An assumption sometimes made regarding immigrants to the U.S. is that as a group they are more likely to engage in criminal behavior than are U.S. citizens. An analysis performed using data collected by the Justice Center reveals that in Anchorage there is no empirical basis for this assumption.
Immigrants—both those without citizenship and those who have assumed citizenship—form a low percentage of the Anchorage arrestee population—substantially less than their representation in the population as a whole. Moreover, non-citizen arrestees are not significantly more likely to have been jailed for a felonious crime, a violent offense, or a drug offense, and are less likely than those with citizenship to have prior criminal histories.
The data used in this analysis were collected as part of the Anchorage Arrestee Drug Abuse Monitoring (ADAM) program. ADAM is a national, multi-site drug monitoring program that measures the extent and nature of alcohol and drug use among those who have been recently arrested. It is funded by the National Institute of Justice (NIJ). ADAM researchers conduct in-depth interviews with recent arrestees and collect urine samples.
In addition to questions about drug and alcohol use, respondents to the survey are asked about prior arrests and previous incarcerations, and ADAM collects data on the criminal charges for which arrestees are held at the time of the interview. Respondents are also asked numerous demographic questions concerning age, ethnicity, citizenship status, education, employment status and marital status. Because ADAM notes citizenship status and also measures several criminological variables for each respondent, it is possible to examine the link between arrestee citizenship and criminal conduct.
ADAM respondents are fully informed that their responses are not traceable to them as individuals; the questionnaire is conducted anonymously. Even if authorities sought out such information, government officials could not identify those individuals with prior arrests. Participation in the study is completely voluntary. Respondents can refuse to participate at all, or participate but refuse any questions they feel to be too sensitive, too personal, or too threatening.
There are some limitations to the ADAM data worth noting here. ADAM samples only recently arrested adults in Anchorage. Therefore, results are not generalizable to juveniles; are not applicable to those who commit crimes but are not arrested; and are not representative of other localities in Alaska or the nation as a whole. Federal arrestees are not included in the sample.
Non-citizen Immigrants among Anchorage Arrestees
To better understand the relationship between citizenship status and crime, it was first necessary to explore the extent to which non-citizen arrestees were part of Anchorage’s total arrestee population. Between 2000 and 2002, more than 9 out of every 10 arrestees booked into Anchorage jails reported they were born in the United States or one of its territories; only 4.9 percent were foreign-born (see Table 1). When compared to the general U.S. population, or the population of the entire state of Alaska, the foreign born—both citizens and non-citizen residents—formed only a very small proportion of the arrestee population in Anchorage. According to the 2000 U.S. census, 10.4 percent of the American population was foreign-born; for Anchorage the figure was just over 8 percent.
Of those arrestees who were not born in the U.S., better than half (50.5%) had become naturalized citizens by the time of their interview with ADAM site staff. At most, only 47 arrestees in the sample (2.4% of total) were not U.S. citizens at the time of their arrest. (In 5 of these 47 cases, no arrestee response was recorded on this point.) Hence, non-citizen arrestees constituted a very small minority of the jailed population in Anchorage.
When respondents reported to interviewers that they were not born within the U.S., they were asked to report their country of origin. As Table 2 shows, foreign-born arrestees in Anchorage—both citizens and non-citizens—come from a wide spread of countries.
In terms of individual demographic characteristics, non-citizen arrestees in the sample differed in several ways from arrestees (native and foreign-born) with American citizenship (see Table 3). Arrestees without U.S. citizenship were, on average, two years older than citizen arrestees. Non-citizens, like their citizen counterparts, were much more likely to be male than female, although non-citizens were proportionally more usually male than arrestees who were citizens. Those arrestees without citizenship were also more likely to have been married, either at the time of the interview or prior to it, than American arrestees, who reported never having been married in more than half (54.7%) of all cases. Non-citizen arrestees were also more likely than citizen arrestees to have a post-secondary education, and in particular were much more likely to report having attained a four-year college degree or even higher level of education prior to their arrest. Finally, non-citizen arrestees were more likely than citizen arrestees to report having a job, especially for full-time jobs. Thus, in the aggregate, non-citizen arrestees in the ADAM sample were more likely than their citizen counterparts to be older, male, non-white, married, college-educated and employed.
The Association between Citizenship and Prior Arrests
To begin the examination of the association between immigrant status and crime, the citizenship status of arrestee respondents was compared to several criminological variables. The first comparison was between citizenship status and prior arrest record. The analysis revealed that respondents who were not citizens were less likely than citizen respondents to report a prior arrest (72.3% versus 85.4%). Next, three other crime variables which tap the nature of the current arrest, and which are frequently mentioned in discussions of criminal dangerousness, were also examined. The three variables were: (1) whether the respondent was arrested for a felony; (2) whether the respondent was arrested for a violent offense; and (3) whether the respondent was charged with a drug offense (not including alcohol) (Table 4). Although some differences emerged for these variables, statistically, there was no significant difference between citizen and non-citizen arrestees—that is, the differences were not greater than could be attributed to random variation.
To summarize, the observed difference in likelihood of prior arrest was the only significant difference found between the two groups across these crime variables. None of the other percentage differences exceeded what would be expected due to random variation. Overall, this first level of analysis revealed little difference in the criminality, past or present, of citizen and non-citizen arrestees in Anchorage jails.
The next stage of analysis controlled for several demographic variables to ascertain whether citizenship status alone had value as an element predictive of prior arrest—that is, once the demographic factors of age, gender, race, marital status, education and employment were taken into account, were non-citizen arrestees still less likely to have had a prior arrest than their citizen counterparts?
Each of the demographic variables shown in Table 3 has been shown in criminological research to be associated with crime. Age, for instance, has a curvilinear relationship with crime (picture an upside down “U”), peaking in the mid-to-late teens, with the very young and very old not likely to engage in criminal activity much at all. Gender also has a clear association with most crime, with males much more likely than women to be both the perpetrators and victims. The findings concerning the relationship between race and crime are somewhat mixed, with the main effects of race often disappearing once other factors (particularly community-level variables) are controlled. Findings from studies of the criminal life course demonstrate that some significant life events such as marriage have a dampening effect on criminal activity. Finally, socioeconomic indicators, such as education and employment status, are generally believed to be inversely related to crime. Because of their established association with crime, we controlled for each of these factors in this analysis.
Results from this analysis, which used binary logistic regression, revealed that arrestees who are not citizens were less likely to have had a prior arrest, holding constant age, gender, race, marital status, education, and employment. In fact, citizenship status had the greatest predictive power; that is, one was better able to predict whether or not a respondent reported a prior arrest by knowing citizenship status than by having information about age, gender or marital status—although these variables were useful predictors of prior arrest as well.
Using the results of the logistic regression analysis, we could calculate the probability of a prior arrest for any particular individual, provided we have information for each variable. For example, a 29 year-old white male citizen had a 79.4 percent chance of a prior arrest. In contrast, an individual who had the same characteristics, but who was not a citizen of the United States had a probability of prior arrest of 59.2 percent.
Thus, based on citizenship status alone, an Anchorage arrestee with citizenship was about 20 percent more likely to report having a prior arrest than an immigrant without citizenship.
Assessing the Validity of Arrestee Responses
Because ADAM data are self-reported, in this analysis it was necessary to address the question of veracity. By examining respondents’ answers to questions asking them about their recent drug use and comparing them with their urinalysis results, we can estimate a “truth score” for every respondent. A series of measures were developed to assess arrestees’ truthfulness for this study. The truthfulness measures were created by comparing urinalysis results with each arrestee’s response to the following question On how many of the past 7 days did you use [marijuana, powder cocaine, crack/rock cocaine, heroin, methamphetamine]? Arrestees who reported no drug use and who did not have a positive test were coded as being “truthful,” as were those arrestees who reported using a drug and who did have a positive result. In those cases where arrestees reported using a drug, but did not test positive in the urinalysis, the respondents were coded as being truthful because of willingness to report drug use, even if they did not use enough of it to be detected. When arrestees denied using a drug within 7 days of the interview, but drugs were detected in the urine samples, they were coded as “dishonest.” In cases where the respondents refused to answer the question about their drug use in the week preceding their arrest, when the question wasn’t applicable (they had not used in the past 30 days), or if there was a missing response (due to interviewer or data entry error), the case was excluded from the analysis.
The most striking finding reported was that both citizens and non-citizens were, on the whole, very forthcoming about their drug use. On average, well over 90 percent of arrestees told the truth about their drug use in the previous 7 days. To the extent that citizens and non-citizens differed in terms of their honesty with interviewers, non-citizens were somewhat more likely to tell the truth when asked about their recent drug use (though the observed percent differences were not statistically significant). To summarize, there is evidence that non-citizen immigrants were just as likely (if not more so) to respond as honestly as their U.S. citizen counterparts, lending credibility to other results in the analysis.
This analysis focused empirically on the notion that non-citizens are a “dangerous class” within the broader community by examining the association between citizenship status and several measures of criminality.
While the data used for this study have limitations in the study of immigration and crime, they provide a sound empirical starting point. Based on the findings presented here there is little, if any, evidence to believe that non-citizen arrestees in Anchorage, Alaska are any more “dangerous” or “criminal” than their American citizen counterparts. This analysis found that not only was there no support for the notion that non-citizens are more criminal, but the reverse was discovered—that is, arrestees with citizenship demonstrated a greater likelihood of a prior criminal history.
Brad Myrstol is a research associate with the Justice Center.