The Final Push

Estimated Popular Vote Totals from the October–November Wave of the RAND Presidential Election Panel Survey

by Michael Pollard, Joshua Mendelsohn

Voters stand near a voting sign before casting ballots during early voting

Voters stand near a voting sign before casting ballots during early voting

Photo by Shannon Stapleton/Reuters

With just seven days left before Election Day on November 8, the latest RAND Presidential Election Panel Survey (PEPS) (2,269 respondents between October 20 and November 1, 2016) indicates that the popular vote gives Democrat Hillary Clinton a significant lead of 9.1 points over Republican Donald Trump. Libertarian Gary Johnson is estimated to garner 7.7 percent of the popular vote, while others claim 13.1 percent.

Throughout the PEPS data collection, which began in December 2015, the popular vote estimate has favored a Democrat over a Republican, as seen in Table 1. Since March 2016, the Democrats' lead has been outside the margin of error — meaning that the lead is greater than what we would expect from sampling error, so we can be confident in the results.

Table 1. More Respondents Have Favored a Democrat Than a Republican in All PEPS Waves

PEPS Wave Any Democratic Candidate (Percentage of Popular Vote) Any Republican Candidate (Percentage of Popular Vote) Difference
December 2015 46.7 43.1 Democrat +3.6
March 2016 53.0 37.9 Democrat +15.1
August 2016 48.1 39.7 Democrat +8.4
September 2016 43.3 33.4 Democrat +9.9
October/November 2016 43.7 34.6 Democrat +9.1

NOTE: Beginning in the August 2016 wave (Wave 3), we asked respondents if they would vote specifically for Clinton or Trump because they had clinched their respective party’s nomination by then. The difference in all waves except December 2015 was statistically significant.

This trend continued in the October/November data, collected after October 19, when the final presidential debate was held. Clinton's 9.1-point lead in the popular vote is outside the margin of error (±1.9).

Recall from earlier reports in the PEPS series that unlike most other polls, we do not use a standard "likely voter" model to estimate popular vote.[1] Instead, the PEPS employs a probabilistic method in which we first ask respondents to rate how likely they are to vote at all (on a scale of 0 to 100) and then ask, "If you do vote in the election, what is the percent chance that you will vote for [Clinton/Trump/Johnson/Someone else]?" Combining these two pieces of information provides our popular vote estimate.

Figure 1. Who Respondents Think Will Win the Election

Figure 1. Who Respondents Think Will Win the Election

We also asked respondents to estimate the percent chance that each candidate would win the election ("What do you think is the percent chance that each of the candidates for president will win the election?"), and those results also indicate that voters anticipate that Clinton will win by a significant margin. The percentage of voters who think Trump will win has declined significantly since before the first debate: In September, 37.8 percent thought that he would win,[2] but now only 31.4 percent report this belief.

The final wave of PEPS data will be collected shortly after Election Day. The post-election survey data will enable us to compare respondents’ reports of how they actually voted with their reports of how they thought they would vote.

Notes

  • [1] See, for example, Joshua Mendelsohn and Michael Pollard, Americans Are United … in Dissatisfaction with Their Choices: Analysis of the American Electorate at the Start of the 2016 General Election, Santa Monica, Calif.: RAND Corporation, RR-1726-RC, 2016. As of October 4, 2016:
    http://www.rand.org/pubs/research_reports/RR1726.html
  • [2] Michael Pollard and Joshua Mendelsohn, Voter Opinions on the Candidates, the Issues, and the Parties: Results from the September RAND Presidential Election Panel Survey, Santa Monica, Calif.: RAND Corporation, RR-1758-RC, 2016b. As of October 31, 2016:
    http://www.rand.org/pubs/research_reports/RR1758.html
  • [3] See, for example, M. DeBell and J. A. Krosnick, Computing Weights for American National Election Survey Data, ANES Technical Report Series, No. nes012427. Ann Arbor, Mich., and Palo Alto, Calif.: American National Election Studies, 2009.
  • [4] For more on the PEPS methodology, see Michael Pollard and Joshua Mendelsohn, Methodology of the 2016 RAND Presidential Election Panel Survey (PEPS), Santa Monica, Calif.: RAND Corporation, RR-1460-RC/UCLA, 2016a. As of June 10, 2016:
    http://www.rand.org/pubs/research_reports/RR1460.html

Research conducted by

This RAND-initiated research was conducted by RAND Labor and Population.

This report is part of the RAND Corporation research report series. RAND reports present research findings and objective analysis that address the challenges facing the public and private sectors. All RAND reports undergo rigorous peer review to ensure high standards for research quality and objectivity.

The RAND Corporation is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.