In the aftermath of the 2020 election season, pundits and politicians have revived a familiar refrain: Public opinion polling is broken. Though sweeping, this claim certainly holds weight concerning polls of Latino voters. While polling samples have used ethnicity as an effective demographic weight for decades, amalgamating all Latino voters does more harm than good given the incredible ethnic, linguistic, and generational diversity in the Latino community. Homogenizing this diverse community, then, can present an imprecise picture of the political stances of Latino voters. As such, any post-mortem of the 2020 election ought to examine polling techniques for Latino communities, especially the failure to disaggregate Latino voters in samples and the lack of Latino diversity in opinion polling groups. Beyond its consequences for election forecasting, improving the accuracy of polling Latino voters could have substantial impacts on media narratives and policymaking in the United States.
Perhaps the most significant issue in creating Latino voter samples is the failure of opinion pollsters to weigh this ethnic subgroup by other traits predictive of voting patterns. The current system assumes that a sample of Latino voters is representative of the whole Latino community. This aggregation is particularly problematic given lower-than-average response rates from Latino voters, resulting in samples that are more educated and more suburban than the actual population.
Especially important is pollsters’ failure to account for a respondent’s generational distance from the immigration process, an issue that characterizes political identities more significantly for Latino individuals than other minority groups. A 2017 Pew poll indicated that the closer people were to the immigration process, the more likely they were to identify as Hispanic: 97 percent of foreign-born Latinos identified as such but only 77 percent of third-generation Latinos said the same. Heritage and specific place of origin are also key factors that diversify the perspectives of the Latino vote, as Mexican Americans, Cuban Americans, and Puerto Ricans all have distinct political leanings.
Just as pollsters determined after 2016 that white voters could not be substituted for one another without accounting for education, so too must survey groups learn that the variety of factors differentiating Latino individuals renders a simple classification by ethnicity nearly useless. One potential solution to this problem is the usage of larger samples of Latino voters so that polling companies can weigh by ethnicity and distance from the immigration process. These efforts should be directed towards states with particularly heterogeneous Latino communities, like Florida.
Polling groups could also experiment with weighting polls by other characteristics, such as respondents’ employment status. Such a strategy might have been able to foresee enormous shifts in support towards Republican candidates in 2020 along Texas’s southern border, where the percentage of people employed by the oil and gas industries is over 10 times the national average. Whether by decoupling voters from their ethnicity or by disaggregating voters within their ethnic identity, pollsters must pay attention to the factors influencing voters’ political preferences beyond the label “Latino.”
Another issue plaguing surveys of Latino voters is the lack of Latino diversity within the polling industry. This problem is twofold: First, few people are qualified to pose the right questions to respondents and, consequently, few organizations specialize in studying Latino preferences. Hiring callers from Latino backgrounds could decrease the community’s notably higher refusal rate for live-caller polls. Long term, improved hiring practices could also work to fend off voter suspicion. Public opinion polling companies should work to create positions dedicated to calibrating polls in a way that best reflects the actual perspectives of Latino voters, and campaigns ought to fund private Latino-owned polling groups to support the industry’s growth.
Recognizing the shortcomings of current polling strategies to survey Latino voters and adopting new practices to improve precision could alter the results of polls during election season. Policymakers can use more precise polling data to more accurately gauge public opinion and better deliver on promises to voters. From a media perspective, disaggregated polling samples can better inform how reporters frame Latino political preferences and more diverse polling leadership can provide useful expertise to journalists. Public opinion polling of Latino voters in the United States may be broken, but the potential benefits of adopting new strategies are well worth the cost of repairs.