Biden Has Major Election Edge According to Voter Group Analysis

Most polls have consistently shown Joe Biden with a lead ranging from 6-9% over the past few months. As of August 31, the Real Clear Politics Poll Average has Biden up by 6.9% and the 538 Polling Polling Average has Biden up 8.0%. But after Donald Trump’s surprise victory in 2016, there is widespread skepticism about the accuracy of polling. Many people seem to believe that the polls underestimate Trump’s support.

Most pollsters weight their polls according to demographic factors, trying to match their sample to the expected makeup of the electorate in categories such as age, gender, education level, income, urbanicity, etc. This is necessary because the actual respondents to a poll don’t always match the expected makeup of the electorate in these factors, and we know demographics have some predictive value. But one source of polling error is the inability to perfectly forecast the actual demographic breakdowns of the upcoming election. When the pollsters’ conventional wisdom proves wrong, wide scale correlated polling errors can occur.

That’s why I was excited to see a recent poll put out by the USC Dornsife Center for Economic and Social Research. On top of all the general demographic data, this poll asked people if they’d voted in 2016 and who they voted for. This extra layer of information makes it possible to break down the electorate into Voter Groups. And analyzing these voter groups makes predicting the outcome in November much more reliable.

First, let’s take a look at who the Voter Groups are:

By far he largest group of voters is what I’ll call Partisans. These are people who voted in 2016 and plan to vote for the same party’s candidate in 2020. This group will likely include around 70% of voters in 2020.

The next largest group will almost certainly be New Voters. These are people who didn’t vote in 2016. Last election about 15% of the voters were new. Given strong interest in the election, this year could be even higher.

The next three groups are all pretty small. The Switchers are people who voted for a Democrat last time and plan to vote for a Republican this time, or vice versa. The USC poll indicates that we should not expect more than 5% of the votes in this election to come from Switchers.

The Protesters are what I call people who are so unhappy with both major party candidates they vote for a third party candidate. In 2016, third party candidates scored 5.7% of the vote. This year looks to be similar.

The Former Protesters are people who voted for a third party candidate last time around but this time will vote for either Biden or Trump. That will probably be the last 3-4% of voters.

The beauty of grouping the electorate in this way is the ability to make forecasts on a group by group basis, building up to an overall forecast that ought to be far more reliable than a standard poll-based forecast.

The next step is to split each group between the two major candidates as well as third parties.

The Partisans. Joe Biden has small but significant advantage among partisans. Looking only at 2016 voters, 44.3% of them are Biden partisans and 40.6% of them are Trump partisans. Although this advantage is small, it looks to be very stable. People who voted Democratic in 2016 and say they intend to vote Democratic in 2020 are not likely, in today’s ultra-partisan atmosphere, to change their minds. The same goes for Republicans. It seems safe to say that there is not a whole lot of potential for a significant swing within this group, regardless of the news events between now and the election.

The New Voters. According to the USC poll, new voters favor Biden 52% to 32%. This lopsided advantage tracks with demographic factors. Most new voters are people who were not eligible to vote in the prior election. That group is made up of people between the ages of 18 and 22 and immigrants. Both of those groups vote heavily democratic. Some new voters will be people who could have voted in 2016 but chose not to. This group is less predictable. A heavy republican turnout advantage would even this category out. But given that the republicans had a turnout advantage in 2016, stretching that advantage even further seems unlikely.

The Switchers. Because of the intense partisan divide in America today, there just aren’t very many switchers. So while winning this group is very valuable (you both add a vote for your side and subtract one from the other side), there’s not many votes to be had. Biden has about a 2-1 advantage among people who plan to switch. This is without question the most unstable portion of the race. Winning those intended switchers back gives Trump has best chance of closing the gap. But only 3.5% of Biden’s ultimate voters look like they’ll be switchers. If Trump could pull even among the switchers, he’d reduce Biden’s vote by 1% and increase his by 1%, gaining a net of 2% in the process.

The Protesters.  This group doesn’t affect the race much. If they all decided to vote, and they all moved in one direction, they could swing the election, but there’s no reason to think that would happen. These are the people who are the most disenchanted with the state of the government and the country.

The Former Protesters. This group is similar to the New Voters, in that they didn’t vote for a major party candidate last time, although they were all eligible. Because they were eligible, they don’t have the same demographic factors that give Biden an advantage. Still, they claim to favor Biden by 58% to 42%.

Totaling all these groups produces the following outcome: