We report results from a survey of 603 returned migrants and potential first-time migrants who were interviewed in their homes in Mexico by a team of bilingual US and Mexican interviewers during January 2005. The research sites were Tlacuitapa, Jalisco, and Las Animas, Zacatecas, rural communities with high rates of migration to the USA, located in states that traditionally have sent large numbers of migrants to the USA. The research communities were chosen purposively to take advantage of extensive baseline data from previous surveys of migration behavior conducted in these towns (Cornelius 1976, 1991, 1998; Mines 1981; Goldring 1992).
A standardized questionnaire was administered to at least one adult in every dwelling unit that was occupied during the fieldwork period. Because of the small sizes of the populations of the research communities (800–1,500), no sampling was necessary. In each dwelling, the interviewer was instructed first to interview the male head of household. If the male head of household was unavailable throughout the fieldwork period, interviewers were instructed to interview his wife about her husband’s migration experiences. If at that time the wife volunteered that she had migration experience of her own, she was interviewed concerning her own migration experience as well. After interviewing the head of household, the questionnaire was next administered to all sons and daughters of at least 15 years of age. We administered the standardized questionnaire only to people aged 15–65, as we expected to find most of the current and potential migrants in this age range. Of the 603 persons interviewed, 68% were categorized by interviewers as having their principal base in the sending community, whereas 31% were based primarily in the USA and were making short visits to their hometowns at the time of our fieldwork.
The questionnaire contained a total of 143 items (see Cornelius & Lewis 2006, appendix A). In addition to questions pertaining to basic demographic attributes, the questionnaire contained sections on employment and residency in 2004; the migratory history of the family from 1995 to 2005; the migratory history of the interviewee; intentions to migrate in the 12 months following the interview; employment and life in the USA; perceptions of the interviewee’s hometown and his economic situation; and plans for the future. Although most of the questions were closed, open-ended questions were included to elicit more fine-grained information on various aspects of the US migration experience. The average administration time was 50 minutes.
In general, if the “prevention through deterrence” strategy were effective, we would expect people to become less inclined to migrate as: (i) their information about enhanced US border enforcement measures increases; (ii) perceptions of risk and danger increase; (iii) actual negative experiences during past crossings increases. A show of force at the border can only be effective if people are aware of heightened restrictions and that they perceive and/or have actually experienced that such policies make crossing much more difficult. An ideal test of deterrence theory would gauge people’s attitudes before and after the implementation of border enforcement policies. We cannot do so with our cross-sectional research design. However, we are able to determine if our respondents’ knowledge, perceptions, and experience with border enforcement policies are important determinants of their decisions to migrate. Thus, we can estimate the relative significance of economic/demographic factors and immigration control policies.
Our main dependent variable for the analysis that follows (Q71) asks whether the person being interviewed intends to migrate at some time during 2005.5 While we realize that there may be some slippage between stated intentions in January 2005 and actual migration outcomes during the year, we believe that this question can reliably get at the type of person who is most likely to migrate. Of the people who responded to this question, 51% responded in yes, that they have at least considered migrating to the USA.
One way of ascertaining a deterrent effect is to simply ask the “no” respondents why they do not wish to migrate. As such, we asked these people to give the main reason why they were not willing to migrate (Q77). Although lack of economic need, lack of interest, and family considerations dominated the responses to this question, 41 people answered that difficulty crossing was their main reason for staying home, whereas an additional 14 people answered that they could not afford coyote and/or transportation costs. In all then, 55 out of 603 survey respondents indicated that they were deterred from crossing because of the direct or indirect effects of US Border Enforcement policies.6 However, it is also important to consider the perceptions, information, and experiences of those who do wish to migrate along with additional control variables.
As our main variables of interest, we include survey items dealing with perceptions of difficult and danger crossing the border. One question (Q80) asked about perceptions of difficulty in evading the US Border Patrol in the current period. Six percent of our respondents indicated that it is not more difficult to cross; 23% answered that it is now somewhat more difficult; 66% responded that it is much more difficult to cross; and 5% answered that it is now virtually impossible to cross. A second question (Q78) asked interviewees about their level of information regarding current US Border Patrol policies. People were asked if they were aware of current efforts to make unauthorized crossings into the USA more difficult. In answering this question, 72% of the respondents indicated that they were aware of heightened security at the border. Third, we asked a question (Q84) about the perceptions of danger in crossing without legal documents. The overwhelming majority (80%) of survey respondents answered that it is very dangerous to cross; only 20% believed that it is only somewhat dangerous or not at all dangerous. Although this question is subjective, we also asked people if they actually knew someone who had died while attempting to cross into the USA (Q85), as people who knew someone who died may be more directly attuned to this extreme risk. Sixty-four percent of those who answered the question indicated that they did know someone who died en route to the USA.
To summarize, we found our interviewees to be well informed about Border Patrol efforts; indeed, a large majority believed that it is much more difficult to surmount the obstacle course at the border. Moreover, a majority believe that it is much more dangerous to cross the border clandestinely today as compared with previous periods. Nevertheless, more than half (51%) reported that they were considering a journey north.
Although these perceptual factors are important, we also asked people with a previous migration history about their personal experience crossing the border. Perhaps direct experience is more important than perceptions. In our survey sample, 64% or 383 individuals indicated that they had migrated to the USA before (of these, 184 were undocumented). Of those who had previously migrated, we asked whether they were apprehended by the Border Patrol on their most recent trip to the border (Q62). Of the entire subset of people who had crossed before – with or without legal documents – 13% indicated that they had been caught by the Border Patrol; of only those individuals crossing without papers, 25% indicated that they had been caught trying to cross.7 We also asked whether the most recent trip to the USA was more difficult than they had anticipated versus less difficult/about the same (Q58). Twenty-two percent of experienced migrants reported that the crossing was harder than they had expected. Restricting this analysis to only those crossing without papers, 44% indicated that their journey was more difficult than expected.
These variables form the core of our analysis. In the regressions that follow we include these variables about potential migrants’ level of information about Border Patrol efforts, their perceptions of difficulty/danger in crossing the frontier, and in separate regressions restricted to experienced migrants, we include information about their past attempts at crossing. Summary statistics for each of these main independent variables, along with our main dependent variable of interest are presented in Table 1.
Table 1. Summary of key variables
|Considering to migrate (dependent variable)||No (49%); Yes (51%)|
|Information about border enforcement||None (28%); Informed (72%)|
|Difficulty evading Border Patrol||Not at all (6%); Somewhat (23%); Much more (66%); Impossible (5%)|
|Perception of danger||Not at all or somewhat (20%); Very dangerous (80%)|
|Knows someone who died||No (36%); Yes (64%)|
|Caught on prior attempt†||No (87%); Yes (13%)|
|Prior attempt more difficult†||No (78%); Yes (22%)|
We also include several control variables in the analysis. Much of the published work suggests that the typical Mexican migrant to the USA is a working age man.8 Therefore we include a dichotomous variable for sex (female, 1), along with age and age squared to account for a parabolic relation between age and propensity to migrate (the very young and the elderly are less likely to migrate). We include additional demographic controls for marital status (married, 1) as well as the number of children the respondent has. We also include a pair of controls for the respondent’s economic status. Whereas we lack wage data for our individual respondents, we include a subjective self-assessment of economic status in which people were asked to rate their economic welfare on a scale from 1 to 10. We also include a variable for the number of years of schooling the respondent has completed; while education may have an independent effect on the propensity to migrate, education is also expected to be highly related to one’s income. In addition, because there may be unique characteristics of the two towns represented in our study that are not included in the statistical model, we include a “fixed effect” term for the town itself in the form of a dummy variable for “Las Animas”. Finally, when restricting our models to the subset of respondents who had migrated before, we include a dichotomous variable for the person’s legal status (documented, 1; undocumented, 0).
Because our dependent variable is dichotomous, we run our models using a logit estimator with robust standard errors. Because several of our independent variables may be highly correlated with each other, we include them sequentially before presenting a model in which all are included. Additional diagnostic testing shows that multicollinearity does not present a significant problem.9 Our most correlated variables (danger crossing and difficulty evading the Border Patrol) were only correlated at the 0.27 level.
We acknowledge that our survey design may suffer from a particular type of response bias, but we can anticipate the direction of this bias. By conducting our interviews in Mexico, our survey does not include people who have already migrated, did not return to their hometowns during the fieldwork (which was timed to coincide with the towns’ annual fiestas), and therefore were not available to be interviewed. Thus, the sample may overrepresent people who stayed in Mexico and underrepresent people who had left. Clearly, undocumented migrants who were in the USA at the time of the survey were not deterred from migrating. Therefore, if there is a deterrent effect, we are more likely to detect it among those who have been successfully dissuaded by border enforcement efforts. This suggests that our results should indicate something of a bias in favor of a finding that the deterrence strategy has been successful. As we find little evidence of deterrence, this type of response bias is not cause for major concern.