- Bearing Drift - https://bearingdrift.com -

Introducing the McGuire Virginia Turnout Model

As we careen towards Election Day, there has been an understandable focus on early voting in Virginia and what it means. As we don’t have party identification for voters, there’s only so much we can project. We can, however, look at the early voter data to project how many Virginians will vote. That is the humble goal of the McGuire Virginia Turnout Model.

As this is my first projection (it will be updated daily until early voting closes), I’ll provide the methodology and assumptions in this post. Future posts will simply provide relevant updates to the numbers.

So how am I doing this? Well, I’ll first look at early voting totals for 2021, 2020, 2017, and 2016 (courtesy of VPAP’s Early Voting Dashboard [1]), plus total turnout from 2020, 2017, and 2016 (courtesy of Virginia’s Department of Elections [2]). That data is as follows:

The first thing I did was project the remaining early vote for this year based upon the last seven days of 2020. This is done two ways. First, using the last week’s share of the early vote and projecting the same share for this year leads to an additional 235,893 early votes projected. Second, using the last week’s share of the early vote and adjusting it by the ratio of 2017 early votes to 2016 early votes gets us a projection of 267,066 additional early votes. Adding this to the current total gives a projected early vote range of 846,447 to 877,620.

From there, I look at the change in early vote share from 2016 to 2017 (falls from 14.4% to 7.4%) and assume the same drop-off this year (from 62.8% in 2020 to 32.6% in 2021). With that assumption, the early vote totals and projections lead to an estimated turnout between 2,598,196 and 2,6913,882.

Keep in mind, this does not predict who will turn out.

Also, as noted above, these projections will change as new data comes in – and those changes will be posted on BD. Here’s what that will look like (using the current projection as stated above).

Of note: we can expect a large amount of early voting tomorrow, the first major Sunday early voting for Virginia. While Sunday is included in the 2020 data, there wasn’t much in 2020 so these numbers could change substantially over the next couple days.

Of course, I’ll see how well the model did once the election is over, and adjust methodology as needed for next year.