The Impact of Prior Victimization and Socio-Economic Status on People’s Crime-Reporting Behavior

The purpose of this study is to determine whether or not prior victimizat ion, along with socio-economic status affect people’s willingness to report crimes to authorities. This study is limited to addressing the effect of prior v ictimization on willingness to report crimes by 1) the type of victimizat ion (i.e ., being a v ictim of property crimes or a victim of crimes against persons), 2) being a vict im of a crime, regardless of the type of crime versus not being a victim, and 3) the difference in crime-report ing variation between the two. The findings of this research study are based on the analyses of the data that have been collected through a self-admin istered survey questionnaire distributed to 531 undergraduate university students. Overall, the findings that emerged from this study suggest that prior victimizat ion cannot be considered a good predictor of crime-reporting behavior.


Introduction
Crime -reporting behavior is generally considered a form of willingness to report crimes to authorities. The review of the existing literature on crime-reporting behavior shows that people's willingness to report crime to the police is influenced by a host of interrelated factors. Some of those factors have a direct effect on crime-reporting behavior; some of them have a meandering effect through negative or positive attitudes toward the police on crime-reporting behavior. Most prior studies in this area have studied crime -reporting behavior in part. Yet, most researchers have been focused on one type of crime or on factors that have been vaguely conceptualized.
The current study includes prior v ictimization as a primary variable that influences one's willingness to report crime to the police. After all, there are a considerable nu mber of studies that show prior v ictimization is an important factor that affects crime-report ing behavior [1][2][3][4][5][6][7][8][9][10]. Prior victimization in this study, as a variable, has two d imensions, victimization by crime (i.e., crimes against property and crimes against persons) and victimization by the police (e.g., police misconduct).
In this study, victimization by the police, however, will only be considered in reference to the review of existing literature on v ictimization.

Forms of Victi mization
Vict imization takes many d ifferent forms, including domestic abuse by individuals close to the victim, other non-domestic victimization, frequency of victimization events, time of occurrence, location of occurrence (e.g., public location vs. private location), different types of victimization (e.g., crimes against property vs. crimes against persons), and different degrees of severity of victimization (e.g., fro m murder, rape, repeat-assaults, to property loss). Each one of these forms of vict imization differs in terms of propensity to report victimization events to the police, and each form of v ictimization may be explained by different sets of correlates[see, [1][2][3][4]6,9,11].

The Effect of Victi mization Experiences on Crime -Reporting Behavi or
Xie et al. researched the relationship between factors that affect crime-reporting behavior for individuals who have been victimized by crime. The major assumptions in their study were that crime-report ing behavior will be affected by the victim's prior experience with the police, by whether or not an arrest was made by the police in an effort to investigate the crime that has affected the victim, and by the police response to an individual's own prior vict imization rather than victimization of another household member [8].
To test these hypotheses, the researchers analyzed longitudinal data borrowed fro m the NCVS (National Crime Vict imization Survey) of 2002. In their study, Xei et al. found that the greater the police efforts following the most recent victimization of an individual, the greater the likelihood that that individual will report subsequent victimization events to the police. Furthermo re, consistent with Hickman and Simpson's [12] and Ho lmberg's [13] research findings, Xei et al.'s research shows that a positive police response to prior vict imization (i.e., if the offender who committed the crime against the victim was arrested) encouraged victims to report subsequent crimes to the police. Regarding vicarious reporting of crimes 1 to the police, Xei et al.'s research shows that victimization of, fo r example, another household member d id not have an effect on an individual's crime-reporting behavior 2 . This study points out that prior vict imization is strongly associated with crime -reporting behavior. Therefore, those who have been previously victimized by a crime are more likely to report subsequent crimes to the police [8]. Moreover, the findings of an earlier study conducted by Conaway and Lohr also confirm that crime-reporting behavior is strongly associated with one's prior victimization status. Conaway and Lohr's analysis of factors associated with reporting vio lent crimes to the police show that people who have been previously victimized, regardless of the type of victimizat ion, are mo re likely to report subsequent victimization events to the police [5].
In general, research shows that as the frequency of victimization events increases, reporting victimization events to the police also increases. This hypothesis has been emp irically supported by the research findings of Unnever and Cornell's study [14]. According to Unnever and Cornell, who examined factors that influence students' decisions to report being bullied to school officials, victims who reported bullying to school officials increased as the persistency of victimization increased. In other words, the findings of this study suggest that the higher the repetition o f v ictimization is, the higher the reporting rates will be [14,15]. Research findings of Williams and Cornell also show that there is a positive influence on students' willingness to seek help for a threat of violence when they have been previously victimized by crime [16]. This tells us that there is a reason to believe that an increased number of v ictimization events is positively correlated with willingness to report subsequent victimization events to the police. Ho wever, this conclusion is not universally supported by all prio r research. Zhang, Messner, and Liu, for examp le, found that there is a negative effect of prior vict imization on reporting subsequent victimization events. This study shows that individuals who 1 Vicarious reporting of crimes to the police refers to actions taken by other individuals, other than the victim or the witness of a speci fic crime. For example, if someone witnesses a crime but is not willing to report it to the police, rather info rms a friend or a family member of that crime occu rrence, and now the second person (e.g., friend or family member) reports it to the police. This form of reporting is referred to as " vicarious reporting." 2 In this study, crime-reporting behavior refers to people's willingness to report (future) crimes to the authorities. In other words, we assessed the reporting behavior and a speci fic number of factors (e.g., prior victimization and SES) that influence this behavior rather than the actual number of crimes that one has reported to the police in the past. The phrase " crime-reporting behavior" in this study is also used in reference to the existing literature, again in reference to people's willingness to report crimes to the police or other authorities. have been previously victimized by crimes were less likely to report subsequent crimes to the police. The authors explained that following a v ictimization event, v ictims of crime may submerge into an increased isolation fro m the mainstream society, creating a defensive shield, which is perceived by the victims to serve as a mechanism against future victimizations [11]. Nonetheless, Zhang et al. maintain that crime-reporting behavior is exp lained by incident-specific correlates, indiv idual-specific correlates, and environmental-specific correlates.

Socio-Economic Status
Prior research shows that people who live in economically disadvantaged neighborhoods are less willing to cooperate with the police. Lack of cooperation, in this context, is man ifested in the form of unwillingness to report witnessed crimes or victimizat ion events to the police [17][18][19][20][21]. Thus, socio-economic status at the neighborhood level as well as at the individual level affects crime-reporting behavior. This effect is observed in many levels. Caracach [22] argued that people who find themselves in difficult financial situations (e.g., unemployed) are less likely to report crimes to the police[see also 23]. In addition to its effect on crime -reporting behavior, socio-economic status affects attitudes toward the police too, which in turn, increases the total effect that socio-economic status has on crime -reporting behavior [22]. Johnson in his study indicates that attitudes toward the police were the most negative among persons with inco me belo w $20,000, the unemployed, and non-homeowners. Moreover, economically d isadvantaged neighborhoods tend to have a higher level of crime co mpared to more economically developed neighborhoods [24]. In high-crime areas the police then are more likely to exercise coercive means, including police use of excessive force, which negatively affects both residents' attitudes toward the police and their crime-reporting behavior (17)(18)(19)(25)(26)(27)(28)(29).
At the individual level, research shows that lower-income persons, overall, are slightly less likely to report crimes to the police co mpared to high-inco me persons. However, this behavior varies by the type of crimes, i.e., property crime or violent crime. Goudriaan's study, for examp le, shows that property crime, especially, is mo re likely to be reported to the police by high-inco me families compared to lo w-inco me families [3]. Skogan's study also confirms that high-inco me families are more likely to report property crime [30]. According to Skogan, families of high-income reported about 14% more property crimes to the police compared to low-inco me families [30]. This tells us that, in most cases, crime -reporting behavior that is based on socio-economic status is explained by its consequences. That is, reporting certain property crimes to the police is done with intent of recovery, for insurance purposes, or in so me cases, as an expression of anger [3,4,31]. This, however, should not be confused with crime-reporting behavior that is based on the seriousness of crime. Vio lent crimes, on the other hand, are more likely to be reported by lower-inco me persons.
Lowe r-inco me v ictims of rape, for example, are mo re likely to report victimization events to the police than higher-inco me victims [3]. Skogan's study shows that low-inco me families report about 19% more violent crimes than high-income families (Skogan,[1,30,32]. Perhaps low-inco me families are mo re often vict imized by violent crimes than high-inco me families. Research confirms this assumption. Statistics show that low-inco me persons have higher victimization rates for violent crimes (47% for those who make $7,500 or less annually) than high-income persons (18% for those who make $75,000 or mo re annually) [33]. Conversely, high-inco me families are mo re often victims of property crime, simp ly because they possess more property that can be targeted by potential offenders [34]. Generally, research shows that the percentage of reported crimes against persons declines as the household income increases [35,32,11]. And the percentage of reported property crimes increases as the household inco me increases. Needless to say, these crime-reporting statistics vary from city to city.
Arguably, crime-reporting behavior is affected by one's socio-economic status. Yet, socio-economic status has an impact on one's attitudes toward the police, which in turn, affect one's willingness to report crimes to the police. In short, research shows that the effect of socio-economic status on attitudes toward the police is manifested in lower crime -reporting behavior. And the effect of socio-economic status on crime-reporting behavior varies by the type of crime (e.g., property crime vs, vio lent crime).

Seriousness of Crime
The decision whether or not to report a crime to the police, to a large extent, is affected by the severity or the consequences of the criminal event. Thus, individual perception about the seriousness of crime plays an important role in crime-reporting behavior [22,36]. Skogan argued that crime -reporting behavior is a form of indiv idual evaluation of the events in terms of cost-benefit rational [34]. Thus, according to Skogan, crimes that tend to produce some form of personal gain (e.g., property crimes) are more likely to be reported to the police [34,3,10]. 3 For examp le, those who have insured property, and that the property will be fully compensated by the insurance companies, are more likely to report property theft to the police (84%) co mpared to those who do not have property insurance (51%) [37]. However, the seriousness of crime does not stop at the value of property. In fact, it is much broader than property crime; it includes crimes against persons as well. Research shows that the severity of crime becomes a stronger determinant of crime -reporting behavior when including violent crimes (e.g., victim inju ries, repeat vict imization events, etc.) [38][39][40][41]35]. Tanton and Jones's study shows that victims of serious crimes have a higher propensity to report crimes to the police than victims of less serious crimes. Vict ims who had experienced injuries during an assault, for example, were associated with higher reporting of the event to the police [10,42]. According to the Bureau of Justice Statistics, 56% of victimizat ions that resulted in injuries were reported to the police co mpared to 40% of vict imization events that did not involve injuries [35]. Additionally, crime-reporting is further influenced by the degree of the severity of in jury (i.e., severe injuries are more likely to get reported than light injuries). The Bu reau of Justice Statistics's study shows that 70% of v ictims who received medical treatment for their injuries reported the event to the police compared to 46% of victims who did not receive medical t reatment for their injuries [35].
The impetus behind crime-reporting behavior for vio lent crimes is the victims' perception that they will have a greater benefit fro m police involvement. That is, if victims of v iolent crimes report their vict imization events to the police, it means that they can put a stop to their vict imization [43]. This argument is consistent with Skogan's [30,34] cost-benefit rational; a form of rational choice perspective in which victims of crime perceive crime-reporting as beneficial [44].
Furthermore, Watkins argued that the severity of crime differs by age [43]. That is, juveniles are more likely to witness or become victimized by less severe crime. Watkins's findings indicate that juveniles are less willing to report crime to the police, not entirely because of their age, but because of the seriousness of the crime [43]. This translates that juveniles are more likely to engage in less serious crimes; thus, their vict imization is not as severe as that of adults [33,30]. Additionally, people are more likely to report crimes when such events result in injuries and great property loss. For juveniles, property loss, in most cases, is not applicable[see 40, 45,46, for reviews].

The Present Study
In this study, we tested three research hypotheses about the influence of prior victimizat ion on crime-reporting behavior. Research shows that victimizat ion experiences have a positive influence on crime-reporting behavior. In this context, emp irical evidence suggests that those who have been previously victimized by crime are mo re likely to report crimes to the police [8,47,48]. Of the two types of victimizations (i.e., crimes against property vs. crimes against persons), as discussed above, research shows that victims of property crimes are more likely to report crimes to the police in general. Additionally, research also shows that as the severity of crimes and frequency of their occurrence increases, so does the likelihood of reporting those crimes to the police, especially crimes that involve some kind of weapon [5,8,10,13,22,36,37,45,46 see also 14,16,35]. However, what makes the inclusion of prior vict imization relevant to this study is that there is a number of studies that do not support the above conclusions[see 11,34,49,]. Thus, to clarify this issue, further testing is needed. To address the influence of prior vict imization on crime-reporting behavior, the following research hypotheses were tested: People's Crime-Reporting Behavior Ha (1): Indiv iduals who have been previously victimized by crime (regard less of the type of crime: e.g., property crime o r crimes against persons) are mo re likely to report victimization events or witnessed crimes to the police compared to those who have not been previously victimized by crime.
Ha (2): Individuals who have been victims of property crimes are more likely to report victimization events or witnessed crimes to the police compared to those who have not been victims of property crimes.
Ha (3): Controlling for socio-economic status, individuals who have been victims of crimes against persons are less likely to report vict imization events or witnessed crimes to the police co mpared to individuals who have not been victims of crimes against persons.

Participants
The data for this research study were collected in 2009 as part of a larger study on crime-report ing behavior. A 111-item survey questionnaire was distributed to a sample of 531 undergraduate university students (248 males and 283 females) in six colleges, who ranged in age fro m 18 to 52 years (M = 22, SD = 4.5). In terms of race/ethnicity, the majority of the participants identified themselves as Whites (72.3%), followed by African A mericans (16.1%), Asians (3.8%), Hispanic/Latinos (1.3%), and others (6.5%). To draw the sample, we adopted a two-stage cluster sampling procedure, stratified by colleges. That is, we drew six sub-samples, one from each college, assuring an equal representation of all undergraduate university students in the final samp le. The nu mber of participants in each sub-sample was based on the percentage of students enrolled in each of the six colleges in a large public university in Pennsylvania. Since in this study we test three distinctive hypotheses, the research results should be viewed and interpreted within the confines of the scope of this study.

De pe ndent Variables
The dependent variable in this study is people's willingness to report crimes to the police. To measure this crime -reporting behavior, we used three separate composite measures with a total of 24 mini-scenarios/items (see Appendix A). The respondents were asked to read each response and, based on their personal perception, they were asked to mark their answers on a five-point Likert-scale (strongly disagree = 1, disagree = 2, neutral = 3, agree = 4, and strongly agree = 5). In terms of reporting behavior by the type of crimes, these scales were designed to measure reporting of crimes against persons and reporting of crimes against property. Reporting of property crimes was measured using a 3-item Likert-scale with 1 to 5 response category (1 = strongly disagree to 5 = strongly agree). By the degree of the severity of crimes, these three scales measured the reporting of crimes fro m the least severe (e.g., s moking marijuana, selling illicit drugs, paint ing graffit i, etc.), mediu m-level crimes (e.g., physical threats, future terroristic threats, etc), and serious crimes (e.g., kidnapping, rape, murder, etc.), The pool of 24 crime-reporting items was subjected to factor analysis using SPSS version 19.0 (Statistical Package for Social Science). Initially, the factorability of the 24 crime -reporting items was examined. To determine the factorability of these items, we used two criteria, namely the Keiser-Meyer-Oklin Test and the Bartlett's Test of Sphericity. The Kaiser-Meyer-Oklin value, wh ich is a measure of the sampling adequacy, was .931, a value that greatly exceeded the minimu m reco mmended value of .60[see [50][51][52]. The Bartlett's Test of Sphericity [53] for this pool of crime-reporting items also reached the statistical significance of p < .000, which tells us that the analyses supported the elements that contributed to the results of the correlation mat rix[see 54,55]. In other words, the relationships between items in the scale were statistically significant.
Furthermore, the init ial results of the factor analysis that was conducted on the 24 crime-reporting items using the maximu m likelihood extraction with the varimax rotation indicated that there were four factors with eigenvalues exceeding 1.0, explaining 43.83%, 14.43%, 7.33%, and 4.53%o f the variance in crime-report ing behavior. After inspecting the scree plot, a three-factor solution was deemed suitable for further investigation. Thus, a second factor analysis with a forced extraction, limit ing the number of extracted factors to three, was performed. The first factor, Factor 1, was labeled "crime-report ing 3," the second factor, Factor 2, was labeled "crime-reporting 1," and the third factor, Factor 3, was labeled "crime-report ing 2." In this context, the factor analysis was determinant in the formation of three crime-reporting scales in this study. The items that were grouped in Factor 1 measured the reporting of serious crimes (e.g., kidnapping, rape, murder, etc.), items that were grouped in Factor 2 measured the reporting of less serious crimes (e.g., smo king marijuana, selling illicit drugs, painting graffiti, etc.), and items that were grouped in Factor 3 measured the reporting of med iu m-level crimes (e.g., physical threats, future terroristic threats, etc). The main idea here was to extract a min imu m number of factors that can explain the maximu m amount of variance in the crime -reporting behavior.
The internal consistency coefficients computed for each crime -reporting scale was adequate, .89, 87, and .94, respectively, indicating that these measures have excellent reliability for research purposes. The results of scree test for each crime-reporting scale, which are based on principal component analysis, suggest that all three crime-reporting scales are unidimensional.

Inde pende nt Vari ables
The goal of this analysis was to test the effect of prior victimization (e.g., v ictimization of crimes against persons and crimes against property) and socio-economic status on crime -reporting behavior. The variables that measured prior victimization were binary (yes = 1, no = 0). Victimizat ion for crimes against persons was measured by a 3-item index. This included crimes such as robbery, aggravated assaults, and non-physical (verbal) threats. For robbery, the question asked, "Has someone taken something fro m you by force or threat of force?" For aggravated assaults, the question asked, "Has someone assaulted or attacked you personally? (An attack can be anything from being hit, grabbed, to being shot at or beaten)." For non-physical threats, the question asked, "Has someone verbally threatened you in any way?" Vict imization for crimes against property, on the other hand, was measured by a 4-item index. This included burglary, vandalism, grand theft, and patty theft. Each variable had a binary response category (yes = 1, no = 0). For burglary, the question asked, "Has your house or apartment been broken into?" For vandalis m, the question asked, "Has your car been broken into or vandalized?" For grand theft, the question asked, "Has your car been stolen?" And for petty theft, the question asked, "Has anyone tried to steal or stole anything (other than your car) that belongs to you?" Socio-economic status was measured on a 5-point scale (poor = 1, lower middle-class = 2, middle-class = 3, upper middle-class = 4, and rich = 5). Other relevant variables, including demographic variab les (age, gender, and race/ethnicity), public interaction with the police variables, police behavior, attitudes toward the police, fear of criminal retaliation, and crime -reporting anonymity were control variables.

Control Variables
As noted, to effectively test three specific research hypotheses, using mult iple linear regression analysis, this study controls a number of variables, including police behavior, attitudes toward the police, public interaction with the police, fear of criminal retaliation, and crime-reporting anonymity (the desire to remain anonymous when calling the police to report a witnessed crime or victimizat ion event). In this study, police behavior was measured by a 22-item scale (strongly disagree = 1, disagree = 2, neutral = 3, agree = 4, and strongly agree = 5). This scale was highly reliable with Cronbach's Alpha = .92. Attitudes toward the police were measured using a 30-item Likert scale (Cronbach's Alpha = .94). Fear of criminal retaliation was measured by a one single item with 1 to 5 response category (refer to Table 3 for the list of variab les). We measured crime-reporting anonymity using a 4-item Likert scale (Cronbach's Alpha = .71) 4 . Additionally, we controlled for the effect of a number of demographic variables such as age, gender, and race/ethnicity. Age was measured in years. Gender was a dichotomous variable (male coded = 1, female coded = 0). Race/ethnicity was also measured as a dichotomous variable, with whites being the base category. After the recoding procedure took place, we created three categories for race: Black (yes = 1, no = 0), Asian (yes = 1, no = 0), Other (yes = 1, no = 0), with Whites selected as the base category.
In terms of interpreting the coefficients for the scales, a higher score on the police behavior scale, for instance, indicates the presence of police misconduct, and a lower score indicates the absence of police misconduct. A higher score on the crime-reporting anonymity scale indicates that respondents were concerned with their identity being revealed, whereas a lo wer score indicates the respondents were not concerned with anonymity when reporting crimes to the police. A higher score on attitudes toward the police scale indicates positive attitudes toward the police, whereas a lower score indicates disfavorable attitudes toward the police.
Among other control variables in this study was public interaction with the police. There were five variables that were used to measure public interaction with the police. This interaction was measured in terms of the quantity and quality of contacts with the police. The quality of contacts with the police refers to the types of contacts people had with the police, namely police-initiated and citizen-in itiated contacts. To measure police-init iated contacts, we used a 12-item list of possible min i-scenarios pertaining to police-in itiated contacts. To measure citizen-in itiated contacts with the police, we used a 9-item list of possible mini-scenarios pertaining to citizen-init iated contacts. Each item had a binary response category (Yes = 1, No = 0). In some instances, people may not have contacts with the police, but they still have an opinion about the police. Weitzer and Tuch, for example, contend that some people create their opinion about the police based on what they see on television [56,57]. To capture this influence, we used a one 4-point Likert item ranging fro m 0 (never) to 4 (o ften) that asked respondents to indicate how often they heard or read about police misconduct on TV, radio, newspapers, internet, etc., a survey item that was borrowed fro m Weit zer and Tuch's study [56]. This single item was designed to measure the influence of med ia exposure on people's decisions whether or not to report witnessed crimes or victimization events to the police; a vicarious form of contact with the police.

Social Desirability
To determine whether or not the respondents have introduced personal bias in their answers to other inventories (scales), for validity purposes, a 12-item personal reaction inventory (scale) with b inary response categories (Yes = 1, No = 0) was used (see [58][59][60]. People have a tendency to over-report or under-report activit ies that are considered to be socially or cu lturally desirable or undesirable [61][62][63]. The concern with this effect is that the respondents who tend to respond in more socially desirable ways are mo re likely to indicate they will report crimes to the police when in fact the People's Crime-Reporting Behavior reality is otherwise. This 12-item scale has a Cronbach's Alpha of .66, wh ich is considered a min imally acceptable level of reliability [62]. A higher correlat ion between social desirability scale and other scales indicates that the respondents have introduced some b ias in their answers. Conversely, a lower correlation indicates that the respondents have answered truthfully. The correlation analysis shows that the effect of social desirability bias on respondents concerning the truthfulness of their responses was minimal (r <.03), which adds to the validity and reliability of the research findings in this study.

Results
Hypothesis 1 predicted that individuals who have been previously victimized by crime, regardless of the type of crime (e.g., property crime or crimes against persons) are more likely to report vict imization events or witnessed crimes to the police co mpared to those who have not been victimized by crime. In other words, prior vict imization has a positive effect on crime-reporting behavior. To evaluate the effect of prior vict imization on crime -reporting behavior, we used a one-factor between-groups analysis of variance. The analyses in Table 1 show that those who had been previously victimized by crime were slightly more likely to report less serious crimes (M = 23.28, SD = 8.516, p < .284) co mpared to those who have not been previously victimized by crime (M = 22.34, SD = 8.739). Addit ionally, prior vict imization had a positive effect on the reporting of med iu m-level crimes (M = 22.52, SD = 5.632, p < .533) and serious crimes (M = 43.42, SD = 9.186, p < .733). However, the mean differences in crime -reporting behavior fo r those victimized by crime and those not victimized by crime were statistically insignificant. Therefore, hypothesis 1 is rejected.
Hypothesis 2 predicted that individuals who have been victims of property crimes are mo re likely to report victimization events or witnessed crimes to the police compared to those who have not been victims of property crimes. This hypothesis has not been supported by the data in this study at either the bivariate level (Table 2) or the mu ltivariate level (see Table 3). It is noteworthy that the data in this study suggest that being a        Hypothesis 3 predicted that when controlling for socio-economic status, individuals who have been victims of crimes against persons are less likely to report vict imization events or witnessed crimes to the police co mpared to individuals who have not been victims of crimes against persons. To test hypothesis 3, we used hierarchical regression analysis. Since we were concerned that socio-economic status might be related to both prior victimization and crime-reporting behavior, it seemed mo re appropriate to place statistical controls for this effect. Thus, the effect of prior vict imization on crime-reporting behavior is independent of the effect of SES.
The socio-economic status was entered in step 1 (o r Model 1), explaining only 0.1% of the variation in the reporting of less serious crimes, 0.1% in the reporting of mediu m-level crimes, and 0% in the report ing of serious crimes. Prior victimization was entered in step 2 (or Model 2). After controlling for the influence of socio-economic status, the R 2 change for crime-reporting 1 was .009, F change (1, 528) = 4.561, p < .05. This means that prior victimization accounted for appro ximately 1% of the variat ion on the report ing of less serious crimes (crime-reporting 1, see Table 4). For the reporting of med iu m-level crimes (crime -reporting 2), the R 2 change was .000, F change (1, 528) = .191, p < .663 (Tab le 5). And for the reporting of serious crimes (crime-reporting 3), the R 2 change was .000, F change (1, 528) = .129, p < .720 (see Table 6). This means that the influence of prior victimization is greater for the reporting of less serious crimes (exp laining 1% of the variation), but this influence tends to decrease as the seriousness of crime increased, explaining approximately 0% of the variat ion in the reporting of mediu m-level crimes and 0% of the variation in the reporting of serious crimes.
In terms o f the d irect ion of the influence of prior victimization on crime-reporting behavior, the analyses in Tables 4, 5, and 6 show that prior vict imization has a negative effect on crime-reporting behavior (b = -1.655 for less serious crimes, b = -.224 for mediu m-level crimes, and b = -.285 for serious crimes). This means that people who have previously been victims of crimes against persons are less likely to report crimes to the police. Conversely, people who have not previously been victims of crimes against persons are more likely to report crimes to the police. However, the effects of prio r vict imization (for v ictims of crimes against persons) on crime-report ing behavior reached statistical significance of p < .05 only for the reporting of less serious crimes. On the other hand, being a victim of crimes against persons did not have a statistically significant effect on the reporting of med iu m-level crimes and serious crimes. Therefore, hypothesis 3 is rejected.

Discussion and Conclusions
The effect of prior vict imization on crime-reporting behavior largely depends on intervening variables. That is, prior vict imization by itself is not a strong determinant of crime -reporting behavior. So me researchers suggest that crime -reporting behavior is affected by the victim's prior experience with the police, by whether or not an arrest is made by the police in an effort to investigate the crime that has affected the victim, and by the police response to the individual's own prior vict imization rather than victimization of another household member[see 8,12]. This tells us that the victim's decision whether or not to report crimes or v ictimization events to the police is, to so me degree, affected by external factors (e.g., police behavior), factors other than victimization events. However, external factors such as police behavior may have positive or negative effects only on the vict im's mot ivation to report crimes to the police. Needless to say, motivation is not the only necessary element that determines one's crime-reporting behavior. Somet imes, victims of crime are forced to put a stop to, for instance, repeat victimizations [43]. Th is means that, regardless of the negative effects of external factors (e.g., police behavior, fear of criminal retaliation, etc), the decision to report crimes to the police can be affected purely by the victim's perception of the event, which in most cases includes taking into account the seriousness of the victimization events, the type of victimizat ion events (i.e., being a vict im of p roperty crimes vs. being a v ictim of crimes against persons), and the relationship between the victim and the offender[see 5,11,22,34,45,53,64-66, for reviews].
The current study is limited to examining whether or not prior vict imization is a strong predictor of crime-reporting behavior. To address this issue, three research hypotheses were developed and tested in this study. Hypothesis one predicted that individuals who have been previously victimized by crime -regard less of the type of crime (e.g., property crimes or crimes against persons) -are mo re likely to report witnessed crimes or victimizat ion events to the police. The findings of this study suggest that being victimized by a crime had an insignificant effect on crime -reporting behavior. In other wo rds, the data in the current study suggest that the decision whether or not to report a crime to the police is not affected by prior victimization status.
To further examine the effect of prior victimizat ion, this time by the type of victimization events (i.e., by looking at the difference between victims of p roperty crimes and those who never been victims of property crimes), another research hypothesis was tested. In this context, hypothesis two predicted that individuals who have been victims of property crimes are mo re likely to report crimes to the police.
The results of the current study show that being a victim of property crime had an insignificant effect on crime-reporting behavior. By the severity of crimes, this effect was statistically insignificant for all three levels of reporting (i.e., the reporting of less serious crimes, mediu m-level crimes, and serious crimes). Additionally, the effect of vict imization for property crime on crime-reporting behavior did not reach the statistical significance of p < .05 at the mu ltivariate level either. In terms of the positive o r negative d irection of the effect, the research findings of the current study suggest that victims of property crimes are generally more likely to report less serious crimes and serious crimes, but less likely to report mediu m-level crimes.
The last research hypothesis tested in this study predicted that individuals who have been victims of crimes against persons are less likely to report witnessed crimes or victimization events to the police co mpared to those who have not been victims of such criminal events. At the bivariate level of the analysis, the research results of the current study somewhat support this hypothesis. Although the effect was insignificant, the current study suggests that individuals who have been victims of crimes against persons are less likely to report crimes to the police. Unexpectedly, this finding changed when tested at the multivariate level. Thus, the mult ivariate analyses show that being a victim of crimes against persons has a positive effect on crime -reporting behavior.
In summation, this study is limited to addressing the effect of prior victimizat ion on crime-reporting behavior by 1) the type of victimization (i.e., being a victim o f property crimes or a vict im of crimes against persons), 2) being a victim of a crime, regardless of the type of crime versus not being a victim, and 3) the difference in crime-reporting variation between the two. Overall, the findings of the current study suggest that prior v ictimization cannot be considered a good predictor of crime -reporting behavior. Although, an attempt has been made to clarify contradict ing results in previous studies, more research is needed in this area. In this study, for example, we did not examine the relationship between the victim and the offender. Prior literature suggests that the victim-offender relationship is one of the major factors that affect a victim's crime-reporting behavior. Along this line, there are a number of studies showing that an increased social distance between the victim and the offender results in an increased reporting of witnessed crimes and victimization events to the police [35,42,45,[67][68][69], see also 30,70]. Bach man, for examp le, found that victims of rape were mo re likely to report v ictimization events to the police when the perpetrator was unknown to them [45,46,68]. In this context, research shows that most sexual assaults, including rape, are committed by someone known to the victim . This tells us that a great number of vio lent crimes (e.g., rape and sexual assault) go unreported, which can be attributed to the victim-offender relationship. Future research should fill this gap by exploring the effect of victim-o ffender relationship on crime-reporting behavior. Additionally, the seriousness of crime has also been identified as a key factor that affects crime -reporting behavior. Thus, future research should examine whether or not there is a strong association between the seriousness of the offense and the victim-offender relationship, and to what extent these two factors affect crime -reporting behavior. Future research should also examine the difference in the reporting o f v iolence committed by strangers and violence committed by less intimate known offenders (e.g. co workers, friends, and relatives).