Perceptions of Police Legitimacy and Citizen Decisions to Report Hate Crime Incidents in Australia

This article examines the importance of perceptions of police legitimacy in the decision to report hate crime incidents in Australia. It addresses an identified gap in the literature by analysing the 2011-2012 National Security and Preparedness Survey (NSPS) results to not only explore differences between hate crime and non-hate crime reporting but also how individual characteristics and perceptions of legitimacy influence decisions about reporting crime to police. Using the NSPS survey data, we created three Generalised Linear Latent and Mixed Models (Gllamm), which explore the influence of individual characteristics and potential barriers on the decision to report crime/hate crime incidents to police. Our results suggest that hate crimes are less likely to be reported to police in comparison to non-hate crime incidents, and that more positive perceptions of police legitimacy and police cooperation are associated with the victim’s decision to report hate crime victimisation. 


Introduction
Prejudice motivated crime (PMC)-predominantly referred to in academic literature as hate crime 2 -is severely under-reported and under-reporting happens on an unequal basis (Gerstenfeld 2011). Multiple factors have an impact on the under-reporting of hate crime, such as the victim's minority group status (Culotta 2005;Gerstenfeld 2011;Miles-Johnson 2013;Shively et al. 2001;Zaykowski 2010) and the victim's perception of the legitimacy of criminal justice system agencies (Goudriaan et al. 2004;Tyler 2005Tyler , 2011. Citizen perceptions of the legitimacy of institutions play a central role in understanding PMC reporting behavior. Legitimacy is an important aspect for authorities, institutions and institutional arrangements to be successful due to the inherent difficulty of coercing others by possession and power (Tyler 2006). The public needs to believe in the values upheld by the police to consider their actions legitimate (Kääriäinen and Sirén 2011). Legitimacy influences cooperation with police and fair procedures enhance police legitimacy even further (Tyler and Fagan 2010).
The relationship between the public and police is important in battling hate crime (Hall 2012). Existing research explores the influence of police legitimacy on victim reporting behavior (Jackson et al. 2012;Kääriäinen and Sirén 2011;Tyler and Fagan 2010) as well as the influence of minority group status on trust in and cooperation with the police (Miles-Johnson 2015; Murphy 2013; Murphy and Cherney 2010;Sargeant et al. 2014). There is limited research, however, on the importance of perceptions of police legitimacy in the decision to report hate crime incidents. This article addresses this gap in the literature by analysing the 2011-2012 National Security and Preparedness Survey (NSPS) results (Ramirez et al. 2013) and exploring the differences between reporting PMC and non-PMC-comparable crimes that do not have the motive of prejudice, bias or hate (Mason and Dyer 2013)-including how individual characteristics and perceptions of legitimacy shape the decision to report crimes.

Literature review
Australian and international research indicates that ethnic minorities display low levels of confidence and trust in the police, resulting in less voluntary cooperation (Cherney and Chui 2009;Murphy and Cherney 2011;Tyler 2011). Perceptions of over-policing and under-protecting influence these levels of cooperation, as well as prior negative experiences with law enforcement in former countries of residency (Cherney and Chui 2009;Murphy and Cherney 2011). The Leicester Hate Crime Project conducted in the United Kingdom from 2012 to 2014 found that hate crime victims have felt their complaints had not been taken seriously by police (and other authorities) and only a few had declared that they would report hate crime in the future (Chakraborti et al. 2014).
Multiple studies have confirmed an association between trust in the police and the willingness to cooperate (Cherney and Chui 2009;Murphy and Cherney 2011;Tyler 2005Tyler , 2011. Prior research has determined that the general public judges police officers and police procedures not only by the efficacy of crime control but also by their standards of justice, focusing on the fairness of procedures (Tyler 2005). Procedures under evaluation are the 'neutrality of decision making, respectful and polite interpersonal treatment and providing opportunities for input into decisions' (Tyler 2005: 339). Positive perceptions of police officers' authority matter and abuse of authority by police officers results in decreased trust in police and, consequently, in less willingness to report crime (Tyler 2005). Further, different ethnic groups have different expectations and requirements regarding the service of police, and many ethnic groups differentiate between police performance and procedural justice (Sargeant et al. 2014). Sargeant et al. (2014) found that procedural justice is less important for cooperation with police for Vietnamese and Indian respondents compared to the general population, while police performance was more effective in promoting trust in the police for Vietnamese respondents (with no significant differences for Indian participants). This result might be due to historical experiences of conflict and cultural differences as well as more recent incidents of biased policing, especially in Australia (Sargeant et al. 2014).
The matter of government legitimacy also influences the reporting behaviour of victims. The police force is a government agency and, as such, its officers represent the rules and regulations of the government. The state has the responsibility to maintain peace, protect individuals and achieve public legitimacy (Tyler 2003). Individuals who see the government as legitimate will be more likely to accept laws implemented by the government and take on the responsibility to abide by such laws and cooperate with government bodies. Dissimilarities in the confidence in the police (and government) additionally undermine social integration and create social differences (Cao 2011). Bradford (2014) has examined the links between procedural justice, social identity and police cooperation and has tested if people are more cooperative with the police if they feel included in the social group that the police represents and identify with this group. The study found significant evidence for police fairness, legitimacy and social identity to influence cooperation with police (Bradford 2014). Oliveira and Murphy (2015) even established that social identity is more important than ethnicity or race in predicting views of the police.
Certain attributes of individuals not only contribute to the chance of PMC and other types of crime victimisation but also play a more specific role in PMC reporting behaviour. As Culotta (2005: 23) states, '[t]he very reason a victim may have been singled out (that is, ethnicity) may also create an obstacle for reporting the incident'. Indigenous Australians especially have specific barriers when reporting a crime. Next to cultural and language barriers, members of Indigenous communities-especially in those communities that are small and isolated-fear negative repercussions, stigmatisation and banishment. Additionally, these persons may experience deficiencies in victim support, government and non-government services and police resources, all of which are barriers to reporting crimes to police (Taylor and Putt 2007;Willis 2011). An unawareness of relevant authorities or distrust towards those authorities as well as a feeling of resignation influences the (un)willingness of Arabs, Muslims and individuals of Middle Eastern appearance to report PMC (Poynting and Noble 2004). Those individuals who have reported an incident in the past have often been dissatisfied with the response and service they have received, resulting in victims avoiding public places and feeling excluded from public life (Poynting and Noble 2004).
Accordingly, being part of a minority group and experiencing a PMC further decreases the reporting of such crime. PMC victims are more likely to refrain from filing reports and less likely to involve the police (Perry 2001). Victims of PMC are more prone to be in powerless situations as well as more likely to have poor relations with law enforcement officials (Gerstenfeld 2011). Personal barriers such as language, culture, sexual orientation, hate crime law knowledge, trust in police and responses to hate crime have an effect on reporting behaviour (Culotta 2005;Shively et al. 2001). The literature points to people with perceptions of lower levels of police legitimacy than the majority of the population as less likely to report crime. The literature also suggests that minority groups have lower perception levels of police legitimacy compared to the majority group. The study on which this article is based hypothesised that perceptions of police legitimacy influences the decision to report PMC. A detailed discussion of the methods used for this assessment is outlined next.

Methods
Data were drawn from the NSPS, a national probability sample, collected under the auspices of the Australian Research Council Centre for Excellence in Policing and Security (CEPS). The study randomly sampled 6,590 Australian residents via Computer Assisted Telephone Interviews. These individuals completed a short two-minute phone survey and 6,098 of these respondents agreed to go on to complete the main survey (online or via hardcopy). Of those recruited, 4,258 people returned completed surveys 3 and of these, 762 had been victims of crime within the previous twelve months. The NSPS primarily benchmarked attitudes and perceptions towards national security and disaster preparedness (Ramirez et al. 2013), but the survey also included items designed to measure self-reported crime and victimisation as well as accounting for hate crime occurrences. 4 Further, the survey allowed for distinction between violent and property victimisation as well as whether non-PMC victims or PMC victims reported a crime to police.

Outcome variable
Two of the questions included in the NSPS asked respondents, firstly, if they had been victims of a property or violent crime in the past twelve months and, secondly, if they had reported the crime to police. Some of the 762 respondent victims had experienced multiple crime incidents and, for each crime incident, we recorded whether the victim had reported the crime to police. We created a multilevel dataset to reflect types of crime victimisation for individuals in order to account for property and violent crimes experienced by each person. In some cases, a respondent experienced both types of crime but only reported one and not the other. The newly created dichotomous reporting variable included 871 incidents of crime nested within the 762 individual victims.

Explanatory variables Perception of police legitimacy
The NSPS measures police legitimacy consistent with existing long-standing literature on the topic (Bradford 2014;Gau 2011;Hinds and Murphy 2007;Murphy and Cherney 2012;Reisig et al. 2007). Items measuring perceptions of police legitimacy in the survey asked about how much respondents strongly disagreed (1) to strongly agreed (5) with the following statements about the police: The above scale (α = 0.95) includes police legitimacy items (that is, trust in the police, police performance), as well as procedural justice items (that is, quality of treatment, quality of decisionmaking) because of high loadings onto one factor during a factor analysis, which specified the items best fitted for the scale. Gau (2011) has tested the assumption that procedural justice and police legitimacy are distinct from each other using a confirmatory factor analysis and has found, to the contrary, that a tendency for trust to load with procedural justice items exists (see also Tankebe 2013;Tankebe et al. 2016).

Willingness to cooperate with police
An additional scale was created to inquire whether respondents have noted that they will cooperate with police and, if they were victims, whether they were willing to report the crime to police (see, that is, (a) call the police to report a crime; (b) help police find someone suspected of committing a crime by providing them with information; (c) report dangerous or suspicious activity to police; and (d) willingly assist police if needed (α = of 0.90).
Identifying with Australia and its community Bradford (2014) has established a link between procedural justice, social identity and police cooperation, with people being more cooperative if they feel included in the social group that the police represents and with which they identify. The NSPS instrument asked respondents: 'How much do you agree or disagree with the following statements?', with response categories ranging from 1 for 'strongly disagree' to 5 for 'strongly agree' for the following items: (a) I see myself first and mainly as a member of the Australian community; (b) it is important for me to be seen by others as a member of the Australian community; (c) I am proud to be Australian; and (d) what Australia stands for is important for me (α = 0.84).

Perceptions of government legitimacy
As Goudriaan et al. (2004) point out, opinions toward the government and its responsibility can influence the reporting behaviour of victims. Perceptions of federal government legitimacy were established through asking respondents 'How much of the time can you trust the Australian government to do what is right?' (on a scale of four from 1 for 'just about always' to 4 for 'just about never'; reverse-coded) and asking about their level of confidence in the Prime Minister of Australia, federal politicians and federal parliament (1 for 'hardly any confidence', 2 for 'only some confidence' and 3 for 'a great deal of confidence'; α = 0.82).
Due to a factor analysis indicating loading onto two separate factors, we distinguished between federal and state government legitimacy and explored perceptions of state government legitimacy with respondents' level of confidence in their State Premier and their State Politicians (1 for 'hardly any confidence', 2 for 'only some confidence' to 3 for 'a great deal of confidence'; α = 0.81). Prior research utilises similar items to tap into the legitimacy of authority figures and government agencies (Useem and Useem 1979;van der Toorn et al. 2011;Weatherford 1992).

Perceptions of law legitimacy
Similar to prior research (see, for example, Murphy, Murphy and Mearns 2010; Murphy, Tyler and Curtis 2009), we measured attitudes and obligations toward the law by items asking respondents how much they strongly disagree (1) to strongly agree (5) with the following statements: (a) you should always obey the law even if it goes against what you think is right; (b) I feel a moral obligation to obey the law; and (c) people should do what our laws tell them to do even if they disagree with them (α = 0.86).

Control variables
Variables that may influence victims' reporting behaviours include linguistic difficulties of respondents, such as speaking another language than English at home (LOTE dummy) (Culotta 2005;Lockyer 2001); the Indigenous status of respondents (ATSI dummy) because Indigenous communities in Australia experience cultural and language barriers (Willis 2011); and Australian citizenship (dummy), because citizenship status guarantees equal rights and services to individuals. Further potential barriers to reporting behavior may include the immigrant status of a respondent (foreign born dummy) because immigrants and ethnic minorities are more prone to victimisation and less likely to report crime to authorities (Culotta 2005); the perception of social isolation in the community (preference for Anglo-Saxons as neighbours; 5 1 for 'strongly disagree' to 5 for 'strongly agree'; α = 0.82); and religion (a dummy coded variable indicating 'Christian' or 'Other').
We also controlled for demographics such as age, gender, dependent children, income (ranging from less than $20,000 to $150,000 or more), education (seven levels from 'No school' to 'Postgrad'), homeownership, marital status, and employment status. 6 Table 1 displays the percentages of the potential barriers for reporting crime to police by victim group (No-victim, Non-PMC victim and PMC victim).

Analytic approach
We used Generalised Linear Latent and Mixed Models (Gllamm) with the dependent variable indicating whether the victim reported the crime incident to police, controlling for all control and explanatory variables. In this multilevel model, units of observations were captured at two different levels. The level one units were crime incidents (property and violent crime) and level two units were individual respondents. The NSPS multilevel dataset included 871 crime incidents nested within 762 individuals, which will vary depending on the model and the missing data on certain variables. According to Pardoe (2004: 298), a hierarchical model 'can account for lack of independence across levels of nested data' (that is, crime incident nested in individuals). The binary response for the multilevel model for the ith observation for the jth individual is the following: = 1 for crime reported; = 0 for crime not reported. According to Grilli and Rampichini (2006: 10), the two-level model assumes the form below: with g(.) indicating the link function.

Limitations
According to Shively et al. (2001), future research may only adequately deal with the limitation of infrequent hate crime events and predominant minority group victimisation by oversampling minority groups until a sufficient sample for analysis exists. The outcome variable is a dichotomous one, indicating whether the crime has been reported to the police, instead of a process variable, explaining why the victim has decided to report the crime to police, which limits the understanding of the decision-making process around reporting-behaviour. Further, the explanatory variables (for example, perception of police, law and government legitimacy and willingness to cooperate with police) have been collected after the dependent variable of reporting the crime to police; therefore, it is possible that respondents reported the crime to police which led to an unsatisfactory result or unfair treatment, and which would have consequently influenced responses to explanatory variables. If experiences with reporting the crime to police were satisfactory, this could have similarly shaped their responses to perception of legitimacy.
We therefore cannot rule out the possibility of reverse causality; that is, the respondents' experiences of reporting crime could shape their perceptions of legitimacy, rather than their perceptions of legitimacy shaping their reporting behaviour.
The limitations around the utilisation of surveys including hate crime items, such as the NSPS, indicate that better measures are necessary to observe and record hate crime victimisation and, consequently, the reporting of hate crime to police, and thus to capture PMC in Australia. Better data collection processes and recording of PMC incidents by police could increase the visibility of PMC incidents, while further research into the utilisation of different legislative approaches and police strategies could have an impact on the reporting behaviour of PMC victims. With Australia's lack of official recording practices in regards to hate crime victimisation, surveys including victimisation items may be the best possible way to explore hate crime victimisation and victims' reporting behaviour at this time.

Results
In this analysis, we were especially interested in the influence of police legitimacy on victims' reporting behaviour. A hate crime variable was included on the assumption that people who experience PMC are less likely to report crime to police than people who experience a crime without a prejudice motive. An indicator for experiencing a property crime incident versus a violent crime incident was also included, as people experiencing property crime may be more likely to report the crime to police due to monetary incentives than people experiencing a violent crime. The model assumes that all of the above constructs could potentially have an impact on the reporting behaviour of victims.
Using the NSPS survey data, we created three Gllamm models (outcomes displayed in Table 2), which explore the influence of individual characteristics and potential barriers on the decision to report crime incidents to police. As this is a multilevel model, the table contains the crime incidents (level one units) and the individual respondents (level two units). Model 1 includes the demographic control variables and the crime incident flags, Model 2 includes the demographic and potential barriers control variables and the crime incident flags, and Model 3 includes all control variables, the policing explanatory variables and the crime incident flags.

Model 1 interpretation: Demographics influencing reporting behaviour
Model 1 suggests that, out of all demographic variables, age alone was significantly associated with the likelihood of reporting crime incidents to police. Every one unit increase in individual age is associated with an estimated 0.4% increase in the odds of reporting crime to police (OR: 1.004; p-value: 0.004). Prior findings from Carcach (1997) and Goudriaan et al. (2004) support conclusions that victims who are older are more likely to report crime to police. According to Carcach (1997), gender is an additional important factor for reporting behaviour but reporting depends on the type of crime (for example, assault versus robbery). In this study, none of the other demographic variables was significantly associated with reporting crime to police.
IJCJ&SD 98 Online version via www.crimejusticejournal.com © 2018 7(2) These multilevel models allow for a distinction between hate crime and crime without a prejudice motive and between violent crime and property crime. Model 1 indicates that the variables hate crime and property crime are significantly associated with whether the crime incident was reported to police. Experiencing a hate crime incident is associated with an estimated 14.9% (1 -0.851) decrease in the odds of reporting crime to police (OR: 0.851; p-value: 0.027), while experiencing a property crime incident is associated with an estimated 14.7% increase in the odds of reporting crime to police (OR: 1.147; p-value: 0.011).

Model 2 interpretation: Potential barriers influencing reporting behaviour
Model 2 explores the potential barriers that may influence the reporting of crime to police and displays similar results to Model 1: the variables age, hate crime and property crime are significantly associated with reporting behaviour. The results show that a one unit increase in age and experiencing a property crime incident is associated with an estimated 0.4% (OR: 1.004; pvalue: 0.004) and 13.8% (OR: 1.138; p-value: 0.018) increase, respectively, in the odds of reporting crime to police. Experiencing a hate crime incident is again associated with an estimated 15.5% (1 -0.845) decrease in the odds of reporting crime to police (OR: 0.845; p-value: 0.021). Although prior research indicates that minority group status influences the decision to report crimes to the police, the potential barrier variables are not significantly associated with reporting or not reporting crime incidents to police. This result may indicate that the potential barrier variables are less likely to impact on the reporting behaviour of victims or could suggest that the NSPS sample of PMC victims does not fit the typical hate crime victim often featured in the hate crime literature. Our sample suggests that PMC victims were primarily native-born, English speaking Australians and therefore may not experience the same potential barriers that exist in other international contexts. 7 PMC victims in the NSPS self-reported that they (or anyone in their household) were victimised because of their skin colour, ethnicity, race or religion, which suggests their perceptions as victims of hostility was associated with some part of their identity.

Model 3 interpretation: Policing scales influencing reporting behaviour
Model 3 seeks to explore if the explanatory variables may have an influence on reporting crime to police. As Table 2 shows, the findings of Model 3 mirror the findings from Model 1 and 2, indicating that variables age, hate crime and property crime are again significantly associated with crime reporting behaviour. An increase in age and experiencing a property crime incident was significantly associated with an estimated 0.4% (OR: 1.004; p-value: 0.011) and 13.8% (OR: 1.137; p-value: 0.016) increase, respectively, in the odds of reporting crime to police, and experiencing a hate crime incident is associated with an estimated 14.9% (1 -0.851) decrease in the odds of reporting crime to police (OR: 0.851; p-value: 0.025). In addition, none of the potential barriers variables was significantly associated with reporting behaviour and a partial mediating effect between the variables Australian citizenship and property crime exists. 8 In regards to the explanatory variables, only a respondent's willingness to cooperate with police is significantly associated with crime reporting behaviour. Every one unit increase in individual willingness to cooperate with police is associated with an estimated 16.2% increase in the odds of reporting crime to police (OR: 1.162; p-value: 0.000). This suggests that respondents who stated that they were likely to cooperate and assist police were also more likely to report the crime to the police. Perceptions of police legitimacy was, at one point, also significantly associated with reporting behaviour until willingness to cooperate with police was added into the model. Every one unit increase in individual perceptions of police legitimacy was associated with an estimated 5.2% increase in the odds of reporting crime to police (OR: 1.052; p-value: 0.032). Respondents who indicated having higher perceptions of police legitimacy were also more likely to report the crime to the police.
We checked for a moderating and mediating effect and found a true mediation effect between perceptions of police legitimacy and the willingness to cooperate with police. 9 The willingness to IJCJ&SD 100 Online version via www.crimejusticejournal.com © 2018 7(2) cooperate with police completely mediates the police legitimacy and crime reporting relationship. This suggests that the willingness to cooperate with police explains the relationship between perceptions of police legitimacy and the reporting of crime to police. People who have higher perceptions of police legitimacy are also more willing to cooperate with police and, ultimately, more likely to report the crime to police.

Discussion
The analysis above shows that perceptions of police and the willingness to cooperate with police along with certain demographic characteristics play important roles for victims' decision-making processes to report hate crime to police. Consistent with international literature, our results suggest that PMC is less likely to be reported to police in comparison to non-PMC incidents, and that more positive perceptions of police legitimacy and police cooperation are associated with the victim's decision to report PMC victimisation. Interestingly, potential barriers that prevent victims from reporting hate crime as indicated by prior research were not associated with reporting behaviour in our analysis. However, there is some indication that citizenship status may partially mediate the likelihood of reporting property crime to the police.

Potential barriers to reporting hate crime
Our results suggest that the potential barriers to reporting PMC indicated by prior researchreligion, speaking a language other than English, foreign born residents, and so on-did not reach significance in their association with reporting behaviour. These victim characteristics are considered barriers to reporting hate crimes because they highlight the isolation from mainstream culture-in this study, Australian culture-that many victims face. These barriers highlight language difficulties as well as problems with cultural differences, and understanding and normalisation of behaviour (Taylor and Putt 2007;Willis 2011). Instead, we suspect that the lack of association in our study between these potential barriers and hate crime reporting reflect the unique nature of Australian crime and immigration patterns.
Hate crime victims in this sample are typically English speaking Australian citizens. Those who indicated they were a victim of a PMC and who were also foreign born where typically from White English speaking countries. Therefore, the PMC victims in this sample did not represent the typical PMC victim that is often the focus of hate crime research. Thus PMC victims in this survey did not necessarily experience those barriers as they were native speakers of English, and were part of, or visibly akin to, the majority White population. While there is little information regarding who PMC victims are in the Australian context, there is some evidence that the majority of people who indicate they are the victims of a PMC are primarily native-born, English speaking Australians. Research by Benier (2017) on the Brisbane, Queensland, sample of the Australian Community Capacity Study (a longitudinal study of community processes, crime and disorder) also shows similar patterns in self-report data. Benier's (2017) study showed that only 35 per cent of PMC victims were foreign born, and only 29 per cent of PMC victims spoke English as a second language. Thus, in the Australian context, the majority of self-reported hate crimes are not from victims who experience barriers due to their belonging to minority groups. This distinguished our results from prior research in this area.
So why do these hate crime victims perceive that they are the victim of a crime motivated by their own national, ethnic or racial identity? If the majority of these victims are native-born English speaking residents, perhaps it is the context of their victimisation that leads to the perception they are the victims of hate motivated crime. One possibility is that their perceptions are influenced by (perceived) changes to population diversity in their neighbourhoods and communities. Australia's population is becoming increasingly diverse, which means that individuals are more likely to come into contact and interact with others who identify as a different race or ethnicity (Forrest and Dunn 2011). This is consistent with work by Lyons (2007) who found that White residents typically report their victimisation as a PMC when it occurs in IJCJ&SD 101 Online version via www.crimejusticejournal.com © 2018 7(2) disadvantaged neighbourhoods that have greater populations of race and ethnic minorities than more wealthy neighbourhoods. Thus Lyons (2007) argues that 'anti-White hate crime' looks typically like regular crime more so than it looks like anti-Black hate crime. Perhaps PMC victims in this study perceived their victimisation to be motivated by hate simply because the perpetrator is of a different race or ethnic identity.
Another possibility is that these hate crime victims are second-generation residents who have been born in Australia and speak English more fluently than their immigrant parents. As such, they too may not experience the same barriers to reporting that are often associated with reporting PMC victimisation. There is some evidence to suggest that second generation residents actually have higher rates of PMC victimisation than first generation immigrants (Perry 2002). Perry (2002) suggests that second generation residents in the United States of America, born and raised in that nation, may be targets of hate motivated crime because they represent those who are 'not real Americans', implying that their visible ethnic minority status overrides fluency in English language or their native-born status. So, while they may not experience the barriers to reporting that are often associated with the most marginalised minority groups, their willingness to report appears to be most associated with perceptions of police legitimacy and cooperation.
The exploration of the NSPS data in the multilevel models demonstrates that PMC victims are more likely to report property crime than violent crime to police. In accordance with prior literature, victims are more likely to report crime to police when a certain gain or outcome is apparent (Schneider et al. 1976). The partial mediating effect between property crime and being an Australian citizen for explaining the reporting of crime suggests that victims who are not Australian citizens are less likely to have contents insurance and, therefore, as Schneider et al. (1976) assert, have less incentive to report the crime to police.
Research conducted on the Leicester Hate Crime Project found that hate crime victims experienced their victimisation as a 'routine reality of being "different"', as opposed to an incident in need of reporting (Chakraborti 2015: 5). The 2016 Census data indicated that over 28 per cent of Australian residents were born overseas and over 34 per cent of people had both parents born overseas (Australian Bureau of Statistics 2017, 2018). Even though Australia is a multicultural country, lack of support from criminal justice authorities may lead to victims normalising their everyday experiences of hate crime. According to Gillis (2013), little is known about police responses to hate crime incidents and often community awareness and willingness to embrace and utilise PMC legislation stems from criminal justice system agents (police and prosecution). What may be necessary for the successful prosecution of hate crime cases may lie in the adoption and implementation of relevant police strategies and procedures. Mason et al. (2014) point towards enhancing police members' understanding of PMC by integrating such information into police manuals, recording systems, recruit training and policy documents, which should lead to the comprehensive investigation and successful prosecution and sentencing of hate crime incidents. Such a strategy should enhance community confidence in police, ultimately leading to prevention of hate crimes (Mason et al. 2014).

Police legitimacy and cooperation
The multilevel analysis indicates that perceptions of police legitimacy and the willingness to cooperate with the police have a significant association with reporting behaviour. These explanatory variables are key indicators for whether hate crime is reported to police. In interviews with PMC victims, McCaffery (2013) found that, amongst the inhibiting factors for reporting crime to police, all respondents indicated lack of confidence in the broader justice system. Doubt in the judicial system referred to lenient sentencing, punishments reinforcing prejudice, revolving door justice, delays in trials and court officials lacking experience, as well as the need for adequate legislation addressing anti-social behaviour. McCaffery (2013)  that PMC victims' past interaction with police as well as the overall image of the police service are key factors in deciding whether to report the hate incident to police.
Existing literature demonstrates the challenges for police officers in commanding legitimacy and cooperation from minority communities (Cherney and Chui 2009;Murphy and Cherney 2011). Immigrants and race and ethnic minorities are less likely to have positive views of police in terms of legitimacy, procedural justice and cooperation due to real or perceived tensions between formal social control agents and marginalised groups. In deciding whether to report victimisation to police, perceptions of legitimacy and cooperation remain important factors. However, it is clear from this study that police legitimacy and cooperation are important influences for all groups in their willingness to report, not just for those who belong to marginal groups.
Expanding our understanding of PMCs is of continuing important in the current political climate in Australia and globally. Crime motivated by hate of a person's ethnicity, national identity, religion or sexuality, or for political reasons, has consequences not only for individual victims but also for communities; when groups become disconnected and excluded, signs of hate can foster radicalisation (Spalek 2007). Local law enforcement plays a vital role. Fostering perception of legitimacy and cooperation with the most vulnerable groups in our communities is important for encouraging connection and inclusion. This is particularly critical for victims of PMC. Police administrators and departments need to engage more effectively with community groups different from the mainstream population, incorporate diversity awareness and the importance of police legitimacy and cooperation into police training, and recruit more persons from minority groups into police forces.
Correspondence: Susann Wiedlitzka, Lecturer in Criminology, School of Law, Politics and Sociology, University of Sussex, Freeman Building, Falmer, Brighton BN1 9QE, United Kingdom. Email: S.Wiedlitzka@sussex.ac.uk 1 This work was supported by the Australian Research Council (ARC) Centre of Excellence in Policing and Security (CEPS RO70002) that funded the National Security and Preparedness Survey. This article is part of a PhD thesis by the corresponding author, which was supported by the ARC Linkage Project 'Targeted Crime: Policing and Social Inclusion' (LP110100585). The article was not solicited as part of the special issue ′Exploring Discourses of Hate' but the journal's Chief Editors thought it appropriate to publish within the issue due to its fit with the theme. 2 The academic literature uses the term 'hate crime' predominantly. According to Walters (2014: 58) '"hate" is a highly emotive term and narrow in meaning'. Other terms include bias crime and targeted violence. For the purpose of this paper, any of these terms can be used interchangeably. We will, however, predominantly resort to the term prejudice motivated crime (PMC). According to Mason and Dyer (2013: 874), parallel crimes consist of comparable crimes that do not have the motive of prejudice, bias or hate. Lewis (2013: 57) refers to parallel non-bias-motivated offenses. We will refer to non-PMC in this context. 3 The NSPS sample showed similarities to the 2011 census; however, there was an oversample in the Australian Capital Territory. One PMC victim supplying insufficient information and one under-age respondent were dropped from the survey, resulting in a sample of 4,256 people. 4 The NSPS asked respondents if 'this incident occurred because of the skin colour, ethnicity, race or religion of anyone in the household?'. We acknowledge that this survey did not capture crimes targeting people because of their sexual orientation, transgender identity, disability or any other group identity or characteristic. 5 The NSPS used the following question to explore the marginalisation and perception of isolation from the community: 'How much do you agree or disagree with the following statement?' A factor analysis indicated that item (a) people in this community prefer that residents in the area are mostly Anglo Saxon IJCJ&SD 103 Online version via www.crimejusticejournal.com © 2018 7(2) and item (b) people in this community do not like having members of other ethnic groups as next door neighbours, loaded highly onto one factor. 6 For further information, see Wiedlitzka (2016). 7 We detected a partial mediation effect between property crime and identifying as an Australian citizen, indicating that being victimised by a property crime may partially explain why Australian citizenship is related to reporting crime to police. People of other nationalities might not come forward to report crime to police because they may be unfamiliar with the Australian criminal justice system, have negative experiences with criminal justice authorities in their home country, may not want to cause any trouble in their host countries or may fear for deportation due to illegal status. 8 Before the addition of property crime, identifying as an Australian citizen was associated with an estimated 18.5% (OR: 1.185; p-value: 0.041) increase in the odds of reporting crime to police. 9 The willingness to cooperate with police variable tapped into respondents' general attitudes towards reporting a crime to police, while the crime report variable directly tested if people have reported a crime to police in the past twelve months. We ran a simple correlation to test if these constructs were too similar and found that, with a Pearson's r score of 0.23, this did not seem to be the case.