houstonians held attitudinal distinctions between police in general and the police who serve their immediate neighborhood (Zhao and Ren, 2015). Alternatively, the predictive value of the study's rural variable may be enhanced by the attitudinal homogeneity of their residential environment versus the diversity among more urbanized communities. For instance, conservatives were proportionately more prominent among the study's rural respondents than their counterparts.

Finally, attitudes toward police and their services are a function of their contacts and interactions (Brown and Benedict, 2002; Cao and Wu, 2019; Dai et al., 2018; Graziano, 2018; Graziano and Gauthier, 2017; Huang and Vaughn, 1996; Ren et al., 2005; Schafer et al., 2003; Weitzer and Tuch, 2002, 2005; Wozniak et al., 2021; however, see Bolger et al., 2021; however, see Koper et al., 2022). Valuable insight may be garnered when assessing the effect of social distance to police on attitudes toward policing. For instance and analogously, Dai et al. (2018) found that public satisfaction with the police is associated with having a military family member. Scholars should also ascertain whether personal experiences with police have a differential effect on the perceived satisfaction with, confidence in and legitimacy of police by conservatives and liberals, as Wozniak et al. (2021) argued that group identities like political ideology may influence attitudes toward police to a greater degree than personal experiences.

Conclusion

These results help us understand why certain policies directed at policing may garner more support than others. The support of policing among conservatives is stable, unlike the fluctuating support by liberals, suggesting that their needs and demands are not being satisfied or met, thus undermining their diffuse support of policing. The services often provided by police and their general demeanor and decorum align with the contemporary conservative rather than the liberal. Although the results show that conservatives and liberals do not align closely on many aspects of police interactions, there is some attitudinal alignment, which can be leveraged in closing the gap about preferred policing styles.

Figures

Figure 1

Confidence in police by the public who indicated “quite a Lot/a great deal”

Figure 2

Distributions – attitudes of police interactions

Figure 3

The effect of political ideology on attitudes of police interactions

Figure 4

Ideology effect on police interactions – comparing very liberal to very conservative

Table 1

Descriptive statistics of attitudes of police interaction statements

Statement: Police officers …	Mean (SD)
… do not let personal biases interfere with their work	5.58 (2.92)
… generally do an adequate job policing neighborhoods	6.19 (2.53)
… could adequately protect citizens from harm	6.61 (2.34)
… make citizens feel at ease when they are around	5.92 (2.66)
… do not let emotions impact their ability to do their job	5.37 (2.94)
… are knowledgeable about the law	6.65 (2.40)
… do not bully members of the public	5.60 (2.78)
… are not dismissive of the public's concerns	5.72 (2.74)
… treat everyone with dignity	5.66 (2.79)
… respect individual citizens' rights	6.04 (2.68)
… are courteous to citizens they come into contact with	6.12 (2.50)
… provide the same quality of service to all citizens	5.68 (2.99)
… always act in accordance with the law	5.92 (2.78)

Note(s): The level of agreement ranged from 0 = “not at all” to 10 = “a great degree”; SD = standard deviation

Source(s): Table by authors

Table 2

Models predicting attitudes of police interactions

	No bias	Adequate job	Adequate protect	Ease citizens	No emotions	Knowledgeable	No bully	Not dismissive	Treat dignity	Respect rights	Courteous	Equal service	Act legally
Constant	0.485	2.089***	3.691***	2.058**	0.325	3.252***	1.061	1.297*	0.873	0.981	1.770**	0.175	0.459
	(0.673)	(0.601)	(0.582)	(0.656)	(0.686)	(0.646)	(0.661)	(0.636)	(0.598)	(0.630)	(0.624)	(0.708)	(0.651)
Age	0.003	0.023***	0.012	0.008	0.009	0.018**	0.020**	0.011	0.004	0.015*	0.017**	0.003	0.004
	(0.008)	(0.006)	(0.007)	(0.007)	(0.008)	(0.006)	(0.007)	(0.007)	(0.007)	(0.007)	(0.006)	(0.008)	(0.008)
Woman	−0.029	0.201	0.396**	0.228	0.169	0.595***	0.194	0.204	0.155	0.181	0.107	0.154	0.143
	(0.172)	(0.153)	(0.147)	(0.163)	(0.178)	(0.147)	(0.169)	(0.168)	(0.157)	(0.155)	(0.151)	(0.172)	(0.164)
White	0.697**	0.220	−0.241	0.184	0.358	0.070	0.507	0.331	0.466*	0.521*	0.384	0.784**	0.498
	(0.265)	(0.242)	(0.218)	(0.253)	(0.278)	(0.218)	(0.268)	(0.264)	(0.233)	(0.245)	(0.217)	(0.262)	(0.269)
Education	0.675***	0.400**	0.407**	0.502***	0.679***	0.350*	0.327*	0.488***	0.642***	0.608***	0.449***	0.716***	0.787***
	(0.141)	(0.123)	(0.125)	(0.134)	(0.139)	(0.137)	(0.139)	(0.133)	(0.133)	(0.136)	(0.131)	(0.149)	(0.137)
Income	−0.055	−0.031	−0.031	−0.026	−0.065	0.006	−0.028	−0.039	−0.064*	−0.038	−0.021	−0.074*	−0.019
	(0.034)	(0.028)	(0.028)	(0.031)	(0.034)	(0.028)	(0.031)	(0.032)	(0.029)	(0.028)	(0.027)	(0.031)	(0.003)
Suburban	−0.385	−0.073	−0.096	−0.392	−0.229	−0.261	−0.197	0.058	−0.363	−0.219	−0.120	−0.296	−0.166
	(0.210)	(0.190)	(0.185)	(0.201)	(0.220)	(0.188)	(0.204)	(0.201)	(0.192)	(0.187)	(0.179)	(0.213)	(0.205)
Rural	0.555**	0.489*	0.513**	0.277	0.571**	0.083	0.652**	0.485*	0.246	0.475*	0.295	0.474*	0.606**
	(0.214)	(0.191)	(0.179)	(0.199)	(0.217)	(0.172)	(0.210)	(0.212)	(0.194)	(0.189)	(0.188)	(0.207)	(0.202)
Party ID -	−0.576*	−0.368	−0.521*	−0.773**	−0.755**	−0.393	−0.601*	−0.357	−0.762**	−0.534*	−0.511*	−0.361	−0.584*
Independent	(0.239)	(0.222)	(0.221)	(0.244)	(0.255)	(0.220)	(0.258)	(0.249)	(0.242)	(0.239)	(0.227)	(0.249)	(0.243)
Party ID -	−0.613**	0.315	0.159	−0.091	−0.114	0.344*	−0.181	0.113	−0.031	0.345	−0.010	0.325	0.084
Republican	(0.201)	(0.191)	(0.169)	(0.199)	(0.216)	(0.169)	(0.204)	(0.204)	(0.191)	(0.189)	(0.185)	(0.207)	(0.192)
Political	0.406***	0.249***	0.179***	0.301***	0.374***	0.183***	0.375***	0.320***	0.393***	0.311***	0.303***	0.396***	0.326***
Ideology	(0.036)	(0.033)	(0.031)	(0.035)	(0.036)	(0.030)	(0.034)	(0.035)	(0.035)	(0.034)	(0.031)	(0.036)	(0.035)
Observations	984	984	984	984	984	984	984	984	984	984	984	984	984
Log Likelihood	−2,324	−2,229	−2,186	−2,270	−2,353	−2,200	−2,305	−2,308	−2,251	−2,245	−2,206	−2,331	−2,286
R2	0.254	0.165	0.117	0.182	0.225	0.127	0.223	0.179	0.273	0.233	0.191	0.276	0.235

Note(s): *p < 0.05. **p < 0.01. ***p < 0.001. Urban is the reference category for the residential environment of the respondent. Democrat is the reference category for the party identification of the respondent

Source(s): Table by authors

Table A1

Descriptive statistics – independent variables

Variable	Min	Median	Mean	Max	SD
Age	19	36	39.02	81	11.58
Education	1	4	3.84	5	0.74
Income	1	5	5.81	12	2.72
Political ideology	0	6	5.71	10	3.11

Variable
	Rural	Sub	Urb
Residential environment	31.19%	34.00%	34.80%
	Dem	Ind	Rep
Party ID	60.18%	15.55%	24.27%

Variable	0	1
Gender	55.77%	44.23%
Race	14.81%	85.19%

Source(s): Appendix by authors

Notes

1. The value “5 – neither agree nor disagree” was displayed on the scales so that respondents could easily discern negative sentiment from positive sentiment in value selection.

2. Less than 4.6 percent of respondents selected the same value across the 13 statements, which indicates that “straight-lining” was not an issue. The variance in individual selections across the statements indicates the quality of the data and appropriateness for further analysis.

Declaration of conflicting interests: The authors report that there are no competing interests to declare.

Funding: This work was supported by the Field Impact Grant from the College of Criminal Justice and Criminology at Sam Houston State University.

Appendix 1 Variable coding and descriptive statistics

Age – continuous measure, respondent's age at the time of the survey.

Gender – binary measure, 0 = men; 1 = women.

Race – binary measure, 0 = nonwhite; 1 = White.

Education – continuous measure, 1 = less than high school; 2 = high school/GED; 3 = some college; 4 = bachelor's degree; 5 = graduate/professional degree.

Income – continuous measure, 1 = Under $10,000; 2 = $10,001 - $20,000; 3 = $20,001 - $30,000; 4 = $30,001 - $40,000; 5 = $40,001 - $50,000; 6 = $50,001 - $60,000; 7 = $60,001 - $70,000; 8 = $70,001 - $80,000; 9 = $80,001 - $90,000; 10 = $90,001 - $100,000; 11 = $100,001 - $150,000; 12 = more than $150,000.

Residential Environment – 3-category nominal measure, Urban; Suburban; Rural.

Party Identification – 3-category nominal measure (created from 7-point measure with leaners coded as partisans), Democratic; Independent; Republican.

Political Ideology – continuous scale from 0 = very liberal to 10 = very conservative.

Attitudes of Police – continuous measure, respondents were provided 13 interaction statements and asked about their level of agreement. The level of agreement consisted of self-placement on a Likert-scale from 0 = not at all to 10 = a great degree.