In the last four decades, China has experienced radical economic transformation. New infrastructure and commercial buildings mushroomed in urban China on a massive scale and at an astonishingly rapid pace, enticing many agricultural workers in the rural areas to rush to these fast-growing Chinese cities in search of better employment opportunities and higher incomes. Due to the demanding hours and other disadvantages at their urban destinations, many migrant workers have no choice but to leave their children behind at their rural places of origin, often in the care of the children’s grandparents or relatives. These children have come to be known as the “left-behind children (LBC).” LBC are a highly vulnerable social group in China, as they experience multiple economic and social disadvantages such as poverty, malnutrition, and improper (and sometimes a total lack of) adult care (Chen et al., 2021a; Chen et al., 2021b; Zhao, Liu, & Wang, 2015).

There is well-documented evidence that parental rural-to-urban migration significantly reduces children and adolescents’ psychological well-being, resulting in problems such as increased risks of depression, anxiety, suicidal ideation, substance use, conduct disorders, and various kinds of victimization (Fellmeth et al., 2018; Liang et al., 2016; Wu et al., 2015). Moreover, several studies separated the types of parental migration. They found that compared with paternal migration, maternal migration has a more severe adverse effect: children with maternal migration experience higher levels of caretaker mistreatment and neglect (Liu et al., 2009) as well as sexual victimization (Chen et al., 2021b). In contrast to the abundance of studies on LBC’s general mental well-being, LBC’s fear of crime has largely been overlooked by researchers and policymakers. Investigation into this issue is important for several reasons. First of all, even though the prevalence of crime victimization among Chinese children is lower than that among Western countries (Chen et al., 2017; Chen et al., 2021a; Chen et al., 2021b; Finkelhor et al., 2013), there is preliminary evidence that LBC is at a much higher risk of crime victimization than the average Chinese children (Chen et al., 2017; Chen et al., 2021a; Chen et al., 2021b). Severe physical and sexual abuse, as well as homicide victimization of LBC, have been periodically reported by Chinese media (Chen et al., 2017; Liu, 2017), amplifying public concerns for LBC’s safety. The compounded effects of social and economic disadvantages and crime victimization may significantly impact LBC’s fear of crime. This link has not been sufficiently explored. Second, as a construct, fear of crime also taps into the psychological anxieties and perceived insecurity associated with the economic and social disadvantages of being LBC, as well as secondary social disadvantages (such as bullying and discrimination due to one’s LBC status) that children experience in school and the community. Studying LBC’s fear of crime can help us develop further insight into the complex mechanisms that influence the psychological well-being of LBC.

Analyzing survey data from a probability sample of middle school students in a rural Chinese province, this study aims to empirically investigate the impacts of parental migration on fear of crime in contemporary rural China and provides evidence for designing effective intervention programs to reduce crime fear. This study contributes to the existing literature in a couple of ways. First, the existing literature shows that the different patterns of parental migration (i.e., paternal migration, maternal migration, and dual-parent migration) have heterogenous impacts on children’s risk of victimization (Chen et al., 2017; Chen et al., 2021a; Chen et al., 2021b). It is plausible that they may also affect fear of crime differently. This study first disaggregates the different patterns of parental migration. Then it explores the relationship between parental migration and fear of crime, with the intention of identifying the most vulnerable subgroup of LBC and its associated characteristics. Second, despite numerous studies (e.g., Gibson et al., 2002; Liu, 2013; Scarborough et al., 2010) indicating that prior crime victimization, vulnerability to victimization, and neighborhood characteristics were significant predictors of fear of crime, few empirical studies tested the generalizability of these theoretical propositions in the context of rural China. This study attempts to fill these gaps in the literature.

Literature Review

Parental Migration and Children’s Exposure to Victimization

Since the initiation of the economic reform and opening-up policy, especially since the 1990s, China has witnessed rapid economic development, leading to a large number of people migrating from rural to urban areas or from small cities to “first-tier” megacities such as Beijing, Shanghai, and Guangzhou, in search of employment opportunities and higher incomes. The number of internal migrants was nearly 376 million in 2020 (Duan et al., 2022). However, the existing household registration system and social security system still exhibit serious regional and urban-rural divisions to this day, making it difficult for internal migrants to obtain permanent resident status in the destination cities. Due to these institutional barriers, internal migrants in the cities often suffer poor economic conditions, high labor intensity, and unstable employment, forcing them to leave their children behind in their rural hometowns (Zheng & Wu, 2014). This results in a large number of LBC in rural areas. Although statistical reports from different government agencies at different times have adopted varying definitions of rural LBC depending on the number of migrant parents, the duration of parental migration, and the age of the children, these statistics consistently documented a considerable number of rural LBC in China (Lyu et al., 2024). The most conservative estimate released by the Ministry of Civil Affairs in 2021 revealed 6.436 million rural LBC who could not be fully supervised by their parents at the end of 2020 (Ministry of Civil Affairs, 2021), and a census-based estimate using inferential statistics even put the country’s LBC in 2020 at a staggering 19.06 million (Lyu et al., 2024).

For LBC, parental migration can increase the family income, improve the family’s financial situation, and increase family investments in children’s education, medical care, housing, and nutrition, thus potentially having a positive impact on their physical and mental development. However, in reality, the long-term migration of either or both parents results in the temporary separation of family members, leaving LBC unable to receive adequate parental care, monitoring, guardianship, and protection. As a result, they face a series of physical and mental development problems, such as anxiety, depression, emotional instability, and victimization (Chen et al., 2017; Zhou et al., 2020).

A national survey reveals that 35% of all LBC live with one parent, 59% live with grandparents, 4% live with other relatives, and the remaining 2% stay at boarding schools (Kong & Meng, 2010). In dual-parent migration families, children often stay with grandparents. Due to their low educational attainment and weak health, they often needed help to provide sufficient and effective guardianship and protection. Sometimes, LBC must even care for their grandparents in addition to studying (Duan et al., 2017). In single-parent migration families, the stay-behind father or mother not only bears the heavy responsibility of taking care of the children and the elderly but also the demanding agriculture work and household chores. They invested limited time and energy in caring for children, decreasing monitoring, guardianship, and protection. Due to insufficient and ineffective monitoring and supervision, some LBC may commit delinquent behaviors, associate with delinquent peers, or engage in unstructured and unsupervised social activities, increasing their proximity to potential offenders and risk of victimization (Chen et al., 2021a).

The existing empirical studies explored the association between parental migration and victimization and found that compared with children who live with both parents, LBC experienced more general victimization (Chen et al., 2017; Chen et al., 2019; Zhou et al., 2020), peer victimization (Wang & Liu, 2020; Xiong et al., 2019; Zhang et al., 2019), neglect and abuse (Chen et al., 2019; Wen et al., 2020; Yang et al., 2020), and sexual assault (Chen et al., 2021b; Yan et al., 2018). Comparatively speaking, mothers play a more important role in educating and accompanying their children, especially young children, than fathers (Hao, 2022). Research has shown that maternal migration has a greater negative impact on children’s development (Chen et al., 2019) because a stay-behind father often works nearby to earn money to subsidize household expenses, in addition to taking care of children and doing agriculture work. Additionally, due to the differences in traditional gender roles between men and women, fathers often adopt authoritarian parenting styles, even yelling at or scolding their children and neglecting communication with their children, which can easily lead to a rigid parent-child relationship. Therefore, several studies have taken a step further to explore the impact of different types of parental migration on victimization and found that compared with children living with both parents, children with maternal migration and children with both-parent migration experienced higher levels of caretaker mistreatment and neglect (Liu et al., 2009) as well as sexual victimization (Chen et al., 2021b).

Although there is currently no research directly examining the relationship between parental migration and children’s fear of crime in rural China, existing studies have revealed that parental migration may increase children’s actual and perceived vulnerability and risk of victimization, which have been identified, among others, as salient predictors of fear of crime. In addition, parental migration may represent a unique form of concentrated disadvantage, potentially influencing fear of crime above and beyond these conventional predictors. Therefore, further investigation is warranted to tease out the direct and indirect effects of parental migration on children’s fear of crime. Moreover, different patterns of parental migration may exert heterogeneous effects on children’s fear of crime, considering previous research on these effects on crime victimization (Chen et al., 2017; Chen et al., 2021a; Chen et al., 2021b).

Fear of Crime and Theoretical Perspectives Explaining Fear of Crime

Research on fear of crime first emerged in the United States in the 1960s (Doran & Burgess, 2012; Farrall et al., 2000), and ever since then, it has become an important research topic in criminology. Recently, research on the fear of crime in China has been proliferating as public security has become a genuine concern with rapid modernization (Jing et al., 2021). Ferraro and LaGrange (1987, P. 73) defined fear of crime as “negative emotional reactions generated by crime or symbols associated with crime.” This definition, which emphasizes the emotional component of fear, has been widely adopted by many scholars.

The existing literature suggests that fear of crime is correlated with a series of potentially severe negative health consequences. For instance, it has been found to increase depression and anxiety, impair interpersonal distrust, and weaken physical and mental well-being (Scarborough et al., 2010). These adverse effects can be further exacerbated by socioeconomic disadvantages due to a higher risk of victimization and a lack of sufficient social support to alleviate the deleterious consequences. Given the adverse health outcomes of fear of crime, many scholars have attempted to explain the factors contributing to fear of crime with the hope of disrupting these fear-inducing mechanisms. Several different yet related theoretical and empirical models have been developed, the most prominent of which are the victimization, vulnerability, and neighborhood models.

The Victimization Model

Past crime victimization may cause lingering psychological trauma in the victims and, therefore, elevate their crime fear. Additionally, past victimization experiences may reflect personal vulnerability to future victimization, reminding individuals of their vulnerability, which may reinforce their fear of crime (Gibson et al., 2002). It thus stands to reason that those individuals with past victimization experience may report a higher fear of crime. However, the evidence on the victimization-fear link is mixed. For instance, based on data from Wave 7 of the World Value Survey, Lin (2023) found that prior criminal victimization experience significantly predicted higher levels of crime fear among residents from 36 countries, even after controlling for country-level variations. Podaná and Krulichova (2023) analyzed data collected from the Urban Youth Victimization Survey in the Czech Republic and also found that victimization experience significantly affected respondents’ fear of crime, although the effect sizes vary across crime types: cyber-victimization and bullying had the strongest fear-inducing effects compared to conventional violent and property crimes. Jing et al.’ (2021) urban China-based study demonstrated that victims of violent crime exhibited a higher level of fear of violence than other residents, but burglary victimization had no significant effect on fear of burglary. Utilizing a multi-stage cluster sample in Tianjin, China, Liu (2013) found that violent and property victimization had no significant effect on fear of crime after controlling for individual and neighborhood characteristics. Similarly, Roccato et al. (2011) analyzed a sample of the Italian adult population and found that none of the direct or indirect victimizations exerted a significant effect on fear of crime.

The Vulnerability Model

In addition to personal experience with crime, factors that make people vulnerable to crime also play a role in shaping their fear of crime (Goodey, 1994; Stanko & Hobdell, 1993; Walklate, 1994). While it seems logical that the physical risk of victimization would be the main predictor of fear of crime, research has shown that this is not always the case. For example, young men, who are at the highest risk of victimization, tend to have the least fear of crime (Gialopsos, 2011; Melde et al., 2016; Melde & Esbensen, 2009).

Recent research on delinquent youths has revealed more complex relationships between actual risk, perceived risk, and fear of crime. While proximity to motivated offenders (such as delinquent peers) increases both actual and perceived risk of victimization, it does not necessarily lead to higher levels of fear (Gialopsos, 2011). In fact, some studies have found that delinquent youths and gang members may experience lower levels of fear of crime than non-delinquent youth, even though they are at greater risk of victimization (Melde et al., 2016; Melde & Esbensen, 2009). This may be because gang members are expected to appear tough and fearless and because they may feel a sense of belonging and solidarity from their gang affiliation.

Related to, but distinct from, the risk of victimization is an individual’s vulnerability to victimization (Killias, 1990). Vulnerability refers to the inability to defend oneself from crime and the severity of the potential consequences (e.g., death or serious injury). Research has shown that women, people with low socioeconomic status, people with disabilities, and the elderly are more vulnerable to crime and, therefore, experience higher levels of fear of crime (Clemente & Kleiman, 1977; Hale, 1996; Killias & Clerici, 2000; Pantazis, 2000; Skogan & Maxfield, 1981; Warr, 1994).

Self-perceived vulnerability is not simply based on actual physical vulnerability but also on social-psychological factors (Jackson, 2009). For example, women may be more worried about crime than men not only because they feel less able to defend themselves physically but also because they have lower perceived self-efficacy (i.e., the belief in one’s ability to cope with challenges and achieve one’s goals) and higher perceived negative impact from victimization. Similarly, white people in the United States who perceive themselves to be in the racial minority in their neighborhood may experience higher levels of fear of crime than black people, even though black people are actually more likely to be victims of crime (Chiricos et al., 1997).

The Neighborhood Model

This model focuses on the neighborhood environment and argues that neighborhood disorder influences an individual’s fear of crime. In the context of fear of crime research, disorder refers to “low-level breaches of community standards that signal an erosion of conventionally accepted norms and values” (LaGrange et al., 1992, p. 312), often reflected by minor offenses against the law or community standards (e.g., vandalism, drinking in public, and drug crimes). Citizens residing in more disorderly neighborhoods tend to express a higher fear of crime. Most existing studies demonstrated that neighborhood disorder had a positive and significant effect on fear of crime (Gibson et al., 2002; Roccato et al., 2011; Scarborough et al., 2010). For example, McGarrell et al. (1997) analyzed the data from a random sample of residents in a US city and found that perception of neighborhood disorder significantly increased fear of crime. Using data from 2,599 citizens nested within 21 cities in the USA, Franklin et al. (2008) found that perception of neighborhood disorder (for example, noise and traffic problems) was the strongest predictor of fear of crime both within and across cities. Based on a sample of Kansas City, Missouri residents, Scarborough et al. (2010) found that perception of social disorder and physical disorder emerged as two significant predictors of citizens’ fear of crime after controlling for race, age, gender, and education. A study by Liu (2013) also showed that individuals living in neighborhoods characterized by a high level of disorder were more fearful than others after controlling for individual characteristics and victimization experiences.

Related to neighborhood disorder, collective efficacy refers to the “social cohesion among neighbors combined with their willingness to intervene on behalf of the common good” (Sampson et al., 1997). Neighborhoods with high levels of collective efficacy perceived by their residents were found to exhibit lower crime rates. Several empirical studies investigated the association between collective efficacy and fear of crime (e.g., Gibson et al., 2002; Yuan & McNeeley, 2015). Based on data from three cities in the USA, Gibson et al. (2002) found that the perceptions of collective efficacy had a significant reductive effect on fear of crime. Similarly, using the Seattle Neighborhoods and Crime Survey, Yuan and McNeeley’s (2015) study showed that perceived collective efficacy was negatively associated with fear of burglary. In summary, fear of crime is influenced by a complex set of factors, including actual experience of victimization, neighborhood characteristics, and vulnerability to victimization.

The Current Study

As was illustrated in the literature review, parental migration may constitute a unique type of vulnerability and cast an influence on rural Chinese children’s mental well-being, including their fear of crime, above and beyond traditional measures of vulnerability, as well as other crime fear predictors. Furthermore, the different patterns of parental migration may, in themselves, represent varying degrees of vulnerability and exert heterogeneous effects on fear of crime. Thus, the purpose of this study is to extend prior research on LBC’s well-being by exploring the effects of different patterns of parental migration on rural Chinese children’s fear of crime while controlling for traditional predictors of fear of crime. Drawing on the survey data from a sample of 1,985 rural Chinese adolescents, this study aims to answer two research questions: First, does parental migration affect rural LBC’s fear of crime above and beyond traditional predictors of crime fear (i.e., prior victimization experience, vulnerability to victimization, and neighborhood characteristics)? Second, do different parental migration patterns (i.e., mother-only, father-only, versus dual-parent migration) cast different influences on LBC’s fear of crime? These questions lead to the formulation of the following hypotheses:

  • H1: Parental migration predicts higher levels of fear of crime among rural LBC, above and beyond influence from conventional predictors such as victimization experience, vulnerabilities to victimization, and neighborhood characteristics.

  • H2: Different patterns of parental migration (mother-only, father-only, versus dual-parent migration) exert different influences on rural LBC’s fear of crime.

Methods

Sample

The data for this study are from the Mental and Behavioral Health of Adolescent Survey, a research project conducted in Guizhou Province, China. Carried out from November to December 2019, this project aimed to investigate the prevalence and correlates of Chinese adolescents’ mental and behavioral problems (e.g., depression, anxiety, conduct disorder, and substance abuse). Guizhou Province was selected as a sample site for two reasons. First, located in southwest China, Guizhou is a landlocked province representing one of the least developed regions in China. Since China’s market economic reform in the 1980s, Guizhou has been a major emigrant province with a large number of LBC. Therefore, it is an ideal sample site for our research on the LBC population. Second, as one of the few multiethnic provinces in China, Guizhou boasts 17 ethnic minorities (National Bureau of Statistics of China, 2018), which enables the researchers to draw a large probability sample of children from ethnic minority groups, who have often been overlooked in past studies.

The research team purposely selected two districts of Guizhou, Guiyang and Qiannan Buyi, and Miao Autonomous Prefecture (Qiannan) to draw the research sample. As the capital city of Guizhou, Guiyang has a higher average income than Qiannan and is inhabited mostly by the majority Han ethnicity. Qiannan is a less developed region than Guiyang, with 55% of the population being from ethnic minority groups (e.g., Buyi, Miao) (Bureau of Statistics of Qiannan, 2010). Using a multistage random cluster sampling method, the research team obtained a list of all middle schools in Guiyang and Qiannan from the local Education Administrative Department. At the time of the survey, there were 338 middle schools in Guiyang (including 183 urban middle schools and 155 rural middle schools) and 108 middle schools in Qiannan (including 47 urban middle schools and 61 rural middle schools). Considering the urban-rural division of schools, five middle schools from rural counties and five from urban areas were randomly selected. A total of 10 schools participated in the study. Second, two classes were randomly selected in each grade from the previously selected schools, and all students from the selected classes were invited to participate in this survey study. The number of students in these classes ranged from 36 to 52.

After obtaining consent from the teachers and the students, we distributed questionnaires to participating students on regular school days. During the survey period, none of the school personnel were present in the classroom. Researchers explained to the students the purpose of this survey, assured them that participation was voluntary, and informed them that all personal information would be kept confidential. They were also told that they could withdraw from the survey at any time.

In total, 2533 students were selected to participate in this survey. Thirteen students refused to participate, and 22 provided invalid data, resulting in a final sample size of 2498. The return rate was 98.6%, which was consistent with the high return rate of school-based research on juvenile delinquency and victimization (Webb et al., 2011).

Measures

Dependent Variable

Fear of Crime

The dependent variable, fear of crime, was measured by a 5-item scale. The respondents were asked how worried they were about these events: (1) worried about home invasion and burglary, (2) worried about being robbed, (3) worried about being beaten up by strangers, (4) worried about being insulted on streets or in other public places, and (5) worried about having valuables stolen on the streets or in other public places (Ren et al., 2015). We summed the responses on a Likert-type scale ranging from “not worried at all” (=1) to “very worried” (=5). The reliability of the measure was excellent (Cronbach’s alpha = 0.95).

Independent Variable

Parental Migration Status

As the key independent variable in this study, parental migration status was measured by asking respondents whether one of their parents currently worked in other cities or provinces.Footnote 1 Then, we recorded this variable into four dummy variables: paternal migration only, maternal migration only, dual-parental migration, and no migration.

Vulnerability to Victimization

Perceived Economic Pressures. We asked the respondents during the past year if they had experienced these events: (1) Family had inadequate money to purchase rice, (2) family had inadequate basic household commodities, (3) family had inadequate money to pay for a medical bill, (4) family had inadequate money to purchase clothes or shoes, and (5) family had inadequate money to purchase stationery and other materials (Shek, 2005). The Response categories for each of the five items were 1 = never, 2 = rarely, 3 = sometimes, 4 = often, and 5 = always. We summed up the responses, with higher scores indicating higher levels of economic pressure (Cronbach’s alpha = 0.85).

Caregiver Monitoring. To measure caregiver monitoring, respondents were asked about the level of monitoring that caregivers exercise over them, including (1) they know where I was after school, (2) they know when I went out at night, (3) I told them whom I was with when I was out, (4) they asked where I had been, and (5) they would call me when I was late (1 = never, 2 = rarely, 3 = occasionally, 4 = often, 5 = always) (Chen et al., 2017). We summed up the responses, and the higher scores indicated a higher level of caregiver monitoring (Cronbach’s alpha = 0.80).

Perceived Neighborhood Characteristics

Perceived Disorder. Similar to Gibson et al.’ (2002) measure, we used five survey items to measure neighborhood disorder. The respondent was asked the extent to which they agree (1= “strongly disagree” and 5 = “strongly agree”) to the following statements: (1) The neighbors are selfish and only think about themselves; (2) garbage can be seen everywhere; (3) burglaries frequently happen; (4) cars, motorcycles, battery cars, and bicycles are parked at random; and (5) often seen neighbors quarrel. The responses were summed across items, with higher scores indicating perceptions of greater disorder. The disorder scale has a Cronbach alpha of 0.77.

Collective Efficacy. Collective efficacy was measured by four survey items taken from Sampson et al. (1997, p. 921), which has been used by a considerable number of studies (e.g., Gibson et al., 2002; Zhao, Lawton, & Longmire, 2015). The scale consists of the following items: (1) Residents will notify the police about illegal activity occurring in this neighborhood; (2) if there is a suspicious person hanging around your block, someone is likely to call the police; (3) when you do a favor for a neighbor, you can generally trust the neighbor to return the favor; and (4) if you were in need of help with your car stuck in front of your residence, your neighbors would come to your assistance. Response categories to these items ranged from “very likely” (= 1) to “very unlikely” (=5). Factor analysis showed that the four-item scale loaded onto the same dimension with acceptable inter-item reliability (Cronbach’s alpha = 0.79). We summed up the responses, with higher scores indicating lower levels of perception of collective efficiency.

Prior Victimization Experience

We asked the respondents during the past year whether anyone (1) robbed you, (2) stole something from you, (3) broke or ruined any of your belongings on purpose, (4) assaulted you with a weapon (e.g., a knife, stick, dagger), (5) assaulted you without a weapon, and (6) attempted to assault you but do not succeed (Finkelhor et al., 2005). Response categories to these items ranged from “never” (= 0) to “8 times and more” (=4). We summed up responses to items (1), (4), (5), and (6) to measure violent victimization experience, with higher scores representing having more violent victimization experience during the past year (Cronbach’s alpha = 0.71). Similarly, we summed up the responses to items (2) and (3) to measure property victimization experience (Cronbach’s alpha = 0.77).

Demographic Variables

Demographic variables included gender (0 = female, 1 = male) and age (measured in years).

Results

Descriptive Statistics

Table 1 reports the descriptive statistics (mean, standard deviation, range) of all variables used in the analyses of this study. This study revealed that most rural households (65%) did not go out to work and still followed the traditional lifestyle of farmers. A total of 16% of the students reported that both of their parents left their homes to work outside in exchange for better material living conditions for the family. Another 13% of families have fathers going out alone, while the proportion of mothers leaving alone is relatively small (6%). Among them, 49% of respondents were male; the average age was 13.55 years. The mean value of fear of crime is above the median value of the scale (mean = 15.36), suggesting a considerable fear of crime among the rural Chinese children surveyed.

Table 1 Descriptive statistics of variables (N = 1985)

Bivariate and Multivariate Analysis

Before proceeding to multivariate analysis, we computed a Pearson correlation matrix to evaluate the relationship among all variables included in the regression. As expected, the association between gender (p < 0.001), age (p < 0.01), dual-parental migration (p < 0.01), caregiver monitoring (p < 0.01), perceived economic pressures (p < 0.001), property victimization (p < 0.001), violent victimization (p < 0.001), perceived disorder (p < 0.001), collective efficacy (p < 0.001), and fear of crime were significant. Detailed results are presented in Table 2.

Table 2 Pearson correlation matrix for all variables (N = 1985)

The results of the nested OLS regression analysis appear in Table 3. Before the regression analysis, we performed collinearity diagnosis on each variable, and the results showed that the maximum VIF was less than 1, indicating that there was no multicollinearity among the variables. Model 1 examines the relationship between migration status and fear of crime, controlling for gender and age. All patterns of parental migration have a significant impact on fear of crime. Compared with no parent migrating, those with paternal migration (β = 1.10, p < 0.05), maternal migration (β = 1.68, p < 0.05), and dual-parent migration (β = 1.64, p < 0.001) exhibited a higher level of fear of crime. In the first model, being male (β = −2.11, p < 0.001) and older (β = −0.50, p < 0.01) were negative and statistically significant, suggesting that females and younger people exhibited a higher level of fear of crime. The pattern generally held across all four models. This model explained 4% of the variation in fear of crime.

Table 3 Nested regressions on fear of crime (N = 1985)

To test the vulnerability model, caregiver monitoring and perceived economic pressures were added to the model (see Model 2). Results revealed that perceived economic pressures (β = 0.34, p < 0.001) were positive and statistically significant. Specifically, those suffering greater family economic pressure exhibited a higher level of fear of crime. The relationship between dual parental migration and fear of crime in this second model continued to be significant (β = 1.35, p < 0.01), but the effect slightly diminished. The effects of paternal migration only and maternal migration only become insignificant. This model explained a 6% variation in fear of crime.

Model 3 includes prior victimization experience to assess whether the victimization model can be generalized to adolescents in rural China. Results revealed that property victimization (β = 0.44, p < 0.001) and violent victimization (β = 0.31, p < 0.001) had significant effects on fear of crime. More specifically, those experiencing more property and violent victimization exhibited a higher level of fear of crime. All the previously significant predictors were still significant, but the effects slightly diminished. The adjusted R2 for this model increased, suggesting that 9% of the variance in the dependent variable was explained by this model.

The final model includes two measures of perceived neighborhood characteristics. It was found that perceived disorder (β = 0.19, p < 0.001) and collective efficacy (β = 0.12, p < 0.01) had significant effects on fear of crime. More specifically, perceiving greater neighborhood disorder and weaker collective efficacy elevated crime fear. All the previously significant predictors were still significant, but the effects slightly diminished. The adjusted R2 for this model further increased, suggesting that 10% of the variance in the dependent variable was explained by this model.

Discussion and Conclusion

This study represents a preliminary attempt to investigate the relationship between parental migration and fear of crime based on a probability sample of rural Chinese adolescents from a province experiencing significant population outflow within China. First and foremost, this study affirmed the impact of dual-parent  migration on fear of crime above and beyond conventional predictors, supporting Hypothesis 1. Although single-parent (i.e., paternal or maternal) migration, on their own, had a significant impact on fear of crime, such an impact was mostly mediated by other measures of vulnerability, such as perceived economic pressures. This finding stands in contrast with past research, which found that even single-parent migration significantly impacted LBC’s mental well-being (e.g., Chang et al., 2019). This does not seem to be the case when it comes to fear of crime. This finding also supports Hypothesis 2, signaling the different effects of single-versus-dual-parent migration.

The multivariate analyses demonstrated that several traditional predictors significantly elevated fear of crime among rural Chinese children. Specifically, adolescents with prior victimization experience were more fearful of crime, consistent with findings from some previous research (Gibson et al., 2002; Lin, 2023), yet contradicting findings from urban China (Jing et al., 2021; Liu, 2013). Perceived family economic pressure and poor caretaker monitoring, two measures of vulnerability were also found to significantly predict fear of crime, in congruence with the existing literature (e.g., Pantazis, 2000). Children with the perception of greater neighborhood disorder and lower levels of collective efficacy exhibited a higher level of fear of crime in this study, a finding that also echoes past research documenting the various impacts of neighborhood characteristics on children’s well-being (Hurd et al., 2012; Jones-Rounds et al., 2013; Wu et al., 2015). Despite differences in social and cultural contexts between China and the West, where much of the fear of crime scholarship has been conducted, this study found that predictors of fear of crime identified in the Western literature are largely generalizable to predict fear of crime among rural adolescents. However, it is noteworthy that the mechanisms by which these risk factors induce fear may be entirely different, as much of the Western literature was based on urban samples and findings from urban China (Jing et al., 2021; Liu, 2013) do not always agree with these Western “lessons.” More research is needed to unpack these lingering divergences.

Meanwhile, the current study found that dual-parent migration had a unique and significant elevating effect on fear of crime, even after controlling for other known predictors of fear of crime: compared to those living with both parents, adolescents with both parents migrating expressed higher levels of fear. The negative impacts of dual-parent migration on the well-being of rural Chinese adolescents seem independent of those exerted by economic stress or poor adult care due to the absence of parents. This finding partially resonates with findings from previous research that parental migration has a series of deleterious mental health consequences (Chen & Zhou, 2021) yet simultaneously accentuates the compounded vulnerability of LBC who experience dual-parent migration. Studies on fear of crime found that perceived vulnerability, rather than actual vulnerability, also influences fear of crime (Chiricos et al., 1997; Jackson, 2009). In the context of the current study, it could be that adolescents with both parents migrating developed a sense of emotional insecurity in the absence of their parents, which increased their crime fear above and beyond that resulting from perceived economic stress or the lack of quality adult supervision. Further research is warranted to explore a hypothesis like this.

Overall, this study adds to the existing literature on the social-psychological impacts of parental migration among rural Chinese adolescents. However, several limitations should be considered when interpreting these findings. First, the sample of this study was drawn from one province in China only. While Guizhou has a large rural population and shares many common characteristics of other rural provinces in China, caution should be applied when generalizing the findings to other parts of China. Second, the multistage cluster sampling method applied in the current study results in the clustering of participants (as students are clustered within schools), a limitation that should be addressed in future studies adopting more sophisticated designs. Third, this study relied on self-reported survey data and only focused on the predictors of fear of crime at the individual level, macro-level factors such as objectively measured neighborhood characteristics (e.g., the percentage of parental migration in the neighborhood, poverty rate, and crime rate) were not included in the analysis. Future research should explore the correlates of fear of crime at both the individual and macro levels and further examine their potential interaction effects. Related to this, the current study did not control for the innate personality traits of the individuals, which may influence how they subjectively perceive their neighborhoods, a limitation that should be addressed in future research. Finally, this study only identified the direct effect of dual-parent migration on fear of crime. Future research should explore the origin as well as any mediating or moderating mechanisms of the fear-inducing effects of dual-parent migration with appropriate research designs and analytic approaches.

Despite these limitations, findings from this study carry several policy implications. First, the findings underscore the social vulnerability of LBC in China and the urgency for the national and local governments to strengthen support for LBC, especially for adolescents who experience dual-parent migration. In particular, mental health counseling for the LBC, as well as financial and family support for the non-parent guardians, should be prioritized. This would mean that government departments should provide social assistance and labor skills training and recommend employment opportunities for those who are physically able to increase family income and alleviate family economic pressure. Free or low-cost health care and education access should also be made easily available to these adolescents. Second, governments should strengthen social cohesion and reduce the crime rate in rural communities by improving the quality of rural environments, such as removing graffiti and garbage, promoting collective efficacy, and consequently reducing adolescents’ fear of crime.