Polls vs. results in analyzing elections

According to the Google Genie, well-designed pre-election polling has a margin error of plus-or-minus 3 points. What can we learn from pre-election polling and post-election results from RealClearPolitics data which is independent of any political party affiliation?

Nationally, of all states combined, Biden did more poorly than the polls predicted: +7.2% down to 3.8%

What about the polls vs. election results in the 12 battleground swing states?

In two states, the election results matched the polls - NV (Dem +2.4 to +2.4), PA (Dem +1.2 to +1.2)

In one state only, Biden did better than the polls - MN (Dem +4.3 up to +7.2)

In three states, Biden's results were less than predicted - AZ (Dem +0.9 down to +0.3); MI (Dem +0.9 down to +0.3); WI (Dem +0.9, down to +0.3).

In four states, Trump did better than the polls predicted IA (Rep +2.0 up to +8.2); NC (Rep 0.2 up to+1.3); OH (Rep +1.0 up to +8.2); TX (Rep +1.3 up to +5.6).

Two states switched political parties from prediction to result: Florida switched from Dem +0.9 to Rep +3.3 and Georgia switched from Rep +1.0 to Dem +0.3.

Conclusion: Using Bayesian logic (Thomas Bayes, 18th century mathematician), the varied nature of the data with no consistent trend would indicate that it is more likely that there was no vote stealing by either side - that the election was free and fair. If the election had been marred by vote theft, then Bayesian reasoning would conclude that it is likely that the Republicans stole more votes than the Democrats.

In only two states did Biden do better than the polls. Trump did better than the polls in five states and by bigger gains. Based on favorability before the election: Trump 44%, Biden 51% closely matched vote totals of Trump 46.7%, Biden 51.3%.

David Thiessen

Woodstock