November is 6 months away. At this point, no one can confidently predict exactly what is going to happen on election day. Luckily, we have polls, elections, and special elections to guide us. However, these three statistics are vastly different, and each one paints a different picture of how big, if at all, the “Blue Wave” will be. Let’s look at each piece of data.
First off, generic ballots polls, as of April 22, give Democrats about a 7 point lead over Republicans. This is good for Democrats, but not nearly as good as other figures. An advantage of 7 points would make the House tilt Democrat and would likely keep the Senate in Republican hands. However, these polls will, most likely, trend further blue between now and November. This is due to 2 factors:
- Congress hasn’t done anything substantial recently, but they have done something. Major votes by Congress tend to sway people’s opinions of the two parties, often punishing the party in power and favoring the party without. The reverse of this is that, when nothing gets done in Congress, people tend to blame the people in power. Basically, if Congress does something major, people will polarize and hyper-partisanship can occur, favoring the Democrats based on the current makeup of the country; if Congress goes a long time without doing something, whether it be because of gridlock or if they just have nothing to do, people will turn against them and want change. This was evident at the Presidential level in 2016 when the entire Rust Belt flipped. The Government wasn’t doing anything and the people wanted a change in leadership.
- Polling firms change who they poll. At this point in time, most polling firms are polling registered voters, meaning anyone who is able to vote in an upcoming election. As November draws closer, firms tend to switch to a likely voter model, meaning they poll people who not only can vote, but are likely to vote. This should change the current generic ballot average in favor of the Democrats due to the fact that Democrats are energized and more likely to vote in 2018. The midterms are seen as a referendum on the President and his party, and voters are historically more likely to turn out against something they don’t like, if they think their party has the advantage, than turn out for something they like, if they think their party is going to lose the election.
Second, Trump approval polls give Trump a net -14% approval margin. This data is useful because, as I stated earlier, midterms are seen as a referendum on the President. A net -14 margin means that 14% more people disapprove of Trump than approve of him. If this translated into a 14 point advantage for Democrats, which it won’t, then both houses of Congress would be getting a new leader. This won’t happen, though, as a significant amount of Trump disapprovers are Republicans and will probably vote Republican. Also, a Trump -14 approval now doesn’t mean a Trump -14 approval in November. Approval ratings can and do change. Trump could gain more support if he continues negotiations with North Korea and keeps quiet on Twitter. That said, he could also lose the support he has now if farmers are negatively affected by a trade war, or if more information about Russian collusion and/or Stormy Daniels is released. For now, we can assume a slight reversion to the mean and project Trump to have around a -12% approval margin.
Third, we have special elections. Unlike polls, special elections involve real people casting real votes in real elections. They are, to an extent, more accurate in capturing the partisan mood of a group of people, especially when they have higher turnout. Higher turnout equals more people which equals larger sample size which ultimately equals a more accurate prediction. When looking at special elections, pundits don’t look at individual results, but rather the average swing over many elections. If one were to look simply at the result of an election, they would see that, for example, the Montana House seat special election stayed in Republican’s hands despite heavy campaigning on the left, implying that Republicans were favored. However, if they were to look at the vote swing in that election and others, they would see something completely different. Special election since 2016 have swung roughly 15 points towards the Democrats, on average. Averaging swings is difficult because some special elections (GA-06) have extremely high turnout, while others (SC-05) have low turnout. It’s safe to say that the average swing towards Democrats is 15%, give or take 5%. The other thing that makes averaging difficult is the strength of the individual candidate running. For example, the Alabama Senate election had a huge swing towards Democrats, giving them the seat. Taking this swing for face value portrays that Democrats should be able to win in states like Texas and Nebraska. However, if you look at the candidates, you’ll see that,
- Roy Moore was extremely flawed and almost certainly lost because of the sexual assault and child molestation allegations. Even members of his own party refused to support him. Jeff Flake (R-AZ) went so far as to donate money to Doug Jones, who ultimately won.
- Doug Jones was a conservative Democrat who appealed to both the Democratic base and moderate Republicans. While liberal on some issues, Jones is a moderate and gained support from both parties. Since elected, Jones has voted with Trump 58.3% of the time, making him the second most conservative Democrat in the Senate (Joe Manchin is the most conservative).
To reiterate, special election swing is about 15%, but could fall anywhere between 10% and 20% depending on how you average.
Fourth, we have the normal, scheduled elections that took place in November of 2017. These elections included but were not limited to the Virginia Governor election, the New Jersey Governor election, and the Virginia state legislature elections. In New Jersey, Democrats flipped the seat from a Republican Governor to a Democratic Governor. In Virginia, Democrats easily beat their Republican opponent and kept the seat blue. Democrats were expected to win both, but not by as large a margin as they did, particularly in Virginia. Lastly, the Virginia state legislature unexpectedly swung by large margins towards the Democrats, making the chamber favor Republicans by a slim 51-49 margin, much less than their previous 66-34 margin. This was largely unexpected as nearly no one thought the chamber was even in play. Alas, the control came down to a single seat, which had to be determined by a draw of the hat. The Republican candidate was pulled, meaning the chamber stayed Republican by a slim margin.
All of these data points are different, and they all have positives and negatives to go with them. Figuring the Democrat’s chances in the midterms is hard to do, and can change based on how you weight the data, and what else you consider. In my opinion, which is based on my personal belief on how to weight this data, I think special elections and elections should be weighted the most, generic ballot next, and Trump approval last. Doing this would give you about a 10-12 point advantage for Democrats. This would, most likely, give the House to Democrats and make the Senate a tossup.
Tomorrow, there will be a special election in the state of Arizona. The swing in that election will help us to figure the midterm probabilities, but is still only a point in a set of data. I’ll be updating my Senate and Gubernatorial predictions and will post them by the end of the month.