Overall assessment for polling through October 31, 2024

Harris was likely to have won Pennsylvania, Michigan and Wisconsin and, speculatively, Nevada if the model corresponded to the election results. This would have resulted in a 270-268 win, assuming no changes from the 2020 election in other states and electoral districts. Trump would have won the remaining states.

The number of "likely voter" polls increased substantially in the second half of September. As a result, the model was revised to excluded "registered voter" surveys and to include all likely voter polls from August to late September. As before, the beginning assumptions include a close contest in the range of Harris taking between 48% and 52% of vote in each swing state. This is a "strong" prior and pronounced polling results will be required to move the credible interval outside that range. The model is based on the Bayesian analysis described in Methodology.

More new polling than was anticipated before the election prompted this final update.

Final result

Harris lost all of the swing states.

Evaluation

States where model and results coincided

Arizona

In Arizona 21 pollsters conducted a total of 37 polls from Jul 30 to Oct 31. Overall, the weighted mean poll of 49.17% indicated a victory for Trump over Harris in the two-candidate race. Weights were assigned by number of respondents. Of the pollsters, 17 corresponded to a victory for Trump with 50.83% of the two-candidate vote. The other 4 corresponded to a victory for Harris in the two-candidate vote, overstating her actual share, which was 47.22%.

Georgia

In Georgia 17 pollsters conducted a total of 31 polls from Jul 30 to Oct 30. Overall, the weighted mean poll of 49.10% indicated a victory for Trump over Harris in the two-candidate race. Weights were assigned by number of respondents. Of the pollsters, 13 corresponded to a victory for Trump with 50.90% of the two-candidate vote. The other 4 corresponded to a victory for Harris in the two-candidate vote, overstating her actual share, which was 48.89%.

Nevada

In Nevada 16 pollsters conducted a total of 26 polls from Aug 2 to Oct 31. Overall, the weighted mean poll of 49.75% indicated a victory for Trump over Harris in the two-candidate race. Weights were assigned by number of respondents. Of the pollsters, 8 corresponded to a victory for Trump with 50.25% of the two-candidate vote. The other 8 corresponded to a victory for Harris in the two-candidate vote, overstating her actual share, which was 48.42%.

North Carolina

In North Carolina 26 pollsters conducted a total of 44 polls from Aug 2 to Oct 30. Overall, the weighted mean poll of 49.48% indicated a victory for Trump over Harris in the two-candidate race. Weights were assigned by number of respondents. Of the pollsters, 17 corresponded to a victory for Trump with 50.52% of the two-candidate vote. The other 9 corresponded to a victory for Harris in the two-candidate vote, overstating her actual share, which was 48.28%.

States where the model did not coincide with the results

Pennsylvania

In Pennsylvania 25 pollsters conducted a total of 51 polls from Jul 30 to Oct 30. Overall, the weighted mean poll of 50.02% indicated a victory for Harris over Trump in the two-candidate race. Weights were assigned by number of respondents. Of the pollsters, 11 corresponded to a victory for Trump. The other 14 corresponded to a victory for Trump in the two-candidate vote, but overstating the actual share for Harris., 48.42%.

Michigan

In Michigan 21 pollsters conducted a total of 37 polls from Aug 2 to Oct 30. Overall, the weighted mean poll of 50.42% indicated a victory for Harris over Trump in the two-candidate race. Weights were assigned by number of respondents. Of the pollsters, 5 corresponded to a victory for Trump with 49.58% of the two-candidate vote going to Harris//t//. The other 16 corresponded to a victory for Harris in the two-candidate vote, overstating her actual share, 49.29%.

Wisconsin

In Wisconsin 17 pollsters conducted a total of 35 polls from Aug 1 to Oct 31. Overall, the weighted mean poll of 50.34% indicated a victory for Harris over Trump in the two-candidate race. Weights were assigned by number of respondents. Of the pollsters, 4 had a weighted mean average that corresponded to a victory for Trump with only 49.66% of the two-candidate vote going to Harris. The other 13 corresponded to a victory for Harris with 50.88% in the two-candidate vote, overstating her actual share, which was only 49.56%.