AAPOR Releases Report Evaluating 2020 Election Polling
The American Association for Public Opinion Research (AAPOR) has released a report analyzing the pre-election polls from the 2020 elections titled “Task Force on 2020 Pre-Election Polling: An Evaluation of the 2020 General Election Polls.” The task force consisted of 19 members chosen to ensure diversity of backgrounds and opinions from various organizations, media outlets, and academic institutions. One of the main points covered in the report relates to polling errors that may have stemmed from issues of noncoverage, nonresponse, or statistical adjustments. The report notes that small errors can add up and have large effects on the predictions of winners. The task force found that in nearly every contest, there was an overstatement of the Democratic-Republican margin in favor of the Democratic candidate.
The AAPOR task force was unable to conclude why the 2020 polling error was so large and widespread; however, they were able to draw a few conclusions from their investigation. According to the report, these errors were not primarily caused by late-deciding voters, failure to weigh by education, incorrect assumptions about the composition of the electorate, reluctance to express support for Trump, incorrect estimates on voter turnout, or too few early voters or Election Day voters. The report states, “it seems plausible that many issues were caused by nonresponse. Nevertheless, it is so far impossible to know the primary issue. Among the possibilities are: too many Democrats and too few Republicans responding to the polls (between-party nonresponse); the Democrats/Republicans who responded had different opinions than those who did not (within-party nonresponse); and new voters and independents unpredictable in terms of both size (too many or too few) and representativeness (i.e., were the new voters who responded similar to those who did not?).”
While the report notes that similar issues with polling errors could occur in future elections, the 2020 election was unique due to its occurrence during a global pandemic and the significant increase in voter turnout. Ultimately, the report concludes it is not possible to predict what will happen in the polls for the next few years but that informing people on the limitations of pre-election polling can aid in decreasing distrust in the polling system from Americans.
This article was contributed by COSSA’s summer intern, Lillian Chmielewska of the University of Wisconsin, Madison.