Just FYI, here are our penultimate state and national numbers for the Presidential race for CHIP50, based on data collected from August 19th to October 26th. (Penultimate because we are actually still in the field; so our future pre-election numbers will change slightly.) This includes weights for demographics and prior vote (see our prior post on weighting). Note that because of the extended period the survey was in the field, it will not capture trends that occurred during this time; and most surveys show a small but potentially very consequential shift in the race towards Trump over the last month. We should expect, also, that these numbers will be somewhat less predictive than top professional pollsters, who will incorporate models of probability of voting, among other things, into the numbers that they produce.
A few quick observations: (1) the national numbers are very close to 2020 (a .3 point shift towards the Democrats); (2) the numbers are significantly more bullish regarding Harris than other surveys; (3) they suggest (with uncertainty) that the shifts that have occurred (e.g. more educated voters shifting towards the Democrats and less educated voters towards Republicans) have eliminated Trump’s electoral college advantage.
Once counting is complete, we will compare the final election numbers to our data, and evaluate the possible explanations for deviations that we find, and post those reflections here.
