Does Income Really Influence Individuals’ Opinions on Undocumented Immigrants?

by Sharon Hoeck

During the Summer of 2014, thousands of undocumented, underage immigrants began arriving in the Southern United States, bringing renewed attention to the U.S.’s antiquated immigration policies. Opinions both of citizens and their representatives in government ranged from demanding that the minors be repatriated to arguing that sending them back to their country of origin would place them in danger and that they should be allowed to stay in the U.S. This study attempts to shed light on one small part of why individuals’ opinions vary so much regarding this topic. Specifically, it asks what effect an individual’s income level has on his or her opinion toward undocumented immigrants currently living in the U.S. Despite the existing body of work on this subject, very little work has been done to examine this relationship since the influx of undocumented minors last summer and President Obama’s current fight to reform U.S. immigration policy.

 
Existing research on this subject does suggest that these two variables are related. Most studies conclude that individuals who are economically vulnerable are less likely to support open immigration policies such as pathways to either permanent residency or citizenship because they see undocumented immigrants as competition for job opportunities or social services (Citrin et al. 1997; Burns and Gimpel 2000; Scheve and Slaughter 2001). These conclusions are despite the fact that non-partisan organizations such as the Congressional Budget Office have published studies showing that undocumented immigrants generally contribute more to the national and local economy than they use. Upon beginning this study and in light of previous research on the topic, I hypothesized that individuals with lower incomes would favor more restrictive policies toward undocumented immigrants, with middle- and upper-class individuals being more open to establishing pathways to permanent residency or citizenship.

In order to study the relationship between an individual’s income and his or her opinion on what should happen to undocumented immigrants, I used the Pew Research Center’s dataset from a political survey conducted at the end of July 2014. The timing of this survey is significant because it coincides with increased public awareness of and media attention on the flow of undocumented minors entering the country from Central and South America. In addition to income level and opinion on undocumented immigrants, I included six control variables in my study that both influence and are influenced by my two primary variables (Table 1; see Appendix for more detailed descriptions).

Table 1. Summary Statistics

Variable	Mean	Std. Dev.	Min.	Max.	Description
Fate of the Undocumented	2.10	0.83	1.0	3.0	What Should Happen? (1=Repatriation, 2=Perm Residency, 3=Citizenship)
Income	5.08	2.36	1.0	9.0	9-point scale where 9 = More than $150K/year
Party Affiliation	2.36	1.52	1.0	9.0	1 = Republican, 2 = Democrat, 3 = Independent 9= Refused/DK
Race	1.74	1.85	1.0	9.0	1=White, 2=Black/African-Am, 3=Asian/Asian-Am, 4=Other, 5=Native Am/Alaskan, 6=Pac. Islander/Hawaiian, 7=Hispanic/Latino, 9=Refused/DK
Highest Completed Level of Education	4.88	1.93	1.0	8.0	1=<high school, 2=Incomplete HS, 3=HS Grad, 4=Some College, 5=2-yr Degree, 6=4-year Degree, 7=Some Postgrad, 8=Postgrad/Professional Degree
Age	52.49	18.73	18.0	99.0	Respondent’s Age
Sex	1.47	0.50	1.0	2.0	1=Male, 2=Female
Border State	2.24	0.81	1.0	3.0	1 = Borders Canada, 2 = Borders Mexico, 3 = Interior
Observations	1805

Source: Pew Research Center, July 2014 Political Survey

In contrast to my initial hypothesis, graphing the relationship between an individual’s income level and his or her opinion toward undocumented immigrants showed a slight parabola, where opinion trends downward toward repatriation as income increases until it reaches approximately $45,000 per year, at which point opinion trends upward toward the approval of providing a pathway to citizenship (Figure A). Each of the boxes in the graph below shows the distribution of opinion toward undocumented immigrants in each of the income brackets, with the black horizontal lines in each box indicating the median—or average—of each bracket’s opinion (see Appendix). Interestingly, the graph shows that opinion regarding immigrants is generally scattered evenly in all income categories except the two most affluent, where opinions seem to favor more welcoming policies such as a path to permanent residency or citizenship. While the results of my analysis are largely insignificant (as I will discuss below), further research may benefit from looking at reasons why the very wealthy are more supportive of assimilation.

Upon examining the parabolic relationship through a regression analysis that includes my six control variables, I found that the slight parabolic relationship in Figure A is statistically insignificant. The regression table below shows that the illustrated parabola is just as likely to be due to chance than to any systematic pattern of opinion. The only two variables within the model that appear to be significant statistically are race and sex. The former of these seems intuitive, as individuals who are closely related to immigrants or are immigrants themselves might logically be sympathetic to the plight of the undocumented. The latter of these variables is less intuitive, as the results show that females are slightly more likely to support government programs allowing undocumented immigrants to stay in the U.S.

Table 2. Determinants of Opinion Toward Undocumented Immigrants

DV: Fate of the Undocumented	(1)	(2)	(3)
Income	-0.04		0.01
	(-0.88)		(0.12)
Income*Income	0.00		-0.00
	(0.73)		(-0.34)
Party Affiliation		0.02	0.02
		(1.34)	(1.29)
Race		0.04**	0.04**
		(2.71)	(2.73)
Education		-0.04	-0.04
		(-0.57)	(-0.49)
Education*Education		0.01	0.01
		(0.93)	(0.92)
Age		0.00	0.00
		(0.22)	(0.30)
Age*Age		-0.00	-0.00
		(-1.08)	(-1.15)
Sex		0.17***	0.17**
		(3.36)	(3.29)
Border State		-0.02	-0.02
		(-0.34)	(-0.38)
Constant	2.21***	1.90***	1.89***
	(20.23)	(6.99)	(6.70)
R-Squared	-0.00	0.03	0.03
Observations	972	972	972

Note: OLS estimates with t-stats in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001

Despite my statistically insignificant results, I do not think that further research should exclude economic factors. It would be helpful to examine these variables differently, however. For example, the income variable used in this model measures individuals’ household incomes from the previous year. A variable that measures respondents’ perceptions of the economy may be more useful in predicting his or her opinion regarding undocumented immigrants because it is an indicator of whether or not he or she feels economically secure. An individual more secure of the future may be more likely to extend welcome to the undocumented immigrants in the U.S. than a less secure individual might be.

Another change that may be helpful for future research would be to modify the regional variable. The border state variable in this model, for example, counts equally the opinion of individuals living on the Mexican border in San Diego, CA with those living almost 850 miles north of the border in Crescent City, CA. A variable that controls for communities within a hundred miles of the Mexican and Canadian borders would therefore be more representative of affected individuals.

In conclusion, although these results refute my hypothesis that lower income individuals would be less favorable toward allowing undocumented immigrants living the U.S. the chance to gain permanent residency status or citizenship, there is plenty of opportunity to refine the model given a more nuanced dataset. Because the influx of immigrants occurred less than a year ago, the data needed for such refinement is not currently available. However, this model could serve as a starting point from which to start a more comprehensive study.

Appendix

1. Control Variables:
   
   
   
   
   
   
   
   1. Party affiliation: the current public stances of the Republican and Democratic parties appear to be divided along socioeconomic lines and the discourse produced by the two parties generally takes very different sides of the immigration debate. In the model, I treated this as a categorical variable
   2. Race: I would hypothesize that individuals of Hispanic, Asian, or African descent may view permissive immigration policies more favorably because or their proximity to immigrant populations. I treated this as a categorical variable
   3. Education: an individual’s level of education both affects his or her future income, but also has some effect on worldview. For example, a more educated individual may be more aware of America’s roots as a largely immigrant population. This continuous variable showed a slightly parabolic relationship with the dependent variable.
   4. Age: younger individuals may make less money than those of middle age, but may have more personal relationships with young immigrants while older individuals may be nostalgic for the European-dominated immigrant patters of their youth.
   5. Border State: not only does the state of the economy vary widely by region, I hypothesize that living in a border state would have a strong impact on an individual’s opinion on what should happen to undocumented immigrants.
   6. Sex: females and males tend to differ in income potential over their lifetimes. This trend is partially due to the persistence of gender-bias in the hiring practices of some industries. Males may dominate the workforce in certain industries that employ more immigrants than the industries women find themselves in (or vice versa).  
      
      
      
      

Note on Figure A:

You may notice that there are two boxes overlapping around the $40,000 to $50,000 income bracket. The box in the foreground represents those respondents who refused to provide their household’s income and those who did not know their income. I moved these observations to the mean of the sample (just above the fifth income bracket) in an attempt to limit any bias in the regression model. Unfortunately, this action may have affected the statistical significance of my results.