Abstract
Donald Trump was particularly vocal in shaping his presidential campaign around policies perceived as being anti-immigration. Consequently, many were shocked that Hispanic support for the Republican Party did not drop in the 2016 presidential election. In fact, our survey, which consisted of 1,080 people of Hispanic descent living in the United States, found that 74% of Hispanic Trump voters were in favor of generally deporting all illegal immigrants. Our results suggest that the population of Hispanics who voted in the 2016 presidential election was, on average, more conservative than the overall population of Hispanics living in the United States. Furthermore, our analysis suggests that issues such as the economy, health care, and education were more important to Hispanic voters than were issues related to immigration.
During his 2016 presidential announcement speech, then-candidate Donald Trump commented on his view of Mexican immigration. In what became arguably one of his most controversial statements (Reysen & Brown, 2015), he said: When Mexico sends its people, they’re not sending their best . . . They’re sending people that have lots of problems, and they’re bringing those problems with us. They’re bringing drugs. They’re bringing crime. They’re rapists. And some, I assume, are good people. (Time, 2015)
This statement set the tone for the strongly anti-immigration stance of Donald Trump’s presidential campaign. Many people expected such a stance to lead to a significant loss in Hispanic votes for the GOP compared to the previous 2012 presidential election (Khalid, 2016). This predicted loss did not happen. According to exit polls, the percentage of Hispanics that voted for the Republican presidential candidate actually increased from 27% to 28% in the 2016 election (Pew Research Center, 2016).
This somewhat unexpected result raises questions about how issues such as immigration, deportation, and the treatment of racial minorities actually affect Hispanic voting preference. In this study, we explore two different theories aimed at explaining the discrepancy between expectations about the Hispanic vote in the 2016 election and the actual outcome:
The population of Hispanics who voted in the 2016 election was significantly different from the population of Hispanics in the United States as a whole at the time of the election.
Hispanic voting preference in the 2016 election was heavily influenced by issues other than immigration, deportation, and the treatment of racial minorities.
The first of these two theories hinges on the idea that Hispanics living in the United States should not be thought of as a monolithic whole in regard to their political views. Instead, this theory assumes that Hispanic political opinion varies widely based on demographic features such as age, language ability, acculturation level, and country of origin. With this difference in political opinion in mind, this theory explains the increase in Hispanic support for the GOP in the 2016 election by assuming that there was a significant difference in the demographic features of Hispanics who voted in the election as compared to Hispanics who did not vote in the election.
The second of these theories explains the increased Hispanic support for the GOP in the 2016 election as a result of Hispanic voting preference being driven more by issues such as the economy, education, or health care than by issues such as immigration, deportation, or the treatment of racial minorities. This theory assumes that, contrary to popular opinion (Kumar & Ordoñez, 2018; Martin, 2018), Hispanics in the United States do not vote solely based on issues related to immigration.
To test these theories, we use data gathered from a survey we conducted in January of 2018 which includes responses from 1,080 people of Hispanic descent living in the United States or Puerto Rico. It is thought that our analysis of these data will provide further insight into why Hispanic support for the Republican Party did not decrease in the 2016 presidential election.
Literature Review
We briefly review prior research on the topic of Hispanic political opinion and voting trends. Based on this review of the literature, we claim that the work of previous researchers verifies two of the main assumptions necessary to the plausibility of our theories: (a) Hispanic political opinion is not homogeneous and (b) there are issues besides immigration that influence Hispanic voting trends.
To start, we address the issue of how the grouping of “Hispanics” came to be and whether or not it can be considered distinct compared to other cultural groups. The word “Hispanic” became popular during the 1970s as an umbrella term used in the U.S. census to encompass all Spanish-speaking or Spanish-surnamed ethnic groups (Cohn, 2010; Fears, 2003; Mora, 2014). From its inception, this term has been met with controversy (Fears, 2003; Gonzalez, 1992). While much of the debate centers around whether or not “Hispanic” is the most accurate or culturally sensitive term that could be used to describe those of Spanish-speaking origin living in the United States (Gonzalez, 1992), there is also contention about the composition of the group itself. The cause of this contention can be seen as a tradeoff between respecting diversity and promoting unity. Some argue that focusing on Hispanics as a single group gives them access to greater political representation and rights in general (Mora, 2014). However, others argue that this imposed homogenization of people from widely disparate backgrounds engenders resentment among those who wish for a more specific identifying term (Gonzalez, 1992). Support for the diversity side of this argument can be seen in the results of a 2012 survey of Hispanics conducted by the Pew Research Center which found that the majority of those surveyed preferred to be described by their family’s country of origin rather than by pan-ethnic terms like “Hispanic” or “Latino” (Taylor et al., 2012).
Prior research on Hispanic political opinion yields support for both sides of this argument of unity versus diversity. While some research shows that there are certain policy preferences that make Hispanics distinct from other cultural groups (Claassen, 2004; Segura, 2012; Taylor et al., 2012), other research highlights differences in political opinion among Hispanics driven by demographic features such as time lived in the United States (Abrajano & Alvarez, 2011), level of acculturation (Branton, 2007; Callister et al., 2019; Hood et al., 1997; Rouse et al., 2010), country of origin (Garcia et al., 1996), language ability (Callister et al., 2019; Newton, 2000), and gender (Bejarano, 2014; Montoya, 1996). Such research suggests that Hispanics who have lived in the United States longer (Abrajano & Alvarez, 2011), prefer English over Spanish (Callister et al., 2019; Newton, 2000), are male (Bejarano, 2014; Montoya, 1996), and have higher levels of acculturation to U.S. society (Branton, 2007; Callister et al., 2019; Hood et al., 1997; Rouse et al., 2010) are more likely to have conservative ideological views than Hispanics who have lived less time in the United States, prefer Spanish over English, are female, and have lower levels of acculturation to U.S. society.
This variance in political opinion according to different demographic features explains why some suggest that Hispanic voters are not unified by either party affiliation or ideological leaning (Nagler & Alvarez, 2004, p. 393). These results suggest that while it may be informative to compare Hispanics as a whole to other ethnic groups, the grouping of Hispanics itself should not be thought of as completely homogeneous.
In addition to not being a monolithic whole with regard to their stances on political issues, prior research has also shown that Hispanics vary in their propensity to vote. While it has been shown that Hispanics typically vote at the same rate as other ethnic groups in presidential elections (Cassel, 2002), foreign-born Hispanics vote at lower rates than U.S.-born Hispanics (Barreto, 2005). This finding, along with the notion that Hispanic political opinion differs according to demographic features, lends credence to the theory that Hispanic support for Donald Trump in the 2016 presidential election could have been the result of a disproportionate number of conservative Hispanics showing up to vote.
Having established the plausibility of the idea that the population of Hispanics who voted in the 2016 election was different from the population of Hispanics living in the United States as a whole, we now turn to the results of prior research on issue salience among Hispanic voters. While such research suggests that a given presidential candidate’s policy stance on immigration and his or her “ability to convey care and concern to the Latino community” affect Hispanic vote choice (Collingwood et al., 2014), it is also thought that issues such as the economy, national security, and moral values influence Hispanic vote choice (Abrajano et al., 2008). Thus, while it has been shown that immigration is an important issue to Hispanic voters, it should not be thought of as the only issue about which they are concerned. This idea that Hispanic voters are influenced by issues other than immigration makes plausible the theory that Hispanic support for Donald Trump in the 2016 presidential election was a result of Hispanic voters placing greater importance on nonimmigration-related issues.
Methodology
Having established the plausibility of our theories given the results of prior research, we now turn to an explanation of the methods we used to test these two theories. In January of 2018, after approval from our university’s institutional review board, we distributed a survey through Qualtrics, Amazon Turk, and Facebook to people of Hispanic descent who were 18 years of age or older and currently living in the United States or Puerto Rico. Between the three platforms, the survey was open to respondents for a total of 14 days. In all, 950 responses were collected from Qualtrics, 38 from Amazon Turk, and 92 from Facebook. While the majority of responses came from Qualtrics, Amazon Turk and Facebook were also used in an effort to reach more people from underrepresented groups in the Qualtrics sample. After the administration of the survey, data from the three platforms was compiled into a single file and analyzed using Python 3.0 software.
Before starting the survey, participants were given the option to take the survey in Spanish or in English. In the survey itself, participants were asked about their age, gender, country of origin, time lived in the United States, and whether or not they currently knew someone living in the United States without documentation. In addition to these demographic characteristics, participants were asked about their vote in the 2016 presidential election. Those who responded that they had not voted in the 2016 presidential election were then asked to select a reason for not voting.
Participants were then directed to a block of questions about their political preferences. Included in this block were the following four questions:
On a scale from 1 to 7, to what extent do you agree with President Trump’s request to build a wall between Mexico and the United States?
Since the campaign and election of Donald Trump, do you feel levels of discrimination toward Latinos/Hispanics have increased? (1–7 scale)
Mark the statement that you most agree with: (a) Generally, illegal immigrants should not be deported. (b) Generally, illegal immigrants should be deported.
Of the following issues, choose the three most important to you: the economy, treatment of racial minorities, education, deportation/immigration, health care, terrorism, social security, the environment, guns, and violence.
To test the theory that the population of Hispanics who voted in the 2016 election was different from the population of Hispanics living in the United States in general, we first split the data by those who voted in the 2016 election and those who did not. We then split these two groups by whether or not a given participant was born in the United States, gender, whether or not a given participant took the survey in Spanish, time lived in the United States, whether or not a given participant knew an undocumented immigrant, and age. For each of these demographic characteristics, we performed a chi-square test with the null hypothesis that there was no difference between the proportion of people with a given characteristic in the “voted” group and the proportion of people with the same characteristic in the “did not vote” group. Based on our first theory, we expect these chi-square tests to yield significant results, indicating that there is a difference between the people in our sample that voted in the 2016 election and the people in our sample that did not vote in the 2016 election with regard to the demographic characteristics mentioned.
In addition to chi-square testing, we performed logit and ordinary least squares (OLS) regressions with whether or not a given participant voted as the response variable and the same demographic characteristics as explanatory variables. We ran these models using both our entire sample and only those in our sample who could legally vote in the 2016 election. The results of the logit regressions are reported as the marginal effect (dy/dx) that a given explanatory variable has on the response variable reported at the means. Thus, the coefficient value of a given explanatory variable in these regressions can be thought of as the change in probability of having voted resulting from a one-unit increase in the explanatory variable for the “average” person in our sample. Similarly, the coefficient terms of the OLS regressions can be thought of as the change in probability of having voted that would result from a one-unit increase in a given explanatory variable. Significance in the OLS models is reported using heteroskedasticity-robust standard errors. Only explanatory variables that had a significant effect on the response variable are reported in each model.
For all four of the models, the percent of correctly predicted outcomes and the mean squared prediction error are given as goodness of fit measures. The percent of correctly predicted outcomes for a given model is calculated at the 0.5 threshold. The mean squared prediction error is calculated by taking the average of the squared differences of the actual values of the data points (0 or 1) and their corresponding predicted values. Having a higher percent correctly predicted and a lower mean squared prediction error indicates a better fit. In addition to these two goodness of fit measures, adjusted R2 is reported for the OLS models.
Based on our first theory, we expect the explanatory variables in each model to have a significant effect on the probability of having voted in the 2016 presidential election. In addition, we expect that explanatory variables that have been shown to be correlated with more conservative views to have a positive effect on the probability of having voted and variables that have been shown to be correlated with more liberal views to have a negative effect on the probability of having voted. Specifically, we expect having taken the survey in Spanish and knowing an undocumented immigrant to make people less likely to vote, and we expect being male, being older, having been born in the United States, and having lived in the United States longer to make people more likely to vote.
While both the chi-square tests and the regressions discussed test the idea that the voters in our sample differ with respect to demographic features from the nonvoters in our sample, these tests do not directly confirm or discredit a difference in ideological preference between the two groups. For this reason, we also performed two difference-in-means tests and a difference-in-proportions test for three of the politically-based questions in our survey. We first compared the mean reported support for Trump’s plan to build a wall among those who voted and those who did not vote. Next, we compared the mean perceived increase in discrimination against Hispanics since the election of Donald Trump among voters and nonvoters. Finally, we compared the proportion of people who agreed that, generally, all illegal immigrants should be deported among those who voted and those who did not vote. Given our first theory, we expect all three tests to yield a significant result. In addition, we expect support for Trump’s plan to build a wall to be higher among those who voted and the perception of increased discrimination to be higher among those who did not vote. We also expect the proportion of people who agreed that, generally, all illegal immigrants should be deported to be higher among those who voted.
In addition, to get a general sense of political ideology across various vote status, we compared mean support for Trump’s plan to build a wall among Trump voters, Clinton voters, those who voted for a third-party candidate, those who chose not to vote, and those who could not legally vote. We also compared the mean perceived increase in discrimination of Hispanics because of Trump’s election and the percent of those who agreed that, generally, all illegal immigrants should be deported among these groups. For the two questions involving means, we used an analysis of variance (ANOVA) model to test the null hypothesis that all means were equal. For the question involving proportions, we used a chi-square test.
To test our second theory that Hispanics who voted in the 2016 election were influenced by issues other than immigration, we report the results of our survey question asking participants to pick their top three most important issues. Since each participant was able to choose three different issues, we report the percent of total responses garnered by each issue. Given our second theory, we expect “deportation/immigration” and “treatment of racial minorities” not to be among the top issues for those who voted in the 2016 election.
Results
As predicted, our analysis of those in our sample that voted in the 2016 election as compared to those who did not vote revealed significant differences between the two groups with respect to country of origin, gender, time lived in the United States, and age (see Table 1). We found that among those who voted in our sample, there were higher proportions of participants who had been born in the United States or Puerto Rico, were male, had taken the survey in English, had lived in the United States more than 10 years, and were older than 50. However, contrary to what we expected, there was no significant difference between the two groups with respect to whether or not a given participant knew an undocumented immigrant living in the United States. In fact, the proportion of participants who knew an undocumented immigrant was slightly higher among those who voted as compared to those who did not vote.
Demographic Features of Those Who Voted and Those Who Did Not Vote in the 2016 Presidential Election.
The results of the logit and OLS regressions in Table 2 provide further support for the idea that those who voted in our sample differed from those who did not vote with respect to various demographic characteristics. These results show that, among our entire sample, taking the survey in Spanish decreased the likelihood that a given participant voted in the 2016 presidential election and that being male, being over the age of 50, and having lived in the United States more than 5 years all increased the likelihood that a given participant voted in the 2016 presidential election. However, given that there were 98 participants who indicated that they did not vote in the 2016 election because they could not legally vote, it could be that these differences exist only among Hispanics in general and are not present when comparing Hispanics voters with Hispanics who chose not to vote despite being eligible to do so. The logit and OLS models under “Only Those Who Could Legally Vote” in Table 2 rule out this possibility. These models show that while those who did not vote but could have voted did not differ in all of the same ways that those nonvoters differed from voters, this group did differ from voters with respect to language preference, gender, age, country of origin, and knowing an undocumented immigrant. Thus, data from our sample suggest that among both Hispanics in general and among only eligible Hispanic voters, those who voted in the 2016 presidential election were more likely to have certain demographic characteristics which have been shown to be correlated with a more conservative political view than those who did not vote.
Models for Whether or Not Participants Voted.
Note. OLS = ordinary least squares.
p < .05. **p < .01. ***p < .001.
Our analysis of three of the politically based questions in our survey further suggests that those who voted in the 2016 election had more conservative political views than those who did not vote. Table 3 shows that, as compared to those who did not vote, those who voted had a significantly higher mean support for Trump’s plan to build a wall, a significantly lower mean perceived increase in discrimination since Trump’s election, and a significantly higher proportion of people who agreed with the statement that, generally, all illegal immigrants should be deported. These results suggest that those in our sample who did not vote in the 2016 election held less conservative views about immigration and the treatment of racial minorities than those who did vote in the 2016 election.
Ideological Differences of Those Who Voted and Those Who Did Not Vote in the 2016 Presidential Election Among Entire Sample.
However, as discussed previously, it could be that this result only holds among our sample as a whole and does not hold among only those who could legally vote in our sample. Table 4 shows the results of these two difference-in-means tests and difference-in-proportions test among only those who could legally vote in our sample. These results show that while there was a significantly higher mean support for Trump’s plan to build a wall and a significantly higher proportion of people who agreed that, generally, all illegal immigrants should be deported among voters as compared to those who chose not to vote despite being eligible to do so, there was no significant difference in the mean perceived increase in discrimination between the two groups. Still, these results suggest a difference in political ideology between those who voted and those who did not vote but were still eligible to vote in our sample.
Ideological Differences of Those Who Voted and Those Who Did Not Vote in the 2016 Presidential Election Among Only Those Who Could Legally Vote.
Table 5 shows mean support for Trump’s plan to build a wall, mean perceived increase in discrimination of Hispanics since Trump’s election, and the percent of participants who agreed that, generally, all illegal immigrants should be deported across various vote status. From the table, it can be seen that, overall, there was a significant difference in political ideologies among Trump voters, Clinton voters, those who voted for a third-party candidate, those who chose not to vote, and those who could not legally vote in our sample.
Ideological Differences by Vote Status in 2016 Presidential Election.
Note. ANOVA = Analysis of variance.
Figure 1 shows the issues that participants thought were most important split by vote status. This figure indicates that “the economy,” “health care,” and “education” were the most important issues among both those who voted and those who did not vote in the election. Overall, “treatment of racial minorities” was the fourth most important issue among participants and “deportation/immigration” was the seventh most important issue among participants. Among those who voted, “treatment of racial minorities” was still the fourth most important issue, but “deportation/immigration” was the eighth most important issue. Among those who did not vote, “treatment of racial minorities” was the sixth most important issue and “deportation/immigration” was the fifth most important issue. These results suggest that while “treatment of racial minorities” and “deportation/immigration” were both important issues for those who took our survey, “the economy,” “health care,” and “education” were more important. Even if responses for “treatment of racial minorities” and “deportation/immigration” were combined into a single category, this category would still only be the fourth most important issue for both those who voted and those who did not vote.

Top issues by vote status.
Discussion
The results of our analysis suggest that a disproportionate number of conservative Hispanics turned out to vote in the 2016 presidential election and that, among those who voted, issues such as the economy, health care, and education took precedence over issues such as immigration and the treatment of racial minorities. Within the context of voter turnout in general in the 2016 election, these findings are consistent with ideas about the election which explain Donald Trump’s victory as the result of an abnormally low turnout of left-leaning voters (Ben-Shahar, 2016; Enten, 2017; Regan, 2016; Waldman, 2016). Proponents of this theory frequently point to a lack of support for Hillary Clinton among less conservative voters as the reason behind why they did not turn out for the election. Whether or not this was the case among eligible Hispanic voters during the 2016 election, it appears that the strongly anti-immigration stance of Donald Trump’s campaign was not enough to motivate many left-leaning Hispanics to vote. As a result, our analysis suggests that the Hispanic vote in the 2016 election was not representative of Hispanic political preference in general.
Perhaps one of the reasons the Hispanic vote in the 2016 election came as such a surprise to many was the prevailing opinion that Hispanics would be heavily impacted by issues related to immigration. This idea often derives from associating Hispanics in general with Hispanic immigrants. The stereotypical idea of Hispanics as a foreign-born, culturally distinct subset of the U.S. population leads many to the false conclusion that all Hispanics place equal importance on the issue of immigration and are generally in favor of proimmigration policies. The results of our survey suggest that while immigration and the treatment of racial minorities are important issues to Hispanics, they may not be the most influential issues in determining Hispanic vote. In addition, our results showed that while “deportation/immigration” was only the seventh most important issue among participants in general, it was the third most important issue among those who could not legally vote in the 2016 election. This suggests that while noncitizen Hispanic immigrants place a high importance on the issue of immigration, Hispanic citizens do not.
Also of interest from our results was the wide gap between Trump voters and non-Trump voters in our sample with respect to different ideological preferences. In our results, it was not surprising to see that, among Trump voters, there was a higher mean support for Trump’s plan to build a wall, a lower mean perceived increase of discrimination since Trump’s election, and a higher percent of people who agreed that, generally, all illegal immigrants should be deported. However, the extent to which Trump voters differed from other groups on these issues was surprising. Whereas Clinton voters, third-party voters, and nonvoters all had a percent of people who agreed that, generally, all illegal immigrants should be deported between 20% and 32%, 74% of Trump voters agreed with this statement. In addition, Trump voters had an average support for Trump’s plan to build a wall that was two points higher (on a 1–7 scale) than all other groups and an average perceived increase of discrimination that was two points lower than all other groups.
Limitations and Future Research
The results of our analysis are limited due to both the size of our sample and the fact that our sample was not randomly selected from the population of interest. Furthermore, our interpretation of the results of our survey relies on the assumption that people with certain demographic characteristics and ideological views would have been either more or less likely to vote for Donald Trump in 2016 had they actually voted. In addition, our analysis looked at only two possible explanations for the Hispanic vote in the 2016 election. There are undoubtedly other factors not discussed in this study that affected the outcome of the Hispanic vote in the election.
Given the results of this study showing that the salience of the issue of immigration varies among Hispanics according to legal status, future research should investigate how both negative and positive immigration messages influence Hispanic political involvement across varying levels of acculturation and legal status. It is thought that such research would further explain how the issue of immigration influences Hispanic political opinion and motivation to vote.
In sum, given the strongly anti-immigration stance of Donald Trump’s presidential campaign, it came as a surprise to many that Hispanic support for the Republican Party did not drop in the 2016 presidential election. Our analysis of our survey of 1080 people of Hispanic descent living in the United States or Puerto Rico suggests that this may be a result of the fact that the population of Hispanics who voted in the 2016 election was, on average, more conservative than the overall population of Hispanics in the United States. In addition, our analysis suggests that Hispanics in general placed more importance on issues such as the economy, health care, and education than they did on the issue of immigration in the 2016 election.
Overall, our findings suggest that Hispanic voters should not be thought of as a unified block. Future analysis of Hispanic voting trends should take into account variation in Hispanic political opinion based on different demographic features such as acculturation level and country of origin. In addition, our findings suggest that such an analysis should also take into account the idea that Hispanics who cannot legally vote are typically less conservative than Hispanics as a whole. It is thought that a consideration of these factors will lead to a more accurate characterization of Hispanic voting behavior.
Footnotes
Authors’ Note
Special thanks to Jessica Strong for her significant contribution to this project.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
