An Election Postmortem: What We Got Right, and What We Got Wrong

What We Got Right

At Elections Daily, we’ve always been skeptical of “mirage elections” – races that look competitive in the polls, but are obviously going to fade down the line. This cycle’s races proved our skepticism justified; none of the races we listed as “Safe” were upsets, continuing our flawless record in that category.

Presidential

Only two other major election forecasters – Sabato’s Crystal Ball and CNalysis – predict every Presidential result. Our ratings, which projected Kamala Harris would win 292-246 – fell squarely in the middle.

While our ratings proved to not be as accurate as we liked, we did accurately predict Republican wins in Arizona and North Carolina. We’ll explain what went wrong with the remaining states later in the article.

House

As of the time of publication, overall control of the House has been called for the Republican Party, but a few races remain uncalled. We’re simply going to cover the highlights here:

• In Alaska, Nick Begich defeated incumbent Democrat Mary Pelota. We correctly predicted this race as Leans Republican, along with the two other Blue Dog co-chair races in Maine (Leans Democratic) and WA-03 (Leans Democratic).
• We were the only election forecaster to have Jim Costa’s CA-21 at anything other than Safe Democratic; we had it at Likely Democratic. As of the time of writing, Costa leads by a margin of 4.8%.
• We were the only election forecaster to have Josh Harder’s CA-09 at Leans Democratic. As of the time of writing, Harder leads by a margin of 3.8%.
• We correctly predicted that Mike Lawler (NY-17) would be the only 2022 “wave baby” to survive in New York.
• We were one of only two election forecasters (the other being Split Ticket) to predict Mariannette Miller-Meeks (R) would win in IA-01.
• Every race listed as “Safe” or “Likely” was called correctly.

Senate

In Maryland, former Republican Governor Larry Hogan put in a once-in-a-lifetime over performance over the top of the ticket. He’s trailing Democrat Angela Alsobrooks by a nine-point margin. This is the best performance in a Maryland Senate race for a Republican since 2006, when Michael Steele lost by a 10-point margin. However, Hogan faced an insurmountable electorate; we were right to list this as Safe Democratic.

On the flip side, Nebraska, independent Dan Osborn proved the polls were right – the race was somewhat close. Incumbent Republican Deb Fischer leads by only a 7.1% margin. Like in Maryland, this was a good over performance, but not enough to truly get close to winning. We had this at Likely Republican – we think this was the right call.

Relative to the national vote, we also think our calls in Arizona and Nevada’s Senate races weren’t bad. While both races turned out to be close, the lower quality of Republican candidates here turned winnable races into narrow loses. It would’ve taken a stronger national wave to flip them.

Gubernatorial

Our most accurate ratings came in the gubernatorial front, where we were the most accurate forecaster. We anticipated only one race – New Hampshire, which we rated as Leans Republican – would be competitive. Among election forecasters, we were the only outlet to have Indiana (which Mike Braun won by 13%) at Safe Republican. We never bought into the hype on this race, and for good reason – the Democratic candidate’s 41% performance here equalled her best public polling performance.

We also were one of only two forecasters (alongside CNalysis) to have North Carolina (which Josh Stein won by a nearly 15-point margin) and Washington (which Bob Ferguson leads by 12%) at Safe Democratic.

What We Got Wrong

Presidential

Let’s start at the presidential race. Out of the seven major swing states, we predicted two (Arizona and North Carolina) correctly while missing five (Georgia, Nevada, Michigan, Pennsylvania, and Wisconsin). Three of these (Michigan, Pennsylvania, and Wisconsin) can be explained by a broader polling error in the midwest; for Georgia and Nevada, we went with a gut feeling that proved to be incorrect.

It’s worth noting that all but one of of the predicted swing states did turn out to be somewhat close: the Midwestern trio were all within two points, while North Carolina, Georgia, and Nevada were within four. Only Arizona (Trump currently leads by 5.5%) proved to be uncompetitive, closer to New Jersey or Virginia than to the major swing states. Polls were broadly accurate at identifying which states would be competitive.

Senate

In the Senate races, we got two races wrong: Ohio and Pennsylvania. Ohio was a response to both broader polling trends in the midwest – which seemed good for Democrats – and to the Selzer poll, which proved to be utterly inaccurate. Pennsylvania was the shocker of the cycle, but only marginally. Polls had shown a tightening race, and we had it in the Leans Democratic column alongside Michigan and Wisconsin.

Two of the more interesting results happened in races we defined as “Likely”: both Arizona and Nevada proved to be closer than expected. However, Trump over performed Kari Lake and Sam Brown by wide enough margins that we think our Likely Democratic ratings here were still justified – especially in Arizona, where Kari Lake is running eight points behind Trump.

House

The House has not finished counting yet, as several races in California (specifically, CA-13, which we rated as Leans Democratic, and CA-45, which we rated as Leans Republican) remain uncalled. However, as of the time of writing, we were wrong in the following races:

• AZ-01 (Leans Democratic)
• CA-27 (Leans Republican)
• CA-47 (Leans Republican)
• CO-08 (Leans Democratic)
• IA-03 (Leans Democratic)
• NE-02 (Leans Democratic)
• PA-07 (Leans Democratic)
• PA-08 (Leans Democratic)
• PA-10 (Leans Democratic)

We had predicted Democrats would have a 219-216 majority; as it stands right now, the Republican majority will have between 220 and 222 members, meaning they out-performed our expectations by between 5-7 seats. The majority of our nine misses came in California (two) and Pennsylvania (three).