Differences in citizen perceptions of interactions with police officers

Introduction

Citizen’s perceptions of the police are important for determining whether someone sees the police in a positive or negative light. Over the past several decades, many studies have been conducted on different aspects of citizen perceptions of the police to determine what makes up these perceptions and opinions on the police, looking at biological and socioeconomic factors civilians can have for themselves, and how the police officers themselves might influence these opinions. Studies have been conducted that have found that particular perceptions play an important role in how citizens see the police, such as perceptions that a person is being treated fairly by the police (Sunshine and Tyler, 2003; Wentz and Schlimgen, 2012), the perception that you can trust the police (Sunshine and Tyler, 2003; Tyler, 2005), the perception that the police are concerned about community problems (Goldsmith, 2005; Hinds, 2009), and the perception that the police are respectful of citizens (Tyler, 1997; Tyler and Wakslak, 2004). These perceptions have been found to be influenced by what the police do when they directly interact with citizens, yet these perceptions can also be influenced by demographic factors of the citizens themselves. The main demographic variables that have been shown to have noticeable effects on perceptions of the police have been the race of a civilian (Lai and Zhao, 2010; Weitzer and Tuch, 2005), household income level (Frank et al., 2005; MacDonald and Stokes, 2006), level of education (Chu et al., 2005; Reisig and Parks, 2006) and employment

status (MacDonald and Stokes, 2006; Wu, 2014). Crime rates have also been known to affect perceptions of the police as well (Schafer et al., 2003; Wu et al., 2009). Two research questions will be answered by this study, the first being: Do citizens from an anonymous city in California differ in their perceptions of interactions with police based on demographics? The second research question is: Do differences in neighbourhood crime rates affect different perceptions of the police? The hypothesis based on the first research question is: Demographic differences are associated with different perceptions of interactions with the city police. The hypothesis based on the second research question is: Neighbourhood crime rates affect citizen perceptions of city police.

While race’s effect on perceptions of the police has been documented in numerous studies (Lai and Zhao, 2010; Weitzer and Tuch, 2005; Wu, 2014), these and other studies have not been conducted in the mid-sized city we collected data from in California, which will be the focus of our discussion. It is important to know how race plays a role in perceptions of the police as it can give researchers a starting point on how to change perceptions. Depending on the results of this study, researchers can help create new policies or programmes for police departments to help their officers be more mindful in their interactions with citizens of different races, especially in heterogeneous neighbourhoods. Since you cannot change a person’s race, knowing how a person’s race can correlate with their perceptions of the police lets us know how race interacts with other determinants of perceptions of the police, and how we might better control for race in future studies.

A citizen’s household income level and employment status are important factors to study for their effect on perceptions of the police because they give potential insight on how citizens are living their lives. Those with low income levels and who are unemployed may be living in more socially disorganized areas of the cities they are living in, and see the cops more often, probably in more negative encounters. Knowing how these two factors affect perceptions can allow programmes targeted at those with low incomes or who are unemployed to better get to know their police officers, so they can be more confident in the police and lessen one stressor in their stressful lives. For citizens who have an employment status that stops them from being employed, such as being retired or homemaker, knowing how these citizens perceive the police can allow researchers to begin to study how these groups come to form their perceptions of the police if they are not in the workforce. Studying these groups of citizens will let researchers know if they have significantly different perceptions of the police since they are not employed. Knowing how citizens with high household incomes might differ from those with low household incomes can allow researchers to study what these different groups might have in common about their perceptions of the police, and how lessons from one group can be applied to the other to potentially create more positive perceptions of the police.

There are several reasons why studying the effects that educational achievement have on perceptions of police is important. Studying this factor will add to past research on the topic (Schafer et al., 2003; Sunshine and Tyler, 2003; Thompson and Lee, 2004; Weitzer and Tuch, 2002), which has mixed results, with the results likely to tip this factor into playing either a positive or a negative role when it comes to determining police legitimacy. This will further spur research on the subject to see how this independent factor can be accommodated or worked with to improve a person’s perception of the police so they are more likely to view the police as legitimate. Having resolved any ambiguity about education’s effects on perceptions of the police, it will allow for researchers to properly account for education in future studies and allow future researchers to consider how other uncertain variables might influence perceptions of the police now that education has a known, stable effect. If education has a positive effect on perceptions of the police, then research can be done to see what specifically makes having a higher education lead to more positive perceptions of the police. When this factor is determined, it can then be extracted and tried on its own outside an educational setting to see if it will improve perceptions of the police in those who do not have a high level of education. If police departments know the effects of education, they may be able to change their policing tactics based on demographic data of certain neighbourhoods, to deal more effectively with high or low clusters of highly educated people in their cities. Since the sample for this study will be from a mid-sized city’s anonymous citizens, further research can be done in different cities in California to see how they are similar and different to city in terms of population and perceptions of the police.

Studying crime rates’ effects on perceptions of interactions with the police will let us know if actual crime rates make a difference in how people see the police. Studies (Bradford et al., 2009; MacDonald et al., 2007) have shown that signs of physical and social disorder in a neighbourhood are the most likely neighbourhood-level variables to affect how citizens perceive the police. Crime rates have also been included in past studies (Reisig and Parks, 2000; Wu et al., 2009), but these usually look at how one particular crime rate type affects perceptions; this study will look at the overall number of crimes reported in a police beat. Knowing how crime rates affect perceptions can be useful information to let officers know how they might be received by citizens in these neighbourhoods when they get calls for services there, and let the officers prepare accordingly. How overall crimes affect perceptions can also let the police department know where they may need to allocate resources that can be used to help increase positive perceptions of the police. This could include providing more outreach programmes to certain police beats or even redrawing certain police beats to make sure they get the appropriate amount of policing and police resources to handle the crimes being reported. Overall reported crime rates can let us know which police beats are more actively calling the police for services, which can lead to further research into how their perceptions of police interact with their willingness to call the police for services.

Literature review

Since police legitimacy and trust in the police is an opinion held by individual people, it makes sense to look at these ideas from a psychological perspective. An article from Tyler (1997) tries to frame the obligation to follow authorities and finding authorities to be legitimate in a psychological context. To determine the psychological context, several different studies were used, the first and second studies were on conflicts students had with their parents or professors, respectively, the third and fourth studies asked participants about recent workplace conflicts they had had with their supervisors, and the fifth and sixth studies asked those sampled about their thoughts on different legal authorities.

Two psychological models were found to influence whether a person defers to authorities and whether they see authorities as legitimate: an identity-based relational model, which had the biggest impact on views of legitimacy, and the resource-based instrumental model, which was found to play a smaller but still significant role in forming views on legitimacy. The relational model shows that people willingly defer to authorities, such as the police, and find them legitimate if people believe they are respected and valued by the authorities, are being treated in a neutral way, and can trust the motives behind how the authorities act. The instrumental model says that people who have more of a stake in the authority being asked about are more likely to see the authority as legitimate and defer to their judgments. While this was not found in many of the studies used in the article, Tyler suggests that the instrumental model can still influence views on legitimacy, just not in the same way as the relational model. People respond best when they believe the police value them, so people will trust and find the police legitimate if they are respected by the police.

Another psychological perspective that can help explain trust in the police is a Durkheimian perspective. Jackson and Sunshine (2007) look at what shapes confidence and trust in the police in a country outside of the United States of America. Conducted in a rural area of England in 2001, 1,023 out of 5,906 questionnaires were sent back to the researchers, with half the questionnaires asking questions about the respondent’s beliefs about social change, authoritarianism, and law and order, which was used in a different paper. The other half of the questionnaires asked questions about the respondent’s beliefs about the police, which is what this paper reports on. The most influential determinants of confidence and trust in the police were a sense of social cohesion and order in the community of the respondent. Respondents also trust the police more when they perceive them as protecting moral structures in the community, not necessarily when they are protecting the community from crime. Likewise, fear of crime and signs of physical and social disorder did not have major impacts on citizens’ trust and confidence in the police, especially when controlling for social cohesion. The results of this study showed that people followed a neo-Durkheimian sociological model of deviance, trusting the police more when they felt that the police were upholding the morals of the community rather than when they were fighting crime.

While it is important to know the psychology behind how people view police legitimacy, it is also important to know what possibly influences those views. Tyler (2001) found that trust and confidence in the police and legal institutions are shaped in the eyes of the public. The studies used for this article were a Chicago study on support for the police and courts in the city, an Oakland study on police performance, a national study asking people about their opinions on the court system, and a statewide Indiana study on the public’s evaluation of the state court system.

Across all four studies, the largest determinant of public opinion on the courts and police was how procedurally fair their actions were, not how fast their decisions were made or what the outcomes of those decisions were. The more fairly a person was perceived as being treated, the more likely they were to have trust and confidence in the justice system. This was particularly true in minority groups that were a part of the different studies, as they were consistently found to be concerned about unfair treatment by the justice system. How the police and courts treat people was also more of a concern for people who had personal experiences with the courts or police than for people who had no experience with these groups. The people in the four studies were also less influenced by outcomes of decisions when it comes to shaping their opinions on the police and courts. Generally, both majority and minority groups were more inclined to positive views of the justice system when they felt they were treated fairly.

While most studies conducted about police legitimacy get their samples from large cities, there are some that got their samples from more-rural towns or cities. Holmes et al. (2017) took a look into the overlooked realm of rural communities and their perceptions of the officers that police them. The communities used for the study were 99 small and rural communities in the state of Iowa. Surveys were sent to households in each community. 9,962 surveys were returned, making a return rate of 67% for this study. The survey results showed that in general, rural communities were more likely to trust their local police than their county sheriffs, which the authors believe to be caused by the smallness of the communities and the police departments in them, making it easier to know the police in their towns and have them react faster to problems. When looking at the demographic control variables used in the surveys, there were only two demographics that had much of an effect on a person’s view of the police, with older people trusting the police more, and Republicans more likely to perceive the police more positively. The results show that rural communities are more likely to trust the local police of their town rather than their county. They are more likely to know their local police since they see them more often, with only age and political views having a great impact on these results.

Social class is one of the major topics of research when it comes to trust in the police, especially in the United States. MacDonald and Stokes (2006) tested the effect of community beliefs, race and socioeconomic status on a person’s trust in the police. The data used in the study came from a 2001 random phone survey in the United States that got a total of 3,003 respondents, with an oversampling of Black respondents, with 500 of the total respondents being Black. While social trust and civic engagement within the community did not significantly alter the relationship between race and trust in the police, the social capital created by these two ideas was a strong predictor that someone would trust the police.

Respondents with higher education were also found to find their local police to be trustworthy. However, it was found that social capital was a much more influential factor in trusting the police than education, enough so that education was found to be an insignificant factor by comparison. So while race was a big predictor of trust in the police, the social capital of a person was also a significant predictor of trust in the police.

MacDonald and Stokes would also write another paper dealing with race and perceptions of the police just one year after the one they published in 2006. MacDonald et al. (2007) found out how race and neighbourhood contexts affected the perceptions of injustice in citizens of Cincinnati, Ohio. The experiment was conducted in 2004 in Cincinnati, Ohio, where surveys were sent to 3,000 people throughout the city, asking about various demographic and neighbourhood variables, and whether they perceived that the police were using racially biased practices. When controlling for similar neighbourhood contexts, Blacks were more likely to perceive injustices being done by the police than Whites. Socioeconomic status was another demographic variable that had significant effects on perceptions of police injustices, derived from education, employment and household income, with those of a higher socioeconomic status being more likely to perceive injustice in the practices of the police. As for neighbourhood contexts, the more neighbourhood disorder there was in a respondent’s neighbourhood, such as signs of physical and social disorder, the more likely the respondent was to think that the police were using racially based practices. Socioeconomic status relating to perceptions of the police shows that those who are better off are the people most likely to be critical of the police, rather than those with a lower socioeconomic status.

Research on police legitimacy also finds differences between minority groups in the US. Weitzer and Tuch (2005) found certain factors influenced satisfaction with the police in whites, Hispanics and African Americans. Conducted between October and December of 2002, 1,792 telephone surveys were collected from Whites, Hispanics and African Americans across the United States from cities with more than 100,000 people living in them. When it came to demographics, Hispanics and African Americans were more likely to be dissatisfied with the police than Whites were, but when demographic variables like race were controlled for, Whites and Hispanics had similar satisfaction levels with the police, while African Americans still showed dissatisfaction. Factors that did determine satisfaction with the police include the citizen’s perceived effectiveness of the police when it came to dealing with crime in their neighbourhood, perceived safety in one’s own neighbourhood, and – for Whites and African Americans – community policing in their neighbourhood generally increased satisfaction with the police. Income had an effect on Whites and African Americans; Whites with higher income were more likely to be dissatisfied with the police, with African Americans of lower income more likely to be satisfied with the police; both effects were very weak, however.

Based on a study by Lawrence (2000), mass media was found to decrease satisfaction in the police in African Americans, which is believed to not affect Whites and Hispanics as much because media tends to report on police conduct with African Americans instead of Whites or Hispanics. When racial aspects are considered, Hispanics are comparable to African Americans when it comes to trusting the police, but when race is not considered, Hispanics have the same level of satisfaction with the police as Whites.

In more recent studies, Hispanics have begun to be added to studies on race and the police to help better understand how different races perceive the police. Lai and Zhao (2010) seek to further our understanding of how citizens view the police, with emphasis on how Hispanic citizens view the police. The study was conducted in Houston, Texas, via random telephone surveys in 2008, gaining a total of 756 useable surveys. The African Americans’ surveys for this study had a generally negative view towards the police and were the least likely group to trust the police; while Hispanics still generally held negative views of the police, these attitudes were less negative than African American views. Besides race, other predictive variables for general views on the police were gender, age and victimisation, while variables that were predictive of trust in the police were race, satisfaction with the police services, and how much concentrated disadvantage a person had, such as being unemployed or living in a poor neighbourhood, which was predictive of distrusting the police more than citizens with less concentrated disadvantages. This study further confirms that while Hispanics still distrust the police, they generally don’t distrust them as much as African Americans do.

Then on occasion you’ll have studies on how four or more races and their views of police legitimacy. Wu (2014) discusses how various races and ethnicities in the United States perceive police activities and what influences these perceptions. Surveys and interviews were used to collect data in Seattle, Washington. Random sampling, ethnic oversampling and replication sampling were used to make sure each race and ethnicity was properly represented. The results of the study showed that on many variables, Blacks were more likely to have negative perceptions of police activities and attitudes, with Hispanics being less likely to have negative perceptions of the police than Blacks, followed by Asians, with Whites being least likely to have negative perceptions of the police. Blacks, Hispanics and Asians, when faced with police harassment or racial profiling, either against themselves or others, were more likely to have negative impressions of the police than Whites were on the same topics. Wu also found that across all racial and ethnic groups, having an education, specifically having a college degree, along with being employed, meant that a person was more likely to notice police bias and racial profiling. The results show that Hispanics and Asians are in the middle of trusting and distrusting the police when it comes to race, with the more educated in all groups likely to notice negative actions of the police.

Many studies will also not limit themselves to demographics of the sample when determining police legitimacy. Bridenball and Jesilow (2008) investigated which variables influence a person’s views on the police and which are the most prominent in forming these views. To determine what these influences are and their strength, interviews were conducted with citizens in Santa Ana, California, with a total of 534 citizens interviewed for the paper. The strongest factor that shaped a person’s views on the police was their age. People become more likely to view the police favourably as they age. Demographics such as socioeconomic status (derived from education level and a person’s occupation), a person’s gender and their ethnicity did not have that much influence based on the data collected. Having personal contact with the police was found to produce both positive and negative views on the police, as they may have liked certain aspects of police conduct when they interacted while finding things they disliked about the interaction as well. One other important factor that influenced views on the police was exposure to visible, persistent signs of neighbourhood disorder and crime like graffiti and prostitution. These signs generally lowered a person’s perception of the police, more than serious, though less common, factors like violent crimes, which also lowered a person’s perception of the police. The amount of local crime in the area was found to have little influence on citizens’ perceptions of the police. While age was a key factor in determining police legitimacy, personal experience with the police was also an important factor.

Another type of study on this topic compared demographics to environmental factors to see which is a stronger predictor of trust in the police. Wu et al. (2009) conducted a study on how social class and race affect a person’s satisfaction with the police. The source of data comes from 2,000 interviews done in two Kentucky towns, Louisville and Lexington, where the 1,963 interviewed people were asked questions on how they felt about various aspects of the police along with questions to determine the demographics of the person interviewed. Personal experiences and demographic details were found to influence a person’s satisfaction with the police when looking at individual aspects, while the context of the neighbourhood a person lives in primarily determines satisfaction with the police when looking at multiple aspects of the police. An example of this would be that the variable of class – a combination of education, employment and being a home owner – was found to influence people at individual contexts, but had very little influence when looking at neighbourhood contexts.

If a person had a negative personal experience with the police, it was the most significant predictor of shaping their satisfaction with the police, seeing them negatively. Age was the most predictive variable of the demographic data, with older people being more satisfied with police services than younger people. Neighbourhood context, for this study meaning racial composition, social class, violent crime rates and mobility of residents in the neighbourhood, was found to be most influential with African American interviewees in shaping satisfaction with the police. An example of the influence of neighbourhood context is that living in a neighbourhood with a high violent crime rate makes a person more likely to be dissatisfied with the police. As this study showed, different factors can be more important when looking at specific levels of satisfaction.

Studies that look predominantly at how the actions of the police influence satisfaction with the police are also common when looking at perceptions of interactions with police officers. Frank et al. (2005) conducted a study to determine why citizens hold the beliefs they do about the police where they live. The sample for the study comes from Cincinnati, Ohio between November, 1997 and January, 1998. The total amount of useable surveys was 613. The surveys studied whether citizens were satisfied with the police and why they were at that level of satisfaction with the police. Most citizens who took the survey had a generally positive attitude for the police, with how positive outcome on a variety of factors, like race, with African Americans holding less positive attitudes than whites. Those with higher income and education were more positive in their attitudes about the police. As for why citizens held these attitudes toward the police, a majority based their attitudes on specific traits a police officer had and the response time of an encounter, with another group in the sample basing their attitudes on outside factors not related to any encounter they had with the police. A third group formed their attitudes based on what happened in their interactions with police, usually excluding the outcomes of these interactions. This study shows what kind of actions people base their satisfaction with police services on, along with the finding that the sample had generally positive impressions of the police.

Sometimes the police officers in these studies do not have to do anything for citizens to change their views on police legitimacy. Tyler and Wakslak (2004) wrote an article that looked at a handful of studies to determine what effect racial profiling has on public support for the police. Four studies were used for this article, the first coming from California on stops by the police and potential profiling factors from them, the second done in New York City about perceptions of racial profiling among young adults. The third study was a survey about the NYPD and how potential profiling affected the opinions of citizens in New York City on the police, and the fourth and final study was a replication of the third. After analysing the results of each survey, it was found that if a person believes their encounter with the police had elements of racial profiling in it, it will negatively affect their views of police legitimacy, predictor of whether racial profiling had actually occurred or if they believe it occurs where they live. Since the perceptions of profiling might not match the reality of what is actually happening, those in the study samples who felt they were treated with fairness and respect by the police were less likely to believe they had been racially profiled in their encounter with police.

Even when race is not a part of the study, inaction or neglect on the part of the police can also affect people’s perceptions of them. Goldsmith (2005) wrote a research paper on the nature of trust and distrust in the police, and what contributes to people trusting or distrusting the police. The paper starts with overviews on trust in general and trust in the government, then goes into why people may not trust the police. Shortcomings in how the police do their jobs, how well they uphold the law and how suspicious the police can be of people were key variables. Finally, the study examined how the police can build their trustworthiness. If people see the police as selectively enforcing laws or disregarding laws and rights entirely, citizen perceptions of distrust emerge towards law enforcement and the broader justice system. Also, an overly suspicious and cynical police creates an us-versus-them mentality between citizens and police, which may cultivate a climate for hostility and resentment towards the authorities.

Several key performance issues can also influence trust in the police. Being neglectful or indifferent to the problems and concerns in the communities they police, discrimination against groups of people, corruption in officers and excessive use of force by officers cause trust issues. One of the suggested methods to help build or rebuild trust in the police is for the police to be fair in their encounters with citizens and to be open about how the police operate and make their decisions, so citizens can know if the police are acting as they should and what actions can be taken if they are not. Another method for increasing trust in the police would be to curtail use of force by the police, which not only means decreasing incidents of police brutality and uses of excessive force, but also decreasing uses of force that lead to coercion of citizens. Police must take care to at least seem like they care about the communities they patrol if they want the people in these communities to trust them.

Fairness is one of the recurrent findings in studies on how police legitimacy is determined. Tyler and Folger (1980) did a study that tests whether the procedures used by the police in their encounters with citizens have an effect on citizen satisfaction with the police. The sample for this experiment consisted of citizens in the town of Evanston, Illinois in 1979, who were interviewed over the phone about instances where they had made calls to the police or had been stopped by the police, and what their experiences in these situations were like. The interviews found that fairness of an experience with the police determined both satisfaction with that specific experience and the overall satisfaction with the police, while the outcome of these experiences did not have an effect on satisfaction with the police held by the interviewee. Those who thought they were being treated fairly by the police were more likely to be satisfied with that experience and the overall job the police were doing in their city. Those who thought they were not treated fairly were more dissatisfied with the police in their encounters and were dissatisfied with the police work in their city. Whether the interviewee had called the police or had been stopped by the police did not matter; as long as they felt they had been treated fairly by the police in their encounter, the findings of this study held true. If a person believed they were being treated fairly by the police, they were more satisfied with the job police were doing.

Finding how police and citizens differ in their views and priorities can help each group trust each other more. Stein and Griffith (2017) compare citizens’ perceptions and police perceptions of the high crime neighbourhoods they live and work in. To get the data for the research, survey and census data was collected from three different Midwestern towns in the United States in these town’s high-crime neighbourhoods. Citizen perception of their high-crime neighbourhoods were consistent across the three neighbourhoods, with low collective efficacy and heightened fear of crime, while police perception of these neighbourhoods differed, specifically perceiving a neighbourhood that was primarily white and had a community policing programme as being a nicer neighbourhood. Stein and Griffith suggest that what an officer sees in a neighbourhood is the primary influence on how they think that neighbourhood is doing, rather than looking at any problems occurring below the surface, which explains the difference between police and citizen perceptions.

On occasion, you can gain insight on police legitimacy by analysing what the officers themselves think influences legitimacy. Vera Sanchez and Rosenbaum (2011) wrote an article on how police officers view Latino and African American communities they police. Forty police officers were interviewed by the researchers, split between ten officers who patrolled four different neighbourhoods that were predominantly either African American or Latino, about their personal observations and beliefs of the minority neighbourhoods they policed. Police officers interviewed generally believed that they were unwelcome in these neighbourhoods, with their opinions of the African American neighbourhoods based more on morals rather than race, except in the case where African Americans were the minority in a Latino community. From their experiences, officers believed that the biggest predictor of a person’s attitude toward the police was their age, finding that the older people in a neighbourhood were more accepting of the police, while every neighbourhood in the study had some tension between young people and the police. The interviews continue to support the idea that Latinos trust the police more than African Americans, while also suggesting that age plays an important role in developing trust in the police.

While it is important for a person to be treated fairly by the police, how the police treat other people is also important in forming police legitimacy. Wentz and Schlimgen (2012) is a research paper that looks at citizens’ perceptions of how police officers interact with other people in their neighbourhoods, among other variables, to determine overall perception of the police. Carried out in an anonymous Midwestern city in the United States, the study collected 426 surveys from households in this city to use as data, asking questions about various signs of social disorder, crimes in their neighbourhood and how safe they felt in their neighbourhood. How a citizen perceived how officers treated other people in their neighbourhoods was found to be the most significant factor in how a person viewed police services overall; this factor was found to be more of an influence than a citizen’s personal contact with the police or the amount of social disorder and crime in the neighbourhood they lived in. As part of the demographic variables used in the survey, the more educated a citizen was, the more likely they were to find that the police were responsive to the concerns and needs of community. While other studies point out that being treated personally with fairness is important, this study also makes clear that seeing other people treated fairly is important for legitimising the police, with additional evidence pointing to how important higher education is to legitimise the police in the eyes of a citizen.

One factor that does not appear often when looking at how trust in the police is influenced, but is still a major factor, is quality of life in a neighbourhood. Reisig and Parks (2000) conducted a study to see if experience with the police, a person’s quality of life and their experiences in the neighbourhood can accurately predict a person’s satisfaction with police. The data used for this study was collected beforehand by the Project on Policing Neighborhoods, who used St Petersburg, Florida and Indianapolis, Indiana to randomly call adults to take part in a telephone interview about policing. Looking through the data collected, Reisig and Parks found that the variables that make up quality of life, such as fear of crime and signs of social and physical disorder, were the most predictive of a citizen’s level of satisfaction with the police, followed by personal experiences with the police. From the demographics collected, the amount of education a citizen had related directly to their satisfaction with their police; the more educated a person was, the more likely they were to be satisfied with the police, with the exception of people interviewed who only had some college education, who did not follow this relationship between education and satisfaction with the police. When crime rates of a neighbourhood is included in the analysis, particularly homicide rates, people in these high crime areas are less likely to be satisfied with the police than those living in neighbourhoods with low rates of crimes and homicides. This experiment shows that the quality of life a person has as regards the neighbourhood they live in is an important factor in explaining their trust in police, along with how education effects levels of trust as well.

Most studies on trust in the police and police legitimacy look at citizens of the country they are conducted in, but it can be beneficial to see how immigrants to a country trust the police. Chu et al. (2005) is a study that examines how Chinese immigrants to New York City view the police, with a focus on what concerns these immigrants have when they interact with the police. With help from a pretest focus group, 151 questionnaires were sent out to Chinese immigrants in New York City in July and August of 2004. Chu et al. found that merely interacting with the police did not increase an immigrant’s opinion on the police, whereas getting services from the police helped increase satisfaction with the police in the eyes of immigrants, as did seeing the police call for assistance. If an immigrant was fearful of crime in their neighbourhood, they were found to be more likely to view the police as ineffective. Education was found to be the greatest demographic variable that influenced an immigrant’s view of the police. Immigrants who had higher education were more likely to see the police in a positive way, when compared to immigrants with lower educational achievement. For these Chinese immigrants, simply getting the police to come to their calls for service was what increased their trust in the police, with those who had a higher level of education more easily trusting the police, boosting the trust gained from police responding to their calls for service.

The effects of police misconduct can lower trust in the police, but sometimes other factors can influence how misconduct is perceived. Weitzer et al. (2008) researched how and why views on police conduct might be different between races of people. To find the answers to their research questions, the study was conducted in Washington, DC, where a majority of the police and citizens are African American. They conducted surveys between 1998 and 2000, to which there were a total of 2,731 respondents. It was found that Blacks were more likely to perceive police actions as misconduct than Whites. It was found that race was not the only reason this difference existed. People living in areas with community policing perceived less police misconduct than those who did not experience community policing. Social class was another important factor in perceiving police misconduct, with the lower social classes believing police misconduct was prevalent and those with higher education not seeing police misconduct as prevalent or a problem in the areas they lived. Here the results were that community policing decreases perceived police misconduct, along with being of a higher social class and higher education.

Many studies will look at multiple races for their sample, sometimes it can be beneficial to look at one race or ethnicity to solely get their perspective on a topic. Priest and Carter (1999) used an African American sample to look into how African Americans evaluate police performance and their attitudes towards the police. Using a telephone survey, the researchers called African American households in Charlotte, North Carolina in June and July of 1996 to obtain their sample, which resulted in 338 complete interviews used for the paper. The largest factor that contributed to the attitudes held by African Americans in the sample was the context of the neighbourhood they lived in, composed of safety in their neighbourhood, the state of their neighbourhood overall and police response times. The more positive these three aspects were rated, the more positive attitudes were of the police. Education and age were also predictors of attitudes and evaluations of the police, with the better educated and the older people in the sample more likely to view the police positively. When it comes to victimisation, whether a person considers themselves to be a victim of primarily lesser offences can have a significant impact on their evaluations of the police, with victims of less serious offences less likely to have their victimisation affect their views on the police. In this study, neighbourhood context and perceptions of victimisation were key components on how African Americans view police performance.

Misconduct by the police can also affect how people trust the police, which has a tendency to affect racial groups differently. Weitzer and Tuch (2004) wrote a research article on how Whites, Hispanics and African Americans view misconduct by the police. The collection of the data occurred from October to December of 2002 on a national scale, where 1,792 Whites, Hispanics and African Americans from cities with populations over 100,000 answered surveys asking questions relating to police use of verbal abuse, police corruption, police use of excessive force, and unwarranted stops by the police. The results from the surveys showed that Whites were the least likely to see various kinds of police misconduct as prevalent, Hispanics were more likely to view police misconduct as prevalent than Whites were, but usually saw it as less prevalent than African Americans, who were most likely to view police misconduct as prevalent and widespread. From the demographic variables collected besides race, it was found that education had a weak and insignificant effect when it came to thinking police misconduct was widespread, the effect being that the more educated a Hispanic person was, the less likely they were to see misconduct as prevalent. The results show once again that Hispanics are in the middle of Whites and African Americans when it comes to how they feel about the police, with the added result that a Hispanic person with higher education does not see misconduct by the police as prevalent.

Prior encounters with the police are also important when people consider present encounters with the police when determining legitimacy. Braga et al. (2014) sought to know how context plays a role in how people assess wrongdoing by the police and how it might affect views on satisfaction with police. The experiment used a survey research firm to get their sample of 1,361 participants from 15 US cities, who were interviewed after watching videos of police encounters with citizens and asked if they believed that what the officers did in the videos was wrong, various questions about satisfaction with the government and the police, and about thoughts on their neighbourhoods and certain beliefs they held. What Braga et al. found was that a person’s prior experience with the police was a bigger determinant of how they felt about the police than the context of the videos they watched; if a person has a negative view on the police because of a prior experience, seeing a good interaction between police and a citizen will not automatically change their views on the police. Besides prior experience, the level of education a person had also shaped their satisfaction with police, with those interviewed who had higher levels of education being more critical of the police in the videos they watched, and less likely to be satisfied with the police. This experiment shows that one encounter with the police, positive or negative, will not drastically change how people feel about the police based on their prior experiences with police, and showing that the more educated a person is, the more likely they are to be critical of the police.

The use of force has become a major factor in how people see police legitimacy, but how much force is too much differs between people. Thompson and Lee (2004) sought to find out which groups are likely to agree with police use of force, and what factors go into those judgements. The researchers got a sample for their study using a national survey in 1998, which asked questions about different situations where a police officer used force, demographics and societal views, getting a total of 498 as the sample size. The groups that were less likely to approve of police use of force, usually regardless of context, were racial minorities and women. Researchers believe this to be caused by police brutality used on minorities and a general distrust in the police for minorities, while the explanation for women is that they do not believe force is necessary to keep social order. Several group factors had no effect on whether a person would support the use of force, such as having a higher education, their political leanings, and fear of crime. With regards to income, those who had low income were more likely to support police use of force than those who had high levels of income. The results of the study overall show that the context of the situations police find themselves in are important to citizens in whether they support the use of force by police, with context also playing a larger role in forming broader attitudes towards the police.

Police legitimacy and support of the police are often used to determine how much a person trusts the police, but these two ideas are not independent from each other. Sunshine and Tyler (2003) conducted research to determine how important police legitimacy is in shaping how much the public supports the police, and what the police do that helps shapes police legitimacy. Two samples were taken for this study, both from New York City, one coming from before 11 September 2001, with a sample size of 586 answering questionnaires, and the other sample coming after 11 September 2001, where the sample size was 1,422 answering surveys. From the results of both samples, it was found that police legitimacy is an important factor in shaping public support for the police, a predictor of police relations with the community, and that the most significant influence on shaping police legitimacy are officers practising procedural justice, treating each civilian fairly and justly when interacting with them. Sunshine and Tyler also found that those in the sample with higher education were more likely to comply with laws and police orders, but were also less likely to see the police as legitimate. From this study, police legitimacy informs support for the police, with one of the key components of legitimacy being procedural justice, another important factor in legitimacy was the education of a person, with more educated individuals finding the police an illegitimate authority.

Racial profiling can be one of the biggest factors that play into dissatisfaction and distrust of the police. Weitzer and Tuch (2002) looked at views on racial profiling, with an emphasis on how races perceive racial profiling. The sample for this study was a national random phone survey conducted in the United States in 1999, which asked about a person’s attitudes towards racial profiling and the police, along with various demographic questions, with the results coming from 903 Whites and 961 African Americans. Those who felt they were stopped by the police because of their race were more likely than those who did not think they were stopped because of their race to see racial profiling as a widespread issue and to be dissatisfied with the police, with this effect being seen by both African Americans and Whites. The findings also show that the more educated a person was, especially if they were African American, the more likely they were to see racial profiling as prevalent, disapprove of the practice and believe that they had experienced racial profiling. Those with higher income were more likely to perceive the police in a favourable way than those who had lower levels of income. These results once again show that perceptions can be more important than reality when it comes to shaping attitudes on the police as was the case with racial profiling, not helped by the results showing those with high levels of education were more likely to see racial profiling as widespread.

Because things like police legitimacy are complicated ideas, sometimes factors that look important by themselves can be insignificant when additional factors are present. Schafer et al. (2003) wrote a research paper trying to determine what prominently shapes perceptions of police services between demographics, neighbourhood contexts and personal experiences. The sample for the experiment came from a Midwestern town in the United States, where telephone interviews were conducted asking the citizens questions about their thoughts on the police and satisfactions with police services, getting a total of 1,269 usable interviews. From the interviews, race was a significant predictor of perceptions of police services until personal contact with the police and neighbourhood context was added in, with race decreasing in significance the more factors that were included by the researchers. Personal contact with the police was another significant factor in determining whether a person was satisfied with the police, with negative encounters leading to decreased satisfaction with the police, and positive encounters having no real effect on satisfaction with the police. Education was found to be an inconsistent predictor of whether someone supported community policing efforts. The effects education did have were that the more educated a person was, the less likely they were to support community policing efforts. Neighbourhood crime rates were part of neighbourhood context, with those living in high crime areas less likely to perceive the police in a positive manner. While the results show race is an important factor in police legitimacy, things like neighbourhood context and personal contact can supersede it, with higher education playing an inconsistent role in negatively viewing community policing.

While trust in the police has been seen as differing between racial groups, some studies find that this is not the case. Tyler (2005) sought to determine whether trust in the police is related with the willingness to cooperate with them, and whether trust is related to procedures and policies used by the police. In 2002, 1,653 citizens of New York City were called on the phone and interviewed about their views on the New York Police Department, asking questions about their trust in the police, how willing they were to cooperate with police, their thoughts on police procedures, and perceptions of racial interactions with the police. The answers from the interviews show that trust in the police is shaped predominantly by whether a person believed that the police were being fair in their procedures and practices, with little variation between racial groups. A person’s willingness to cooperate with the police was also tied into how fair they thought the police were being; if a person believed that the police were abusing their authority, then they were less likely to cooperate with the police when asked. The results of this study also show a slight correlation between having a high level of education and distrusting the police and being less likely to be influenced by procedural and distributive justice, though this correlation was usually not significant. Showing that procedural justice can supersede racial factors means racial factors are not completely accurate predictors of trust in the police, though the results show that distrust in the police has a slight correlation with having a high educational degree.

Methods

Procedures

Table 1 demonstrates the stratification of sampling from different neighborhoods in the mid sized-city by police patrol beats. The data for this study came from a survey conducted in the city in California in November, 2017 through March, 2018 on how citizens view the police officers of the city (Table 1). This study was a joint venture between the criminal justice faculty from California State University, Stanislaus and the mid-sized city Police Department to create and hand out a survey agreed upon by both parties. This survey was mailed to a sample of 1,500 citizens living in the city in California for them to complete, and mailed back to the city Police Department (Table 1). Once the completed surveys were sent back, canvassing was conducted to increase the response rate to the surveys and give some of the respondents who were unable to send the surveys back a chance to get their answers heard. The data from the surveys and canvassing were reviewed and cleaned by faculty members of California State University, Stanislaus, with the process completed in early April 2018.

OPEN IN VIEWER

Table 1.

Police beat sampling statistics (based on 2010 census).

Beats	Houses	Population	Surveys sent out	Survey to house ratio	Total houses/population %
31	2,903	10,750	66	1 per 44
32	2,437	11,053	57	1 per 43
33	2,489	9,455	62	1 per 41
Total	7,829	31,258			12.3/15.4
41	1,598	2,404	38	1 per 42
42	2,840	7,416	67	1 per 43
43	4,401	14,113	105	1 per 42
Totals	8,839	23,933			13.9/11.8
61	4,355	12,928	102	1 per 43
62	4,638	17,687	109	1 per 43
63	5,729	19,824	134	1 per 43
Totals	14,722	50,439			23.1/24.8
71	6,882	21,730	162	1 per 43
72	5,347	18,415	126	1 per 43
73	6,230	20,114	147	1 per 43
Totals	14,722	50,439			29.0/29.6
81	4,910	11,811	116	1 per 43
82	4,400	12,304	104	1 per 43
83	4,437	13,477	105	1 per 43
Totals	13,747	37,592			21.6/18.5
Total	63,596	203,481

The sampling design and data collection process for this study were a multi-institutional effort conducted by both university faculty and the local law-enforcement agency which was being evaluated in the survey. We implemented mandatory training for our data collectors canvassing the community during the house-to-house phase of the data collection process. The training for data collectors was intended to minimize potential biases or harm during the data collection process by reminding all participants, using both a written informed consent form and a verbal reminder, that (1) their participation in the study is anonymous, (2) they do not have to answer questions they are uncomfortable with, and (3) that their continued participation in the study is voluntary. The data collectors informed respondents that their information would be de-identified and that their responses would only be measured using aggregate analytical techniques.

The sample number of 1,500 was chosen based on prior literature about studies about citizen perceptions of the police with similar sample sizes (Frank et al., 2005; Lai and Zhao, 2010; Tyler and Wakslak, 2004), which also happens to be a good, representative number for the study being conducted. To get a representative sample of the population of the city, the city was divided up into its 15 pre-existing police beats, with stratified random sampling done for each of these police beats so each beat got a proportionate amount of surveys sent to them to make the 1,500 sample representative (Maxfield and Babbie, 2012). The reason for choosing police beats is that it is a pre-existing, simple way of sectioning the entirety of the city, and also shows where in the city the need for police officers and their services occur, with the smaller, condensed beats needing more police services than the larger beats.

The random sampling of households was done from a master list of households in the city, starting at a random point on the list, then selecting going down the list by 43 entries, selecting that point, then continuing down the list until each beat had enough entries to be proportional. Households for this study are defined as private houses, owned or rented, and excludes businesses and apartment complexes. The reason for households to be defined this way is that it makes it easier to send the surveys to the correct addresses and makes it simpler to select the sample based on single homes and not an apartment complex with multiple addresses in one location. At each household, an adult, someone over the age of 18, the age of majority in California, needed to be present to answer the survey, as adults can consent to taking the survey, unlike minors. Minors were not included in the IRB forms for who could take this survey, so they are not represented in this study.

The survey that was used for this paper is adapted from the ‘Community Survey on Public Safety and Law Enforcement’, originally developed by The Office of Community Oriented Policing Services (COPS) which is funded by the United States Department of Justice (DOJ). The survey is included in the Appendix A and the informed consent form is in Appendix B, which is a 2017 survey created by California State University, Stanislaus and the city Police Department. The survey begins with questions about how the city Police Department interacts with the community, followed by a question asking about the respondents the top three crimes they are concerned about. Next, a series of questions ask the respondent whether they feel safe in their neighbourhood and how they perceive various aspects about the police officers in the city. Then a series of questions are asked about the respondent’s contact with the police in the last 12 months and their satisfaction with those interactions. Finally, a series of demographic questions end the survey, with questions asking about gender, race, age, education level, household income, whether the respondent knows someone close to them in law enforcement, and the respondent’s employment status.

Predictor variables

This study will use several predictor variables to see what effects they might have on people’s perceptions of their interactions with city police officers. The first predictor variable is race, conceptualized as what a person believes their biological race to be, and operationalized as asking the respondent what their race is, with the options to choose from being: American Indian, Asian, Black or African American, Hispanic, White (non-Hispanic), Other (please specify), Prefer not to answer. In the analysis, Other (please specify) and Prefer not to answer were omitted, to keep the analysis simple and focus the analysis on responses that had given a definitive race. We will be recoding our race variable from a nominal variable into a binary-coded dichotomous variable (0 = non-white; 1 = white).

Household income level is an predictor variable, conceptualized as the current total household income of the respondent, which is operationalized as indicating which of the following income levels the household responding to the survey falls under: Less than $20,000, $20,000 to $34,999, $35,000 to $49,999, $50,000 to $74,999, $75,000 to $99,999, $100,000 to $149,999, $150,000 to $199,999, $200,000 or more. In the analysis of the data for household income, the levels $150,000 to $199,999 and $200,000 or more were collapsed into the $100,000 to $149,999 level to create the new level of $100,000 or more, to help simplify the variable of household income. We will be recoding our household income variable from an ordinal variable into a binary-coded dichotomous variable (0 = Income is $49,999 or less; 1 = Income is $50,000 or more).

Level of education is conceptualised as the highest level of education the respondent has achieved, operationalised as asking the respondent what their level of education is from the following choices: Less than 12 years of schooling, GED (no college), High school degree (no college), Trade Certification, Some college hours (no degree), Specify the number of hours, Associate Degree, Bachelor’s Degree, Master’s Degree, PhD, Prefer not to answer. To help simplify the data analysis of this variable, GED (no college) was collapsed into High school degree (no college), Trade Certification was collapsed into Some college hours (no degree), and PhD was collapsed into Master’s Degree, which was relabelled as Graduate Degree. Prefer not to answer was omitted from the final analysis to keep the analysis on responses that gave an actual education level. We will be recoding our education variable from an ordinal variable into a binary-coded dichotomous variable (0 = High school education or less; 1 = Some college education or more).

Employment status is conceptualized as whether the respondent is in an occupation or some alternative occupation to employment, operationalized as what the respondent’s current employment status is from the following options: Full-time (>35 hours a week), Part-time (<35 hours a week), Self-employed, Out of work and currently looking for work, Out of work but not currently looking for work, A homemaker, A student, Military, Retired, Unable to work, Other (specify), Prefer not to answer. For the analysis of the data, two new attributes were created for this variable, Employed and Unemployed, with Employed containing attributes Full-time (>35 hours a week), Part-time (<35 hours a week), Self-employed, and Military, and Unemployed containing Out of work and currently looking for work, Out of work but not currently looking for work, A homemaker, A student, Retired, and Unable to work. Other (specify) was omitted from the data analysis to keep the analysis focused on responses that gave a definitive answer to this variable. We will be recoding our crime rate per beat variable from a nominal variable into a binary-coded dichotomous variable (0 = Unemployed; 1 = Employed).

The last predictor variable for this study is the crime rates of police beats, conceptualized as the number of crimes reported in a police beat divided by the population of each beat, then multiplied by 100,000, operationalized as the crime rate in each of the 15 police beats in the year 2017, with each beat’s total crime rate seen in Table 2. In the analysis, crime rate was turned into a variable with four attributes to it: Less than 2,999, 3,000 to 4,999, 5,000 to 6,999, and 7,000 or more. The surveys that had their beat numbers recorded were given a value corresponding to one of the four attributes for this variable, depending on what the crime rate in that beat was and what section it fell under. We will be recoding our crime rate per beat variable from an ordinal variable into a binary-coded dichotomous variable (0 = Less than or equal to 4999; 1 = Greater than or equal to 5000). For all the demographic variables, mode imputation was used on all missing data cases to increase the amount of usable cases in the different analyses (Batista and Monard, 2003).

OPEN IN VIEWER

Table 2.

Crime rates by police beat in anonymous city, California for 2017.

Police beats	Crimes reported in 2017/population	Crime rate
31	380/10,750	3,535
32	583/11,053	5,275
33	517/9,455	5,468
41	627/2,404	26,082
42	497/7,416	6,702
43	460/14,113	3,259
61	847/12,928	6,552
62	953/17,687	5,388
63	1,336/19,824	6,739
71	782/21,730	3,599
72	596/18,415	3,236
73	586/20,114	2,913
81	784/11,811	6,638
82	988/12,304	8,030
83	688/13,477	5,105

Results

Descriptive statistics of survey respondents

In terms of demographics for the survey, of the 307 people who responded to the survey, 46.6% were male, 46.6% were female, 1.3% identified as other, 3.6% preferred not to answer, with 2% of respondents leaving the question of gender blank. The median age of respondents to the survey was 56, with the age group aimed for in this study being citizens over the age of 18. The racial composition of respondents was 1% American Indian, 4.6% Asian, 1.3% African American, 16.6% Hispanic, 63.5% White (non-Hispanic), and 13% of the data was omitted after mode imputation for missing data. The educational level breakdown of the sample was 5.5% had Less than 12 years of schooling, 16.6% had a High school degree (no college), 28% had Some college hours (no degree), 14.7% had an Associate Degree, 18.9% had a Bachelor’s Degree, 12.4% had a Graduate Degree, and 3.9% of the sample was omitted after mode imputation. Household income composition of the survey was 9.8% for Less than $20,000, 9.8% for $20,000 to $34,999, 11.7% for $35,000 to $49,999, 30.9% for $50,000 to $74,999, 15% for $75,000 to $99,999, and 22.8% for $100,000 or more. In terms of employment, 52.4% of the sample was employed, while 40.7% of the sample was unemployed, with 6.8% of the sample omitted after mode imputation of the initial missing data. For crime rate per police beat, 9.1% of the sample came from a beat with a crime rate of Less than 2,999, 29.6% came from a beat with a crime rate of 3,000–4,999, 51.5% came from a beat with a crime rate of 5,000 to 6,999, 8.1% came from a beat with a crime rate of 7,000 or more, and 1.6% of the sample did not have a police beat associated with their data and are missing data. For the descriptive statistics for all variables used in this study, see Table 3; for the descriptive statistics of the variables after they have cleaned for this study, see Table 4.

OPEN IN VIEWER

Table 3.

Descriptive statistics before data cleaning.

Variables	Mean	Median	Mode	SD	Min	Max	Missing
Outcome variables
Fairness	3.59	4	4	.873	1	5	14.3
Respectfulness	3.72	4	4	.910	1	5	10.1
Responsive to community concerns	3.20	3	3	.986	1	5	10.7
Trust in police	3.59	4	4	1.057	1	5	4.2
Predictor variables
Race	4.86	5	5	1.111	1	7	2.6
Education level	5.50	6	5	2.159	1	10	1
Household income	4.20	4	4	1.993	1	8	14.3
Employment status	4.96	3	1	3.999	1	12	1
Crime rate	2.596	3	3	.770	1	4	1.6

OPEN IN VIEWER

Table 4.

Descriptive statistics after data cleansing.

Variables	Mean	Median		Mode	SD	Min	Max	Missing
Outcome variables
Fairness	3.64		4	4	.821	1	5	0
Respectfulness	3.75		4	4	.866	1	5	0
Responsive to community concerns	3.18		3	3	.933	1	5	0
Trust in police	3.61		4	4	1.037	1	5	0
Predictor variables
Race	4.58		5	5	.852	1	5	13
Education level	3.64		3	3	1.447	1	6	3.9
Household income	4		4	4	1.568	1	6	6.8
Employment status	1.44		1	1	.497	1	2	0
Crime rate	2.596		3	3	.770	1	4	1.6

Before the data analysis began, the missing data in the outcome and predictor variables were replaced through mode imputation. The purpose behind using mode imputation for this study is that it allows for the different variables to have a better response rate, increasing the data available to analyse. As seen in Table 3, Fairness was missing 14.3% of its data, Respectfulness 10.1%, Responsive to Community Concerns 10.7%, Trust in Police 4.2%, Race 2.6%, Education Level 1%, Household Income 14.3%, Employment Status 1%, Crime Rate 1.6%. For each response that had a missing data value in a variable, the mean of that variable was imputed and then rounded up or down to the nearest whole number to fit in with the coding for the responses in each variable; those variables with means lower than .49 were rounded down, means with .5 or higher were rounded up. The only variable that did not have mean imputation done to them was Crime Rate, as it is based on knowing the police beat a survey came from, and since five responses did not have their police beat, coupled with the 20.4% response rate, it would make accurately imputing the beats difficult for only five additional data points.

Bivariate analysis of perceptions of policy legitimacy

To determine difference between different races and the outcome variables, coded as a nominal variable, 1 for American Indian, 2 for Asian, 3 for African American, 4 for Hispanic, 5 for White (non-Hispanic). For perceptions of fairness, a Kruskal–Wallis H test found a statistically significant difference between racial groups, χ²(4) = 10.639, p = .031, with mean ranks of 91.83 for American Indians, 135.21 for Asians, 69.38 for African Americans, 114.47 for Hispanics and 140.99 for Whites (non-Hispanic). Using a Kruskal–Wallis H test for perceptions of respectfulness of police, the test found no statistically significant difference between racial groups, χ²(4) = 3.878, p = .423, with mean ranks of 105.00 for American Indians, 142.00 for Asians, 117.25 for African American, 118.65 for Hispanic and 138.23 for White (non-Hispanic). Regarding responsiveness to community concerns, a Kruskal–Wallis H test found no significant difference between racial groups, χ²(4) = 5.264, p = .261, with mean ranks of 89.50 for American Indians, 139.29 for Asians, 68.13 for African Americans, 127.54 for Hispanics and 137.35 for White (non-Hispanic). Trust in the police had no significant difference between racial groups based on a Kruskal–Wallis H test, χ²(4) = 5.090, p = .278, with mean ranks of 79.17 for American Indians, 144.93 for Asians, 86.00 for African Americans, 124.13 for Hispanics and 137.63 for Whites (non-Hispanic).

To determine the differences between levels of education and the variables concerning perceptions of interactions with police, levels of education coded as an ordinal variable, 1 for Less than 12 years of schooling, 2 for High school degree (no college), 3 for Some college hours (no degree), 4 for Associate Degree, 5 for Bachelor’s Degree, 6 for Graduate Degree. A Kruskal–Wallis H test on police fairness found no significant difference between levels of education, χ²(5) = 1.592, p = .902, with mean ranks of 141.35 for Less than 12 years of schooling, 154.79 for High school degree (no college), 152.30 for Some college hours (no degree), 138.22 for Associate Degree, 147.31 for Bachelor’s Degree and 144.75 for Graduate Degree. Respectfulness of police was found to have no significant difference where levels of education are concerned based on a Kruskal–Wallis H test, χ²(5) = 1.019, p = .961, with mean ranks of 140.24 for Less than 12 years of schooling, 152.63 for High school degree (no college), 145.21 for Some college hours (no degree), 146.53 for Associate Degree, 145.53 for Bachelor’s Degree and 157.09 for Graduate Degree. Regarding responsiveness to community concerns, a Kruskal–Wallis H test found no significant difference between education levels, χ²(5) = 4.190, p = .522, with mean ranks of 126.56 for Less than 12 years of schooling, 156.75 for High school degree (no college), 146.76 for Some college hours (no degree), 158.07 for Associate Degree, 135.80 for Bachelor’s Degree and 155.37 for Graduate Degree. Trust in the police was found to have no statistically significant difference between levels of education based on a Kruskal–Wallis H test, χ²(5) = 1.882, p = .865, with mean ranks of 149.29 for Less than 12 years of schooling, 150.65 for High school degree (no college), 152.48 for Some college hours (no degree), 146.22 for Associate Degree, 135.86 for Bachelor’s Degree and 154.37 for Graduate Degree.

Kruskal–Wallis H tests were used to determine the differences in various levels of household income and the outcome variables, with household income coded as an ordinal variable, with 1 for Less than $20,000, 2 for $20,000 to $34,999, 3 for $35,000 to $49,999, 4 for $50,000 to $74,999, 5 for $75,000 to $99,999, 6 for $100,000 or more. The Kruskal–Wallis H test for police fairness found no significant difference between the various levels of household income, χ²(5) = 1.737, p = .884, with mean ranks of 147.15 for Less than $20,000, 153.12 for $20,000 to $34,999, 157.61 for $35,000 to $49,999, 149.23 for $50,000 to $74,999, 151.21 for $75,000 to $99,999 and 163.76 for $100,000 or more. Perceptions of police respectfulness found no statistically significant difference between income groups based on a Kruskal–Wallis H test, χ²(5) = 2.409, p = .790, with mean ranks of 150.10 for Less than $20,000, 157.12 for $20,000 to $34,999, 148.58 for $35,000 to $49,999, 163.71 for $50,000 to $74,999, 143.86 for $75,000 to $99,999 and 150.61 for $100,000 or more. Police responsiveness to community concerns had no significant difference between household income level groups when run through a Kruskal–Wallis H test, χ²(5) = 2.786, p = .733, with mean ranks of 166.95 for Less than $20,000, 158.22 for $20,000 to $34,999, 157.72 for $35,000 to $49,999, 142.90 for $50,000 to $74,999, 159.51 for $75,000 to $99,999 and 156.17 for $100,000 or more. No statistically significant difference was found between groups of household income levels when run through a Kruskal–Wallis H test along with trust in the police, χ²(5) = 1.423, p = .922, with mean ranks of 157.60 for Less than $20,000, 161.78 for $20,000 to $34,999, 155.43 for $35,000 to $49,999, 148.61 for $50,000 to $74,999, 147.00 for $75,000 to $99,999 and 160.31 for $100,000 or more.

Since employment status was turned into a binomial nominal variable, 1 for Employed and 2 for Unemployed. For perceptions of police fairness, a Mann–Whitney U test indicated that there was no significant difference between employed (M = 139.29) and unemployed people (M = 148.93), U = 9384.000, p = .279. Regarding police respectfulness, a Mann–Whitney U test indicated that there was no significant difference between employed (M = 141.60) and unemployed people (M = 145.95), U = 9756.500, p = .634. In a Mann–Whitney U test, employed (M = 139.89) and unemployed people (M = 148.15) were found to have no significant difference in their perception of police responsiveness to community concerns, U = 9481.000, p = .374. For trust in the police, unemployed people (M = 154.19) had a statistically significant higher perception of trust in their interactions with city police officers than employed people (M = 135.20), U = 8726.500, p = .042.

For the final predictor variable, Kruskal–Wallis H tests were used to find differences between the different crime rate levels in a police beat, an ordinal variable, with regard to the different outcome variables, with crime rates per beats coded with 1 for Less than 2,999, 2 for 3,000 to 4,999, 3 for 5,000 to 6,999, and 4 for 7,000 or more. A Kruskal–Wallis H test found that there was no difference between crime rates in a police beat and perceptions of police fairness, χ²(3) = .317, p = .957, with mean ranks of 149.77 for Less than 2,999, 155.12 for 3,000 to 4,999, 149.50 for 5,000 to 6,999 and 152.92 for 7,000 or more. No statistically significant difference was found between crime rate groups on their perceptions of police respectfulness based on a Kruskal–Wallis H test, χ²(3) = 1.547, p = .671. with mean ranks of 156.02 for Less than 2,999, 151.13 for 3,000 to 4,999, 153.86 for 5,000 to 6,999 and 132.90 for 7,000 or more. Regarding responsiveness to police concerns, there was no significant difference between crime rate groups based on a Kruskal–Wallis H test, χ²(3) = 2.137, p = .544, with mean ranks of 155.29 for Less than 2,999, 156.47 for 3,000 to 4,999, 151.41 for 5,000 to 6,999 and 129.78 for 7,000 or more. Finally, based on a Kruskal–Wallis H test, there was no statistically significant difference between crime rate groups and their trust in the police, χ²(3) = 1.617, p = .656, with mean ranks of 145.84 for Less than 2,999, 160.18 for 3,000 to 4,999, 149.04 for 5,000 to 6,999 and 141.80 for 7,000 or more.

To determine whether the predictor and outcome variables were correlated with each other, Spearman’s Rho was used to determine how each variable correlated with each other. For the predictor variable of race, there was weak positive correlation between it and perceptions of police fairness that was statistically significant, rs = .163, p = .008, no statistically significant correlation with police respectfulness, rs = .090, p = .143, no statistically significant correlation with responsiveness to community concerns, rs = .078, p = .204, and no statistically significant correlation with trust in the police, rs = .080, p = .911. There was no statistically significant correlation in level of education and police fairness, rs = -.035, p = .550, no statistically significant correlation with police respectfulness, rs = .017, p = .770, no statistically significant correlation with responsiveness of police to community concerns, rs = .000, p = .999 and no statistically significant correlation with trust in the police, rs = -.030, p = .610. Household income did not have a statistically significant correlation with police fairness, rs = .047, p = .413, and had no statistically significant correlation with police respectfulness, rs = -.021, p = .720, no statistically significant correlation with responsive to community concerns by the police, rs = -.015, p = .790 and no statistically significant correlation with trust in the police, rs = -.003, p = .961. Employment status was found to have no statistically significant correlation with police fairness, rs = .064, p = .280, no statistically significant correlation with police respectfulness, rs = .028, p = .635, no statistically significant correlation with responsiveness of police to community concerns, rs = .053, p = .375 and a statistically significant weak positive correlation with trust in the police, rs = .120, p = .042. Finally, crime rate per police beat did not have a statistically significant correlation with police fairness, rs = -.015, p = .791, no statistically significant correlation with police respectfulness, rs = -.032, p = .581, no statistically significant correlation with police responsiveness to community concerns, rs = -.064, p = .269 and no statistically significant correlation with trust in the police, rs = -.046, p = .424. See Figure 1 for a full correlation table between all the variables in this study.

OPEN IN VIEWER

Figure 1.

Spearman’s Rho correlations.

Logistic regression analysis of perceptions of policy legitimacy

In Table 5, the results from our Logistic Regression analysis of perceptions of police legitimacy indicators are reported. We conducted a multivariate logistic regression on four models in order to examine the odds-ratio of predictor variables on perceptions of police legitimacy. In particular, we are interested in examining the magnitude of effect and direction of the relationship of our selected socio-demographic characteristics on predicting citizens’ perceptions of police interactions related to (1) fairness, (2) respectfulness, (3) responsiveness, and (4) trust.

OPEN IN VIEWER

Table 5.

Logistic regression analysis.

	Model 1: To what extent do officers in the Police Department treat people fairly? (1 = A lot to Great Extent)	Modest 2: To what extent are officers in the Police Department respectful? (1 = A lot to Great Extent)	Model 3: To what extent is the Police Department responsive to the concerns of community members? (1 = A lot to Great Extent)	Model 4: To what extent do you trust the Police Department? (1 = A lot to Great Extent)
Variable (groupings)	Exp (B)	Exp (B)	Exp (B)	Exp (B)
N = 307
Race (1 = White)	2.18*	1.43	1.50	1.07
Education level (1 = Some college education level or more)	0.71	1.10	1.10	1.30
Income level (1 = Income is $50,000 or more)	1.39	1.32	0.66	1.03
Employment status (1 = Employed)	0.86	0.99	0.89	0.52*
Crime rate per beat (1 = Greater than or equal to 5000)	1.12	0.84	1.00	0.90
–2 Log Likelihood Ratio	304.61	309.30	315.34	319.18
Chi-square	9.36†	3.25	3.94	7.00

^† p < 0.10 *p < 0.05, **p < 0.01, ***p < 0.001.

In the all of the models, we enter in our socio-demographic predictor variables. We found that only our first logistic regression model is moderately statistically significant for predicting perceptions of police interactions, related to the fairness outcome (p < 0.10). In particular, we are interested in comparing the strength of our individual predictor variables when entering them into the model. We found that respondents who indicated that their race is White have an increase in odds of 2.18 times of rating their interactions with police as being fair ‘A lot or to a great extent’ compared to Non-White respondents (p < 0.05). The remaining predictor variables in our first model were not statistically significant for predicting perceptions of fair treatment by the police.

Discussion

Based on the results of the analysis of the ‘Community Survey on Public Safety and Law Enforcement’, this study’s hypotheses cannot be completely confirmed regarding the people of the city surveyed in California. For the first hypothesis: Demographic differences are associated with different perceptions of interactions with city police, when race was used to examine perceptions of police fairness there was a statistical correlation between the two variables, with the correlation being a weak positive one of .163. The results of our correlation analysis for this result is that, in order from the most likely to least likely to perceive the police as fair: Whites (non-Hispanic), Asians, Hispanics, American Indians and African Americans. Additionally, in our logistic regression analysis we found that White respondents are 2.18 times more likely to perceive the treatment of the police as fair when compared to Non-White respondents.

Another statistically significant correlation that was observed in our results was between perceptions of interactions with police and a demographic variable. More specifically, we found a weak relationship between employment status and perceptions of trust, which had a direction and magnitude of effect that was a weak positive correlation of .120. The correlation indicates that unemployed people were more likely to trust the police than employed people. Besides these two correlations, no other combination of demographic variables and variables regarding perceptions of the police were statistically significant in our analysis of correlational relationships; testing the first hypothesis presented mixed results.

For the second hypothesis – Neighbourhood crime rates effect citizen perceptions of city police – no combination of crime rates and variables regarding perceptions of interactions with the police were statistically significant, so the second hypothesis cannot be confirmed. Since this study used stratified random sampling to obtain its sample, there is an element of generalisability to these results. While previous research has suggested that demographic and crime rates do cause people to differ on their perceptions of the police, the results hint that this might not always be the case or as strong an effect as previously thought, especially in older studies. The results of this study might be more generalisable to smaller, less major cities in a state, or based on the overall demographics of our anonymous city, cities with predominantly White populations. As for the city itself, the results show that what causes differences in people’s perceptions of interactions with the police are not something inherent within the people themselves or the crime in the areas they live, but something else. The most likely source in these differences in perceptions is something that the city police are doing, or potentially not doing, related to their face-to-face community interactions and training policies around communicating with local residents.