Last Friday, President Obama put immigration back on the front pages by announcing an end to deportations for illegal/undocumented immigrants who came to the U.S. as children. The issue of how to handle undocumented immigration is likely to stay in the news in the weeks to come, with the Supreme Court set to rule on Arizona’s 2010 immigration law. But do Americans’ reactions to undocumented immigrants vary with those immigrants’ personal characteristics, such as their appearance or their ability to speak English?
One of the most common answers to that question emphasizes ethnocentrism and inter-group similarity. In this view, we are likely to favor more lenient policies toward immigrants that are more like us, whether in how they look or how they speak. This approach might explain why 20% of respondents to a 1997 Knight Ridder poll thought immigrants from Mexico had done the most to create problems for the U.S., while just 1% said the same of European immigrants.
With support from the Russell Sage Foundation, I ran a survey experiment in 2010 to test the influence of inter-group similarity on attitudes about creating a pathway to naturalization. In all, just under 2,000 non-Hispanic white and black Americans sampled through Knowledge Networks participated in the experiment, which involved watching variants of an ABC News clip that had been experimentally altered. In the news clip, Diane Sawyer first describes a 2007 Senate compromise on immigration that would have required illegal immigrants to pay a $5,000 fine prior to naturalizing. I added a segment featuring “an immigrant who would be affected” at the end of the clip and I randomly varied both the blurred photograph in which he appeared and the voice-over through which he spoke. My respondents then indicated their support or opposition for creating a pathway to naturalization similar to what had just been described. In the control group that saw no video, 45% were somewhat or strongly supportive of the policy.
The experimental results—written up in an unpublished manuscript here—run against some of our most common intuitions about ethnocentrism and in-group favoritism. Check out the right side of the Figure below. A substantial shift in the featured immigrant’s skin tone has no impact on Americans’ support for a pathway to naturalization. To be sure, the immigrant’s appearance was otherwise similar, and given the broader context, respondents were likely to perceive him as coming from Latin America in any case. But on its own, the immigrant’s skin tone did nothing to change attitudes.
How the immigrant spoke was another matter. The videos always showed subtitles while the immigrant spoke. But in some cases, the immigrant spoke English with a very slight accent, while in others, he spoke with a pronounced accent or else in fluent Spanish. If what matters is similarity to the respondents, we should expect that the immigrant with only a slight accent will induce more support for a pathway to naturalization, and that the Spanish speaker will have the opposite effect. The thick accent should fall in the middle, as it is intelligible while clearly identifying the speaker as foreign. But the results, shown on the left side of the Figure just below, undercut these expectations. As compared to a control group that saw no video, hearing fluent Spanish or nearly unaccented English had no discernible effect. But for those who heard the pronounced non-native accent, the effect was striking, and equal to 18% of a standard deviation. Those respondents who heard a pronounced accent were more supportive of creating a pathway to naturalization than any other group, including those who heard the much subtler non-native accent. In the Figure, we see each experimental group’s mean difference from a control group who saw no video, with the thin vertical lines reflecting 95% confidence intervals. The number under each figure is the corresponding two-sided p-value.
What to make of that pattern? In January of 2011, I replicated the experiment in a second Knowledge Networks survey using a different image, to ensure the results’ basic robustness. The pattern emerged again. I’ve examined the treatments’ effects on a variety of different types of immigration attitudes. I’ve also done follow-up studies using Amazon’s Mechanical Turk interface. There isn’t much evidence that the fluent English speaker is perceived as more of a labor market threat. Instead, the accented English seems to be a signal that the immigrant wants to assimilate, and it also increases the perception of the featured immigrant as typical of immigrants generally. The U.S. is a nation of immigrants, and as an immigrant tries to speak English, that effort might cue positive associations about assimilation. What’s clear is that differences in speech or appearance do not always divide—and that in some cases, immigrants who speak with an accent might be viewed more positively as a consequence.
[Note: cross-posted at the Russell Sage Review]