US 2020 Presidential Election Predictions

This is the only site with all 2020 US Election Predictions

Polling Data Update, Oct 28, 2020

Polling data are wrong. They missed in 2016, with polls biased to Democrats. They missed in 2012, with polls biased to Republicans. Please say this 3 times, so you get it out of your system, and can read further. When they miss for a swing state, say Michigan, people take notice.

Conventional wisdom says polling data ought to be better days before the election. In fact, as many people have already voted, it is no longer an issue of who they are likely to vote for, but who they just voted for, either by mail in ballots or in person early voting.

Everyone is focused on the swing states - states or districts too close to call. Polls always come with margin of errors, usually 3 to 4%, due to their sample size. Reliable polls show margins to be twice this error, at least 6% to 8% difference to stay out of the polling noise range. Less than 3%, it can be more noise than anything else. It's called NMR - no meaningful results. .

The analysis from the New York Times is very interesting, because they ask, "What if the polls have the same errors as in 2016 or 2012?". See link.

Correcting polls to 2016 is bad for Biden. States that flipped from blue to red are WI (10), IA(6), FL(29), NC(15), GA(16) and ME2(1). With the 2016 correction, the race becomes essentially tied or slighly in favor of Trump. Biden is doomed if he loses Wisconsin and Pennsylvania, both is the polling noise range with the correction.

Note, correcting the polls to 2012 is good for Biden. I can't say which way it goes.

So, one more maxim, polls necessarily get any better days before election day. Don't trust any poll showing margins of 3% or less. It can be just polling noise.

David Lord