The 2020 Senate Elections: How did I do?

On Wednesday, I took a look at back at my predictions for the 2020 U.S. Presidential election to see how I did. I successfully predicted every electoral vote except for the 29 in Florida and one in Nebraska’s second congressional district. But as important as this year’s Presidential election was – after all, everyone said it was The Most Important Election Of Our Lifetimes(TM) – it was not the only election this month, nor was it the only election I predicted.

I also made predictions for the 35 open seats in the 2020 U.S. Senate elections, and the 435 open seats in the 2020 U.S. House elections (including more detailed predictions in part 2 and part 3). Today, I’ll evaluate my predictions for the Senate. Up next, the House.

How did I do in predicting the results of those elections?

What We Still Don’t Know

The answer is, “we don’t know for sure yet,” because there are a few elections whose results are not yet known. Both of Georgia’s senate elections finished with no candidate claiming more than 50 percent of the votes, meaning they are headed to a runoff on Tuesday, January 5th, 2021. One runoff election will match Republican David Perdue (49.7% election day vote share) and Democratic candidate Jon Ossoff (47.9%); the other will match the Democratic winner Raphael Warnock (32.9% election day vote share) against the higher vote-getter of the two major Republicans, Kelly Loffler (25.9%). Republicans will be favored to win both elections.

So there are those two uncertainties, but that still leaves 33 certainties. How did I do in my predictions?

Hits and Misses

My top line result was that the Senate would be split evenly, 50 seats to 50, and that the Democratic Party would win tiebreakers thanks to Vice President Kamala Harris. That prediction could still come true, if both runoff elections in Georgia go Democratic. That is definitely possible, and as a Democratic not-in-Georgia Party Member I will do what I can to make it happen, but it’s unlikely. I think it’s more likely that the final Senator count will be Republican 52, Democratic 48.

The map below compares my predicted Senate results to the actual results as we know them now. Each hexagon represents a senator. Their colors show the party affiliation of the election winner; red is republican and blue is democratic. The pink, cyan, and lavender hexagons are incumbents (Republican, Democratic, and Independent respectively) who were not up for re-election this year. The as-yet-undecided races in Georgia are dark purple. Races that I mis-predicted are shown in BOLD ALL-CAPS.

Checking my predictions for the 2020 U.S. Senate elections. Click for a larger version.

Not counting my as-yet-unconfirmed Georgia predictions, I missed on two, both Republican incumbents. In North Carolina, Thom Tillis (R-NC) beat lawyer Cal Cunningham; in Maine, Susan Collins (R-ME) beat state house speaker Sara Gideon.

Margins of Victory

What about my predictions for which races would be close? Not including Georgia, I predicted that eight races would be decided by five percent vote share or less, and that two would be decided by one percent or less. In reality, none of the elections were that close: none were decided by one percent or less, and only three were decided by five percent or less. Although was wrong about Tillis’s win in North Carolina, I did predict the election would be within five percent (the final margin was 1.8 percent). I was also right about Mark Kelly in Arizona (2.4 percent) and Gary Peters in Michigan (1.7 percent).

But I missed badly on some other predictions of close elections. I had predicted a less than 1 percent margins for Gideon in Maine; in reality, Collins won by 8.9 percent. I predicted a Joni Ernst win in Iowa by less than 1 percent; the actual margin was 7.9 percent. And I missed on five of the five-percent races as well: John Hickenlooper won by 9.3 percent in Colorado, Steve Daines by 10 percent in Montana, Lindsey Graham by 10.3 in South Carolina, Roger Marshall by 11.9 in Kansas, and Tommy Tuberville by 20.4 in Alabama.

The lesson here? Predicting a close election is good for pageviews, but they can’t all be close.

The bigger lesson? Revisiting predictions to see what has come true and what has not is a powerful way of learning how to do better and better. It’s so easy, I wish more people were willing to do it.

Coming up next: my predictions for the House of Representatives.

Leave a Reply