Fear from a distance: testing a new model of psychological distance and fear of crime

ABSTRACT

This study tests a new model of fear of crime. We hypothesise that community members who perceive crime as psychologically proximal (e.g. likely to happen soon in their immediate environment) will experience more intense feelings of worry about crime. We analyse survey data collected from a convenience sample of N = 719 residents from Queensland, Australia. In support of our hypothesis, psychological distance from crime explains a total of 58% of the variance within worry about crime at a statistically significant level (t = −20.14, p < .001). Our findings provide an empirical foundation for fear of crime reduction strategies designed to increase community members’ psychological distance from crime. Future directions are discussed.

KEYWORDS:

• Construal level theory
• psychological distance
• fear of crime
• risk perception
• victimisation risk

Introduction

It is well known that individuals are capable of worrying about crime in the absence of immediate threats in their environment (Chadee et al., 2007; Gouseti & Jackson, 2015). However, theoretical explanations for this phenomenon are lacking in the existing literature on fear of crime. One plausible reason is that individuals can cognitively transcend their ‘here and now’ and react to crime events that are not in their immediate surroundings. These ideas align with Trope and Liberman’s (2010) Construal Level Theory of Psychological Distance (CLT). Scholars have recently begun to explore whether the ideas of CLT can be applied to understand how people estimate their risk of becoming a victim of crime (Gouseti, 2016; Mellberg et al., 2022). This work has found that individuals who are most worried about crime tend to construe crime events as close to them in time, space, and likelihood. Individuals who are least worried about crime tend to construe crime events as distant from them in time, space, and likelihood.

Despite the above work, there are several questions that remain unanswered about how psychological distance applies to fear of crime in the real world. Proponents of CLT assert that temporal, spatial, social, and hypothetical distance are distinct but highly correlated dimensions of psychological distance (Trope & Liberman, 2010). In the context of fear of crime, it is not known whether each dimension of psychological distance from crime are equally salient to community members’ worry about crime or whether certain dimensions play a more important role in shaping worry about crime. Addressing this gap in knowledge provides a basis to develop new insights on how fear of crime develops within communities and potential ways to address its deleterious effects. The current study addresses this shortcoming in the fear of crime scholarship by testing a new theoretical model of psychological distance and worry about crime using data collected from N = 719 participants living in Queensland, Australia.

Review of literature

Many scholars define fear of crime as a negative emotional reaction to a stimulus perceived as threatening (Hale, 1996; Hart et al., 2022; Henson & Reyns, 2015). Interestingly, the stimulus perceived as threatening is rarely an actual instance of crime (Chadee et al., 2007; Jackson, 2006). Instead, people may respond to a variety of physical and social stimuli within their immediate environment. For example, the perception of incivility (i.e. physical and social disorder) has been known to heighten perceptions of risk and increase feelings of worry about crime (Chataway & Hart, 2016; Jackson, 2004). Perceptions of incivility are thought to heighten perceptions of risk by signalling to the individual that crime could be nearby (Brunton-Smith & Sturgis, 2011; Hunter, 1978). These ideas suggest that people are capable of reacting to stimuli which are not a direct threat of victimisation, but are instead indicative of crime and its proximity to the individual. Indeed, several scholars have noted the weak relationship between crime rates in an area and fear of crime (Ferraro, 1995; Hale, 1996). For these reasons, scholars have concluded that risk perception is subjective and largely independent of the actual prevalence of crime in an area (Chadee et al., 2007). These subjective perceptions of victimisation risk are often higher than one’s actual risk of victimisation and have been associated with increased feelings of concern and worry about crime (Chadee et al., 2007; Jackson, 2006).

Addressing these subjective perceptions of risk is important, since community members who worry about crime may also experience poorer physical and mental health outcomes, impeded social functioning, and an overall decreased quality of life (Gray et al., 2011; Lee et al., 2020; Stafford et al., 2007). The negative consequences associated with worrying about crime can also coalesce at a neighbourhood-level. Neighbourhoods where fear of crime is concentrated tend to have weaker social cohesion, lower informal social control, and a higher rate of perceived disorder (Brunton-Smith & Sturgis, 2011; Franklin et al., 2008; Scarborough et al., 2010). To address fear of crime more effectively, it is critical to understand what mechanisms underpin subjective risk perception and lead to people becoming worried about crime in situations where crime is not occurring or imminent (Gouseti & Jackson, 2015). However, there is a dearth of theoretical explanations for how individuals become worried about crime in situations where there is no immediate threat of victimisation.

Under-theorisation of fear of crime

The under-theorisation of fear of crime has been noted as a long-standing issue within the scholarship (Gouseti, 2016; Hale, 1996; Jackson, 2004, 2006; van der Wurff et al., 1989). Rather than investigating and theorising the complex cognitive processes involved in becoming worried about crime, scholars have tended to focus on building data-driven frameworks to identify correlates of fear of crime (Gouseti, 2016; Jackson, 2004). Most commonly, variables associated with risk perception are identified from survey data and placed into ad hoc frameworks to test whether they are correlated with worry about crime (Jackson, 2004; van der Wurff et al., 1989). These frameworks are considered to be data-driven because there are no theoretical justifications for the inclusion of each variable (van der Wurff et al., 1989). Furthermore, these frameworks often lack hierarchy, meaning that constructs are presumed to have direct and equal influence on worry about crime (van der Wurff et al., 1989). Data-driven frameworks are effective at identifying whether certain variables are associated with worrying about crime, but they are unable to fully explain how these variables reflect the complex theoretical processes involved in becoming worried about crime (Jackson, 2004). The absence of hierarchy also limits the ability for these models to explain the processes underlying risk perception and fear of crime (i.e. the order of each cognitive process and how these contribute to the development of fear of crime).

As an alternative to data-driven approaches to understanding fear of crime, theory-driven approaches allow researchers to better explain how and why certain cognitive processes shape individuals’ worry about crime. Although theory-driven investigation into fear of crime has been increasing in prevalence (Gabriel & Greve, 2003; Gouseti, 2016; Jackson, 2004), there are still limited explanations for how people become worried about crime in situations absent of a direct threat of victimisation. Recently, researchers have suggested that CLT might help to explain how people estimate their risk of victimisation based on their perceptions of proximity to crime (Gouseti, 2016; Gouseti & Jackson, 2015).

Construal level theory of psychological distance

CLT describes how individuals mentally travel beyond their current situation to experience and react to imagined events that are not actually occurring (Trope & Liberman, 2010). There are two distinct but interrelated cognitive processes involved in CLT (Trope & Liberman, 2010). The first is construal level, which describes how detailed or abstract an individual’s mental representation of an event is (Trope & Liberman, 2010). Events construed at a low level are represented as concrete and vivid, with specific details (e.g. being physically attacked by a stranger in the street). Events construed at a high level are represented as abstract, with limited details (e.g. assault).

The second cognitive process involved in CLT is psychological distance. Psychological distance describes how near or far one’s mental representation of an event is perceived to be from the individual’s ‘here and now’ (Trope & Liberman, 2010). According to the theory, the psychological distance of an event can be perceived through four dimensions: temporal, spatial, social, and hypothetical distance. These dimensions refer to ‘ … the perception of when an event occurs, where it occurs, to whom it occurs, and whether it occurs’ (Trope & Liberman, 2010, p. 442). Each dimension of psychological distance is argued to represent a distinct but interrelated cognitive process (Trope & Liberman, 2010). When an event is perceived as proximal along one dimension of psychological distance, it is also perceived as proximal along the other dimensions (Soderberg et al., 2015). For example, if an individual perceives victimisation as a threat in their current location (spatially proximal), then they would perceive victimisation as more likely (hypothetically proximal), more imminent (temporally proximal), and a direct threat to themselves (socially proximal).

Trope and Liberman (2010) assert that psychological distance and construal level share a bidirectional causal relationship. When an event is perceived as psychologically proximal, it is construed with concrete and specific details (Soderberg et al., 2015). Conversely, mentally representing the concrete details of an event (low-level construal) will evoke the perception of that event as psychologically proximal (more likely to occur to the individual in their current situation; Soderberg et al., 2015). Table 1 summarises the interaction between construal level and psychological distance.

Table 1. Trope and Liberman’s (2010) construal level theory of psychological distance.

Close events Construed at a low level	Distant events Construed at a high level
Concrete, vivid, context-dependent	Abstract, vague, limited context
Now, soon	Later, never
Here, nearby	Elsewhere, far away
To me, to people I know	To others, to people I don't know
Likely, realistic	Unlikely, unrealistic

Fear of crime as a proximal threat

In the context of crime, the cognitive processes involved in mentally transcending one’s current situation might explain how individuals perceive a risk of criminal victimisation when there is no credible threat of crime in their immediate surroundings. An individual’s construal level of victimisation has been described as crime construal (Gouseti, 2016). Low-level crime construal involves a concrete, vivid, and detailed mental representation of victimisation. Conversely, high-level crime construal involves an abstract and undetailed mental representation of victimisation (Gouseti, 2016). An individual’s perceived psychological distance from crime should, according to CLT, influence how they mentally represent crime. Individuals who experience low-level crime construal (i.e. vivid thoughts about victimisation) and perceive crime as psychologically proximal (i.e. likely to happen to them in their current location) are theorised to become worried about crime (Gouseti, 2016; Mellberg et al., 2022).

To date, Gouseti (2016) and Mellberg et al. (2022) have measured and examined different components of CLT in the context of fear of crime. Gouseti (2016) conducted a 2 × 3 experiment with 300 participants who were randomly assigned to one of six possible conditions. Level of crime construal was manipulated across two levels (high-level construal and low-level construal), while spatial psychological distance was manipulated across three levels (proximal, neutral, distal). The outcome variable was intensity of worry about crime measured using a 4-point scale. The results of this experiment indicated that participants in the high-level crime construal condition reported a lower intensity of worry about crime (Gouseti, 2016). Furthermore, when controlling for crime construal, participants in the psychologically distal condition also reported lower levels of worry about crime compared to those in the psychologically proximal and neutral conditions (Gouseti, 2016).

Expanding upon Gouseti’s (2016) research, Mellberg et al. (2022) piloted survey measures of psychological distance under non-experimental conditions. Using a convenience sample of 265 residents living in Townsville, Australia, Mellberg et al. (2022) piloted a survey instrument designed to capture perceptions of psychological distance from crime in the community. The survey items demonstrated good reliability and factorial validity, suggesting that community members’ psychological distance from crime could be measured under non-experimental conditions. Each dimension of psychological distance, in line with CLT (Soderberg et al., 2015; Trope & Liberman, 2010), was found to be highly correlated (Mellberg et al., 2022). Furthermore, the results from Mellberg et al.’s (2022) regression analysis demonstrated that those participants who perceived crime as a temporally, socially, and hypothetically proximal threat reported higher levels of worry about crime. However, the authors noted that the survey measures required further refinement (e.g. perceptual measures of spatial distance).

Gaps in existing knowledge and need for further study

As noted in the review of literature above, there are limited theoretical explanations within the fear of crime scholarship for how individuals become worried about crime in situations absent of a direct threat of victimisation. A coherent theoretical framework is needed to understand how psychological distance from crime and each of its dimensions are related to community members’ worry about crime in a real-world setting (i.e. under non-experimental conditions). Existing research exploring psychological distance and fear of crime has been performed under experimental settings (Gouseti, 2016) or with a small sample as part of a pilot study (Mellberg et al., 2022). Despite the emerging research on CLT and fear of crime, there is no established theoretical model of psychological distance and worry about crime which recognises the complex relationships between temporal, spatial, social, and hypothetical distance from crime.

The current study builds and tests a new model of psychological distance and fear of crime. Based on the assumptions outlined in CLT, community members’ psychological distance from crime should comprise their temporal, spatial, social, and hypothetical distance from crime. Furthermore, each dimension of psychological distance from crime should also be highly correlated, such that when crime is perceived as proximal along one dimension, it is also perceived as proximal along the other dimensions (Soderberg et al., 2015; Trope & Liberman, 2010). We propose that these theoretical assumptions are best reflected by the hypothesised model of psychological distance and worry about crime schematically represented in Figure 1. Similar approaches to modelling the effects of psychological distance have been used in previous research applying CLT (Azadi et al., 2019; Blauza et al., 2021; Sharma et al., 2022). We further hypothesise that community members who perceive crime as psychologically proximal (i.e. that crime is likely to happen to them soon in their current location) will express a higher intensity of worry about crime. By testing this new model, the current study expands scholarly understanding of how psychological distance shapes worry about crime and fills important gap in the literature about human risk perception formation. We also provide an empirical basis for managing fear of crime within the community by targeting proximal perceptions of crime.

Figure 1. Theoretical model of psychological distance and worry about crime. Note. +/− denotes hypothesised direction of relationships among latent variables.

Methodology

Current study

The current study tests a new model of psychological distance and worry about crime. First, we detail our approach to data collection and outline our refinements to perceptual measures of psychological distance based on previous research. Then, we confirm the structure of our hypothesised model using confirmatory factor analysis (CFA) and test the assumption that temporal, spatial, social, and hypothetical distance from crime are distinct but highly correlated dimensions of psychological distance from crime (Soderberg et al., 2015; Trope & Liberman, 2010). Finally, we estimate a structural regression model to test our hypothesis that psychological proximity to crime will predict a substantial proportion of variance within community members’ worry about crime at a statistically significant level (p ≤ .05).

Sample

The present study received approval from the Queensland University of Technology Human Research Ethics Committee (QUT Ref.: LR 2021-4865-5792). Data were collected from a convenience sample of N = 1063 residents living in Queensland, Australia using the Qualtrics survey platform (Qualtrics, Provo, UT, USA). Participants were recruited using targeted advertisements on social media (i.e. Facebook, Instagram, and Messenger).¹ This recruitment method was chosen because it is cost-efficient and allows researchers to sample specific audiences according to demographic characteristics of interest across large geographical areas (Liu & Mattila, 2017; Ramo & Prochaska, 2012). The advertisement ran for 50 days. The advertisement appeared in 43,442 newsfeeds with 1,242 unique link clicks, for a cost per link click of 0.65 AUD.

The data were screened for missing cases. Missing cases constituted participants who did not complete all survey items pertaining to worry and psychological distance. Despite more than 5% of the sample containing missing responses, a missing values analysis revealed that there were no patterns within the missing data (ꭓ² = 314.34, df = 312, p = .452; Little, 1988). Missing data (n = 337 cases) were subsequently deleted listwise from the dataset to allow for CFA models to be estimated with complete data (Schafer & Graham, 2002). The final dataset consisted of N = 719 participants.

Demographic information about the sample is provided in Table 2. To assess representativeness, the sample was compared to census data available for the population of Queensland (Australian Bureau of Statistics, 2021). According to the most recent available census data for Queensland, the population was comprised of 50.7% women and 49.3% men, the median age was 38, and 46.5% of the population were married (ABS, 2021). Participants were aged between 18 and 87 (Mdn = 49, SD = 13.81), slightly older than the population of Queensland. The current sample also comprises more women (58.8%) and fewer individuals who report being married at the time of the survey (43.7%). Regarding country of birth, the most common country of birth according to census data was Australia (71.4%), New Zealand (4.0%), England (3.7%) and India (1.4%; ABS, 2021). The current sample comprises slightly more individuals born in Australia (75%) and individuals born in England (8.6%). Overall, our sample is reasonably representative of Queensland, being only slightly overrepresented by older individuals and women.

Table 2. Demographic characteristics for participants (N = 719).

Characteristic	n	%
Gender
Woman	423	58.8%
Man	296	41.2%
Sexuality*
Heterosexual or straight	233	32.4%
Gay or homosexual	7	1.0%
Lesbian	1	0.1%
Bisexual	8	1.1%
Pansexual	2	0.3%
Asexual	6	0.8%
Queer	2	0.3%
I use a different term	3	0.4%
Relationship Status*
Single	168	23.4%
In a relationship	150	20.9%
Married	314	43.7%
Widowed	20	2.8%
Separated	17	2.4%
Divorced	56	7.8%
Highest Education Level
Did not complete year 12	67	9.3%
Completed year 12 or equivalent	165	22.9%
Bachelor’s Degree	269	37.4%
Master’s Degree	113	15.7%
Doctorate Degree	27	3.8%
Other^	78	10.8%
Country of Birth
Australia	539	75.0%
England	62	8.6%
India	1	0.1%
New Zealand	30	4.2%
China	1	0.1%
Italy	2	0.3%
Philippines	1	0.1%
South Africa	6	0.8%
Malaysia	3	0.4%
Other	74	10.3%

Note. Demographic information can only be reported for participants who provided responses to the questions above. * Denotes participants could select multiple options. ^ ‘Other’ included trade certificates I through IV.

Measures

Survey questions were adapted from Gouseti (2016) and Mellberg et al. (2022). Mellberg et al. (2022) examined the scaling properties and factorial validity of psychological distance and worry about crime, finding that each dimension of psychological distance was empirically distinct. However, the researchers recommended that spatial distance items ‘I believe I will fall victim to [a crime] in a location near me’ could be refined to include the egocentric point of reference (Mellberg et al., 2022). The current study, therefore, adapted spatial distance items to instead ask participants to rate their level of agreement to the following statement ‘I believe I might fall victim to [a crime] in my current location’, This phrasing is supported by other research on fear of crime (e.g. Engström & Kronkvist, 2021). Asking participants about crime perceived relative to their immediate environment also aligns with the ego-centric point of reference (i.e. one’s ‘here and now’), as discussed above in the literature review.

Personal crime

The three crime types included in the worry and psychological distance items were: (a) falling victim to a physical attack; (b) falling victim to harassment and/or verbal abuse; and (c) falling victim to a mugging. These crimes are forms of personal crime, which may involve a direct threat to the self in the current situation (i.e. one’s ‘here and now’). Personal crimes were preferred because they do not involve a spatial reference point (e.g. burglary which occurs in the home). Avoiding spatial reference points was important to ensure participants’ psychological distance from crime was not influenced by the phrasing of survey items.

Worry

The outcome variable was worry about crime. Participants were asked how worried they felt about falling victim to three different types of crime. Specifically, How worried are you about falling victim to [a crime]? Where 1 = Not at all worried; 2 = A little bit worried; 3 = Fairly worried; 4 = Very worried (M = 1.8; SD = 0.7). This intensity measure of worry has been validated across several geographical contexts including in the United States (Brunton-Smith & Sturgis, 2011; Scarborough et al., 2010), the United Kingdom (Gouseti, 2016; Jackson & Gray, 2010), and in Australia (Chataway et al., 2019; Chataway & Mellberg, 2021). Intensity measures of worry were used because they do not use temporal reference points (i.e. over the last month, etc.) that could influence one’s perceptions of psychological distance and the proximity of crime to themselves (Gouseti, 2016).

Psychological distance

Temporal, spatial, social, and hypothetical distance from crime were conceptualised as unique dimensions of psychological distance from crime. To measure perceptions of psychological distance, scale items adapted from Mellberg et al. (2022) were used. Participants were asked how much they agree or disagree with statements about their perceived distance of crime from themselves on a 5-point scale where 1 = Strongly Disagree, 2 = Disagree, 3 = Neither Agree nor Disagree, 4 = Agree, and 5 = Strongly Agree. The statements related to each dimension of psychological distance and were reverse coded so that a high score would indicate a psychologically distal perception of crime. Temporal distance: I believe that I might fall victim to [a crime] in the near future (M = 3.4; SD = 1.0). Spatial distance: I believe that I might fall victim to [a crime] in my current location (M = 3.6; SD = 1.1). Social distance: I believe that [crime] happens to people who are similar to me (M = 2.8; SD = 1.0). Hypothetical distance: I believe that it is likely that I will fall victim to [a crime] (M = 3.4; SD = 1.0). The complete instrument containing descriptive statistics for each item is available in the supplementary material. Block randomisation was used so that the survey items corresponding to each dimension of psychological distance were presented in a random order. Within-block randomisation of psychological distance items was also used so that each of the three crime types were presented in a random order. Both methods of randomisation were implemented to minimise the influence of order effects on participants’ responses (Sanjeev & Balyan, 2014).

Analytic plan and model specification

First, confirmatory factor analyses were performed to assess the fit of the proposed measurement model (see Figure 2) to the data and the factorial validity of the survey measures of psychological distance from crime. After confirming the fit and structure of the measurement model, we estimated a structural regression model to assess the relationship between psychological distance and worry about crime (see Figure 3). For model identifiability, the regression weights for each of the first observed variables in the first-order factor were arbitrarily constrained to 1.0 (Byrne, 2010; Kline, 2015). For the higher-order portion of the measurement model, all regression weights between the first-order factors and the higher-order factor were freely estimated to assess psychological distance from crime’s suitability as a higher-order factor (Kline, 2015; Koufteros et al., 2009). For the structural model, the pathway from psychological distance and the highest performing higher-order factor loading (temporal distance from crime) was constrained to one (Chen et al., 2005). No substantial differences were observed when alternative factors were constrained.

Figure 2. Hypothesised measurement model of psychological distance from crime.

Figure 3. Hypothesised structural regression model of psychological distance from crime and worry about crime.

SEM was performed using IBM Statistics Package for the Social Sciences (SPSS) 29.0 and Analysis of Moment Structures (Amos) to assess the fit of the hypothesised model. All models were estimated using Maximum Likelihood (ML) estimation and variance-covariance matrix. Although the data were ordinal and categorical rather than continuous, the items contained more than three response categories (Byrne, 2010). Assessment of normality revealed some departure from univariate normality, with skewness and kurtosis values for the items falling between −0.65 and 0.80 and −1.3 and 0.44, respectively (Kline, 2015). However, the data demonstrated a significant departure from multivariate normality (kurtosis = 69.22; Byrne, 2010). Although Asymptotically Distribution-Free estimation has been recommended for non-normal data, this technique requires large sample sizes (Byrne, 2010; Kline, 2015). Instead, bootstrapping was performed on the data (Byrne, 2010). Bootstrapped samples were set to 500 with bias-corrected confidence intervals of 95%. Analysis of multivariate outliers via Mahalanobis Distance revealed three cases with statistically significant differences (p < .001). Further investigation of these cases did not reveal any non-genuine responses, response errors, or any substantial changes to interpretation of final results. Outliers were therefore retained for further analyses.

Results

First, we examined whether our hypothesised model of psychological distance from crime exhibited good fit to the data. The results from the CFA are presented in Figure 4. The measurement model demonstrated inadequate fit across exact fit indices after error terms were allowed to covary (χ² = 191.60, df = 39, p < .001). When approximate fit indices were considered, the model demonstrated an acceptable fit (RMSEA = .07 [90% CI = .06, .08]; SRMR = .04; CFI = .98; GFI = .96; TLI = .97; NFI = .98).² The standardised factor loadings indicated that the items and their latent variables have reasonably good factorial validity. Although some of the standardised factor loadings fell slightly below .70, none of the lower 95% confidence intervals fell below .50, all estimates were statistically significant (p < .001), and the average variance extracted (AVE) for each latent variable exceeded the .50 threshold recommended by Fornell and Larcker (1981).

Figure 4. CFA results. Note. *** Denotes statistically significant result (p<.001). AVE_{Psychological distance}=.65; AVE_{Temporal distance}=.61; AVE_{Spatial distance}=.62; AVE_{Social distance}=.60 AVE_{Hypothetical distance}=.61.

Next, we investigated whether psychological distance from crime could explain sufficient variation within the survey items measuring temporal, spatial, social, and hypothetical distance from crime (Credé & Harms, 2015). There was strong evidence that psychological distance from crime was a suitable higher-order factor of temporal, spatial, and hypothetical distance from crime. Psychological distance from crime accounted for an average of 46.27% of the variance within the observed variables (i.e. raw scores from the survey items). However, for the items measuring social distance, the higher-order factor of psychological distance from crime only accounted for an average of 21.24% of the variance within these scores. The first-order factor of social distance retained substantial explanatory power, accounting for an average of 45.78% of the variance within participants’ scores across these items. Because there is no recommended threshold for the proportion of variance within observed variables that should be accounted for by a higher-order factor (Credé & Harms, 2015), it is recommended that these values are compared with future research replicating this model (see discussion).

Finally, to assess the reliability of the scale used to measure the constructs contained within the measurement model, higher-order-omega (ω_bo) was calculated (Flora, 2020). Although there is no universal cut off for ω coefficients (Cho & Kim, 2015), the scale measuring psychological distance from crime demonstrated satisfactory reliability, with an ω_bo value of .85. Overall, we retained the measurement model for further analysis, noting the shortcomings regarding social distance items.

Structural regression model

Having determined that our hypothesised model of psychological distance from crime was suitable, the final step of our analysis was to determine whether perceptions of psychological distance shaped worry about crime. A structural regression model was estimated to test our hypothesis that psychological proximity to crime would be associated with increased intensity of worry about crime. The structural model is presented in Figure 5 below. The exact fit of the model was inadequate (χ²  =  496.13, df = 74, p < .001). However, the model demonstrated acceptable fit across most approximate fit indices (SRMR = .06; RMSEA = .09 [90% CI = .08, .10]; CFI = .95; GFI = .91; TLI = .93; NFI = .95). The items measuring worry about crime demonstrated satisfactory factorial validity (AVE = .68) and congeneric reliability (unidimensional ω = .80). Psychological distance from crime was a statistically significant predictor of worry about crime (t = −20.14, p < .001). The standardised path coefficient of −.76 (95% CI = −.82, −.71) also indicated that this relationship was relatively strong and in the expected direction. A more proximal perception of crime resulted in an increased intensity of worry about crime. Overall, the model explained 58% of the variance within worry about crime.

Figure 5. Structural Regression Results. Note. Measurement model not shown. *** Denotes statistically significant result (p<.001).

Discussion

The current study aimed to test a new theoretical model of psychological distance and worry about crime with a community sample. As evidenced by the proposed measurement model, temporal, spatial, social, and hypothetical distance from crime were distinct but highly correlated dimensions of psychological distance from crime. However, social distance from crime was weaker compared to the other dimensions. Our hypothesis that psychological proximity to crime would be associated with an increase in worry about crime was supported. Our findings suggest that community members who perceive crime as likely to occur soon in their current location report more intense feelings of worry about becoming the victim of a crime. These findings can (a) strengthen current theories about the development of fear of crime and (b) inform how we manage fear of crime in the community, as discussed below.

A new theoretical model of CLT and fear of crime

The new model tested in this study explained over half of the variance within worry about crime. Consistent with the assumptions of CLT (Trope & Liberman, 2010), temporal, spatial, social, and hypothetical distance from crime were distinct but interrelated dimensions of psychological distance from crime. Psychological distance from crime and worry about crime shared a statistically significant negative relationship. When an individual perceives crime as psychologically closer to them in time, space, and likelihood they experience more intense feelings of worry about becoming the victim of crime.

Our findings are similar to experimental research by Gouseti (2016) who found that perceiving crime as psychologically distal in temporal, spatial, social, and hypothetical terms resulted participants experiencing less intense levels of worry about crime. Furthermore, Mellberg et al. (2022) found that increased distance from crime in temporal, social, and hypothetical dimensions significantly reduced community members’ worry about crime. In their pilot study, spatial distance from crime also shared a negative relationship with worry about crime, but this relationship was not statistically significant. The current study sought to expand upon Mellberg et al.’s (2022) piloted survey measures. We found that our temporal, spatial, and hypothetical distance items performed well, as evidenced by the factor analyses presented above.

Overall, these findings further demonstrate that CLT can be used to understand how community members estimate their risk of becoming a victim of a crime. The current study offers a theoretical explanation for how individuals become worried about crime in situations absent of crime itself, thereby responding to calls from scholars to enhance theorisation and investigation of the processes underlying fear of crime (Hale, 1996; Jackson, 2006; van der Wurff et al., 1989). Our study also provides an avenue for governments to reduce and manage fear of crime by encouraging psychologically distal perceptions of crime in neighbourhoods where fear of crime is concentrated.

Using psychological distance to reduce and manage fear of crime

Guided by the current evidence, strategies to address worry about crime can be developed to address community members’ perceptions of psychological distance form crime. One approach for such a strategy could be to develop and deliver messaging which encourages psychologically distal perceptions of crime. For example, to address hypothetically proximal perceptions of crime, community members could be informed when crime rates have decreased in an area. This information could also address temporal and spatial proximity to crime by including spatial and temporal reference points (i.e. specific time periods and locations). As evidenced by the results of the current study and previous work (Gouseti, 2016; Mellberg et al., 2022), encouraging psychologically distal perceptions of crime might be associated with less intense experiences of worry about crime among community members.

The findings from the current study also offer another avenue for managing fear of crime in the community using the media. Previous fear of crime research has found that consumption of crime-related media, such as local television news, may heighten perceptions of risk and feelings of worry about victimisation (Callanan, 2012; Callanan & Rosenberger, 2015). Gouseti and Jackson (2015) note that existing perspectives on how media might shape people’s perceptions of crime align with perceptions of psychological distance from crime and crime construal. For example, Chiricos et al. (2000) suggest that crime reports involving a victim perceived as similar to the viewer might heighten their perceptions of risk and worry about victimisation. This aligns with social proximity to crime, which was correlated with other proximal perceptions of crime in the current study and in previous CLT research (Mellberg et al., 2022).

Using current understandings of how the media and fear of crime interact, CLT could be used as a framework to guide how police and news outlets report incidents of crime. As evidenced in the current study, creating psychological distance from crime and avoiding proximal perceptions of crime may lead to less intense experiences of worry about victimisation. Police and news outlets might, therefore, consider reporting less detailed information about crime on public platforms with large audiences, such as social media. Violent crimes might be reported without specific temporal and spatial reference points (e.g. avoiding terms such as ‘a crime occurred in the neighbourhood overnight’) to avoid psychologically ‘proximising’ crime for general audiences. Drawing upon the broader CLT framework and Gouseti (2016), reporting crime using less specific and detailed language (i.e. more abstract language evocative of high-level construal) might also facilitate psychologically distal perceptions of crime and potentially reduce the intensity of worry experienced by community members.

Limitations

The current study has two main limitations. First, psychological distance from crime accounted for a smaller proportion of variance within the items measuring social distance and its items when compared to the other dimensions of psychological distance. This might be due to the way the concept of ‘social distance from crime’ has been measured in our study. We expanded upon the survey research from Mellberg et al. (2022) and Gouseti (2016) by asking participants to indicate whether they believed certain crimes would happen to people perceived by the respondent as similar to themselves. While this is consistent with other research that applies CLT (e.g. Spence et al., 2012), it must be noted that social distance can be measured and manipulated in a variety of ways (Soderberg et al., 2015).

Social distance is frequently manipulated in experiments by asking participants to make judgements about a similar or dissimilar person (Soderberg et al., 2015). However, in experimental research, similarities between actors and observers can be directly manipulated for each participant. For example, Liviatan et al. (2008) manipulated social distance by framing actions as being undertaken by someone who studied the same university classes as the participant. Because the current research is non-experimental and does not manipulate social distance, our survey item, ‘I believe [crime] happens to people like me’, does not contain a clear reference point. As a result, the item could be too broad for participants, leading respondents to feel uncertain about how they determine similarity or likeness (e.g. age, gender, socio-economic status).

One way of addressing this challenge is to change the phrasing of questions about social distance from crime in future research, ‘I believe [crime] might happen to me instead of other people’, ‘I believe [crime] happens to people in my occupation’, or ‘I believe [crime] might happen to someone I know’. These items may help participants to make judgements about crime’s proximity along the social dimension because they contain a clear reference point. These survey items could also be adapted into a scale which measures social distance from crime, rather than using one question as in the current study.

Second, the current study uses cross-sectional data to test models theorising the processes involved in risk perception and worry about crime, and is therefore limited in explaining how these dimensions of distance are affected in terms of exposure to worry-inducing stimuli over time and in different situations. Repeated measures data would allow researchers to capture the ongoing processes involved in risk perception and worry about crime, which can include community members managing their own worry through precautionary behaviours (Gray et al., 2011; Lee et al., 2020), that may be a result of proximal representations of crime. From a CLT perspective, individuals who manage their own worry about crime might engage in behaviours which increase the psychological distance of crime without eroding their quality of life (i.e. functional fear of crime; Jackson & Gray, 2010). Other individuals may instead engage in behaviours which result in a detailed and proximal representation of crime, leading to more frequent or intense worry about crime and a decreased quality of life (i.e. dysfunctional fear of crime; Jackson & Gray, 2010). Understanding the role of psychological distance in the complex processes involved in worrying – or not worrying – about crime requires researchers to move beyond testing process models with cross-sectional data.

Future directions for advancing CLT and fear of crime

There are several further avenues for future research on psychological and fear of crime. For example, the model tested in this study may be useful for understanding in finer detail how perceptions of the immediate environment, such as incivility, shape worry about crime. The incivilities thesis contends that people who perceive signs of physical and social disorder within their immediate environment perceive that they are at a greater risk of falling victim to crime and express higher levels of worry as a result (Covington & Taylor, 1991; Innes, 2004; Skogan, 1986). Fear of crime scholars have hypothesised that these incivilities and signs of disorder act as a visual indication of the prevalence and proximity of crime, even when actual rates of crime are low (Brunton-Smith & Sturgis, 2011; Hunter, 1978). These ideas align with psychological distance. Signs of disorder might communicate that crime is spatially proximal (i.e. that crime occurs in the current location), therefore increasing one’s perceived likelihood of falling victim to crime (hypothetical proximity to crime). Future research could use the theoretical framework proposed in the current study to test these relationships between perceptions of disorder, psychological distance from crime, risk perception, and worry about crime.

Demographic differences, such as age and gender, might also influence the relationship between psychological distance and worry about crime. A substantial body of fear of crime research has found that women (Ferraro, 1996; Jackson, 2009) and the elderly (Ferraro & LaGrange, 1992; Jackson, 2004) report heightened levels of risk perception and worry about crime. Gender might influence the relationships between each dimension of psychological distance and worry about crime. For example, differences in the relationships between worry about crime and social distance across men and women may indicate this dimension of psychological distance is more salient for one group compared to the other, providing audience-specific guidance for fear of crime management strategies informed by CLT.

Similarly, victimisation history could influence perceived psychological distance from crime and worry about crime. Fear of crime research has found some relationship between worry about crime and being the victim of a crime (primary victimisation) or knowing someone who has been the victim of a crime (secondary victimisation). For example, Lee et al. (2020) found that both primary and secondary victimisation were associated with an increase in worry when other demographic variables were controlled. Gouseti (2016) found strong relationships between primary victimisation and increased sensitivity to falling victim to crime. Secondary victimisation also shared significant relationships between most elements of risk sensitivity (control excepted) and worry about crime (Gouseti, 2016).

In the context of CLT, a recent primary or secondary experience of victimisation should be one that is temporally and social proximal to the individual. As found in the current study, proximal perceptions of crime might be associated with increased feelings of worry about crime (Gouseti, 2016; Mellberg et al., 2022). Because the dimensions of psychological distance are highly correlated (Soderberg et al., 2015), a temporally and socially proximal crime event should also be associated with a spatially and hypothetically proximal perception of crime (Gouseti, 2016). Understanding how demographic differences potentially moderate the relationship between psychological distance and worry about crime could guide community safety and wellbeing strategies targeting different audiences.

Conclusion

The current study sought to enhance theoretical understanding about how people become worried about crime. Prior to the current study, there was no known theoretical framework which explained how community members become worried about crime in situations where crime is not actually occurring. Elements of CLT were applied to develop a new theoretical model of fear of crime. Coinciding with emerging research on CLT and fear of crime, we found that people who perceive crime as psychologically proximal (i.e. likely to happen to them soon in their current location) worried more intensely about becoming a victim of crime. Our findings provided clear directions for reducing fear of crime by addressing proximal perceptions of victimisation. Future researchers are now equipped with the tools to investigate how psychological distance shapes worry about crime for different groups of people and across different situations. In doing so, we can continue to advance theoretical investigations into how fear of crime develops in the community and how it can best be managed.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Due to the nature of this research, participants of this study did not agree for their data to be shared publicly, so supporting data is not available.

Notes

1 In line with the National Statement on Ethical Conduct in Human Research (National Health and Medical Research Council [NHMRC], 2018), participants were informed that the submission of the completed survey was accepted as an indication of their consent to participate in the research project.

2 We also investigated whether the measurement model was preferred when compared to other theoretically plausible models. The results of nested model comparison provided strong evidence in favour of our hypothesised measurement model. See supplementary material for model comparisons.