Perceived Criminal Threat from Undocumented Immigrants: Antecedents and Consequences for Policy Preferences

Abstract

Drawing on the group threat perspective, this paper examines the perception of criminal threat from undocumented immigrants and its relation to both contextual measures of threat and public support for enhanced controls against undocumented immigrants. With data from a national telephone survey of non-Latino adults (N = 1,364), we estimate the predictors of perceived criminal threat as well as the effects of perceived threat and other factors on immigration policy preferences. Results indicate that political ideology and education are the strongest predictors of perceived criminal threat. Perceived criminal threat has the greatest influence on support for more punitive controls and partially mediates the effects of race, education, political ideology, and contextual threat on these control preferences. Future social threat research should consider the inclusion of perceptual threat measures instead of relying solely on contextual indicators of threat. In addition, contextual threat should be explored more often in dynamic, rather than static, terms.

Keywords:

• social threat
• immigration

Notes

¹ While scholarly reference frequently makes use of the term “undocumented,” the survey providing data for this project used the term “illegal” for a few reasons. First, claims made to the general public most often use the latter term. Additionally, since our focus is on measuring public perception, we felt the term “illegal” would create less confusion among respondents than either “undocumented” or “unauthorized.” Moreover, it has been shown that the term used does not significantly influence policy preferences of survey respondents (Merolla, Ramakrishan, & Haynes, 2013).

² In this paper, we use “Anglos” to reference non-Latino whites.

³ Individual perceptions may mediate the relationship between context and support for punitive controls or may be completely independent for reasons we discuss in our conclusion.

⁴ The Research Network, Inc., a public opinion polling firm in Tallahassee, Florida, completed the surveys using list-assisted random digit dialing. Using recommended criteria from the American Association for Public Opinion Research (AAPOR, 2009), a response rate of 38.5% and cooperation rate of 73% were obtained. These are generally consistent with rates reported in other recent studies that are using RDD telephone surveys (e.g. Hirschfield & Piquero, 2010; King & Wheelock, 2007). Also, while response rates have decreased in recent years (Pew Research Center, 2004), several researchers have found that response rates are not significantly related to non-response bias (Curtin, Presser, & Singer, 2005; Keeter, Kennedy, Dimock, Best, & Craighill, 2006; Keeter, Miller, Kohut, Groves, & Presser, 2000).

⁵ We used the 5-year pooled estimates instead of single-year estimates for 2008 because of limitations to the single-year data. Single-year ACS estimates are only available for large counties (at least 65,000 population), resulting in a considerable amount of missing data. Further, the larger sample size afforded by multi-year data provides for more reliable estimates, even in the largest counties.

⁶ 2000 Census data were used for this measure of percent Latino.

⁷ We also examined the possibility that the effects of percent Latino and change in percent Latino could be non-linear. This prospect was first raised by Blalock (1967) in relation to minority composition posing either economic or political threats. Quadratic terms for percent Latino and change in percent Latino were not significant in any of the models estimating either perceived criminal threat or support for immigration controls (results available on request).

⁸ There were 18 missing responses on age, 13 on education, 203 on income, and 57 on conservative. For purposes of imputation, OLS regression was used for age; ordered logistic regression was used for education and income; and logistic regression was used for conservative. We also included male, white, black, other race, and perceived criminal threat in the models even though they were not missing data.

⁹ Each of the subsequent analyses was also conducted using the total crime rate. It was not significant in any model, and results for other predictors were substantively unchanged.

¹⁰ Our comparisons to Wang (2012) make reference to her full sample results (Table 2). Her analysis found unemployment to be a significant predictor of perceived criminal threat for Natives as opposed to Immigrant respondents.

¹¹ The significance of slope differences was estimated using the method established by Paternoster, Brame, Mazerolle, and Piquero (1998). These differences used one-tailed tests for significance.

¹² Although border state residence was marginally significant in the full sample analyses for border control, it fails to reach statistical significance for either the Anglo or Black subsamples. This is likely because of the reduced sample sizes for the subgroups.

¹³ It would have been helpful if we could have specified our models predicting perceived criminal threat in ways that corresponded more closely with Wang’s estimates, but our survey unfortunately was conducted too early for that.

Additional information

Notes on contributors

Elizabeth K. Stupi

Elizabeth K. Stupi is an assistant professor of Criminology at the University of West Georgia. Her current research interests include the influence of mentors on both crime rates as well as individual criminality and the effect of social threat, particularly immigrant threat, on the levels of social control.

Ted Chiricos

Ted Chiricos is William Julius Wilson professor of Criminology and Criminal Justice at Florida State University. His current research interests focus on the effects of race, immigration, age, and social threat on justice outcomes as well as the various factors contributing to the extraordinary punitiveness of American culture. He is also collaborating on the examination of threat-related factors that may contribute to intimate partner violence.

Marc Gertz

Marc Gertz is a professor of Criminology at Florida State University. His research interests include the study of public opinion and the criminal justice system, the organization of courts as well as interest groups and voting behavior.