Expert Opinions: What constitutes a good vs. bad Latino poll

During a webinar and telephonic briefing discussion hosted by NCLR yesterday, experts on political polling and Latino voters analyzed how the national exit polls and many polls throughout the election cycle consistently miss the mark when it comes to Latino voters – and discussed why that is problematic for good post-election analysis and resultant political and policy decision-making. See below for summary details from each presenter and link to slide presentations from each panelist.

Gabriel Sanchez, Principal at Latino Decisions and Professor of Political Science at the University of New Mexico, highlighted two problems he consistently sees in Latino polling – polls that have too few Latinos in their sample, leading to a large margin of error; and polls that have the wrong Latinos in their sample, leading to Latino results that are not representative of the Latino voting community as a whole. In particular, many pre-election polls of Latinos rely on sample sizes of Latinos that are way too small; samples that are not drawn to be representative of all Latinos; and limited, or zero interviews in Spanish. Regarding exit polls in particular, Sanchez highlighted four things to look out for: which precincts were selected, and how? What share of Latino interviews are in Spanish? What are the Latino demographics? And what percent are election day voters vs. early voters? As Latino Decisions’ analysis has demonstrated, Latino voters most often “missing” from polls with problematic methodology are those most likely to be Democrats: such as, naturalized immigrants, Spanish-dominant, lower socioeconomic status, and younger Latinos. Click here for link to the slide presentation

Alan Abramowitz, the Alben W. Barkley Professor of Political Science at Emory University, examined consistent problems he encounters in Latino polling this election cycle, including: underestimates of the likely Latino share of the electorate; a lack of transparency about the methodology in the Latino-specific samples; and overestimates of Trump’s share of the vote, based on historical trends and in comparison with Latino polls featuring a greater sample size and higher-quality methodologies. As Professor Abramowitz explained, ensuring the accuracy of Latino polling matters in order to avoid giving a misleading impression of the way Latino voters are reacting to Trump’s extreme racist and anti-immigrant message. In addition, such errors can result in underestimating the overall Hillary Clinton margin in the nation or a key swing state. For example, the difference between a 30-point Clinton margin among Latinos and a 58-point Clinton margin among Latinos would be approximately 3 points of the national vote margin—the difference between a 2-3 point national margin and a 5-6 point national margin. Click here for link to the slide presentation

Michael Frias and Jonathan Robinson, Chief Client & Marketing Officer and Senior Research Analyst, respectively, at Catalist LLC, described their firm’s work to use voter file data to explore Latino voters’ behavior. The presentation included discussion of several problematic examples of Latino voter exit polling – such as the 2008 exit polls in Virginia, which implied that more Latinos voted than the Census Bureau said lived in the state, and the 2012 exit polls in Virginia, which estimated that Latino turnout fell by 40 percentage points compared to 2008. The presenters, who also discussed Catalist’s efforts to marry high quality Latino polling data with information from voter files, predicted that Latinos are poised to make a big difference in the upcoming elections. For example, Catalist analysis shows that first time voter registrations among Latinos is higher in percentage terms than they have seen in either 2008 or 2012. Click here for link to the slide presentation

Clarissa Martinez de Castro, Deputy Vice President at NCLR and the host and moderator of the discussion, noted that while polling of the Latino community has become more robust in the last decade, challenges persist. Sampling challenges in national polls, and specifically in exit polls, can provide an inaccurate picture of how Latinos vote and why. Accurate polling matters because the data and analyses derived from it helps shape political outreach strategies, as well as the policy agendas and priorities of parties and elected officials. She highlighted 2004 exit poll discrepancies as one of the elements leading to the development of Latino election-eve polling of highly likely Latino voters in 2006, which has grown into a large, multi-state poll providing more accurate and detailed information about how and why Latinos vote.  In the last three election cycles, these polls have been sponsored by multiple partners invested in achieving greater accuracy, and have been conducted by Latino Decisions. Click here for link to the slide presentation


, , ,