Can I Beat the Experts?

Senate 56, House 225. I figured I’d do a prediction for the midterms. What’s the point of having elections if we don’t all go out on a limb guessing what will happen? The following are some thoughts I jotted down over the last few days to explain why I think Republicans will hold the House and gain seats in the Senate, interspersed with some interesting screencaps.

The first thing is that early voting looks very good. The normal trend is that Republicans win the absentee ballots (lots of military), Democrats win early voting, and then Republicans win on Election Day. It’s not hard to see why. Democrats haul a lot of people to the polls to get them to vote. They go to churches and colleges and arrange to walk everyone to the polls in a group, and hand them sample ballots showing whom to vote for (just talking about the legal methods here). That takes time, so they’re doing it all through the early voting period, giving them an edge there. Republicans are more independent and likely to get themselves to the polls. They’re also more traditional and inclined to do it the old-fashioned way, voting on Election Day with everyone else.

All that means that, in a close election, Republicans often start out with a lead based on absentee ballots, then lose the lead as early voting goes on, and then the election hinges on whether enough Republicans vote on Election Day to overcome the Democrat lead from early voting.

I should point out that when we’re talking about absentee votes and early votes, we don’t know for sure how they voted. We only know which party they’re registered as. So when you read that Republicans have 50,000 early votes in a race, that means 50,000 votes have come in from registered Republicans. It’s possible that some of them will be for the Democrat or a third party. It’s possible some will be empty or mis-marked and not count at all. But generally you can assume those will balance each other out; there will be mistakes from both sides, and roughly equal crossover from each side.

So we currently know (in states that report it) that Republicans are leading early voting compared to 2016. In Florida, for instance, they’ve maintained a 50,000 vote edge over 2016. In 2016, early voting ended with Democrats 100,000 votes ahead. This year, Democrats will be ahead, but by less than 50,000 votes.

So democrats are now counting on one or more of:

  • Lots of Republicans who normally vote on election day voted early this time
  • Lots of Democrats who normally vote early will show up on election day this time
  • Lots of Republican early voters crossed over to vote Democrat
  • Independents voted Democrat at record rates

There’s no evidence for the first three, and frankly they’re kind of silly. On the first two, people are creatures of habit, and there’s no reason to think hundreds of thousands of people would change their voting habit in one direction in two years. In fact, the one thing that could be affecting it works in favor of Republicans: closed polls in the R-heavy Panhandle due to the hurricane. Those people could show up heavy on Election Day. On #3, Trump’s approval rating with Republicans is over 90%, far higher than in 2016. There’s no reason for Republicans to hand the House to the Democrats and make things harder for him.

So they’re really left with #4, and the only evidence they have for that is their polls. So that’s what many races will come down to. Are the polls wrong (again) or are voters doing something tricksy and weird that’s throwing off the usual pattern of early voting? It might help to know that we’ve been here before. In 2016, their polls also showed higher Democrat output than the early vote reflected, so pollsters convinced themselves of #3 up there. It wasn’t true then, and there’s no reason except “but muh polls!” to assume it is this time.

Another factor is that there’s just no reason for voter upheaval. The economy is good. Jobs are up. Gas prices are down. The president is doing the things he said he would do, to the extent Congress, the bureaucracy, and Hawaiian judges are letting him. His approval rating is higher than Obama’s was at the same point in his presidency. His approval rating with Republicans is sky-high, driving their enthusiasm, added to anger over the Kavanaugh hoax. All those environmental factors lean R.

I see parallels with 1998, when Clinton gained five House seats and held steady in the Senate. That was another time when the economy was pretty good, we hadn’t gotten into any wars for a while, and Americans were tired of the other party’s endless investigations of the president, correct or not.

Another important and under-reported factor could be the 1982 Consent Decree. In that year, Republicans in New Jersey were charged with voting shenanigans. As part of the settlement, the DNC and RNC both agreed to give up their right to seek remedies for voter fraud in many cases. That’s why you’ll hear about obvious cases of voter fraud, as when a district ends up with more votes than its population, and nothing is ever done about it. They legally weren’t allowed to. Well, that decree finally expired last December after the judge who kept extending it died. This is the first election in over 30 years where the parties will be in trouble if they’re caught breaking the rules.

That has to be giving would-be ballot-box-stuffers and defrauders pause, especially with the president proclaiming that voter fraud will be prosecuted according to the law. This attempt by North Dakota Democrats to fool hunters into thinking they could lose their hunting licenses if they vote might be a good place to start.

Finally, this is worth a chuckle. A 2016 poll of “Hard” Trump and Clinton voters found that 55% of Hard Clinton voters admitted they could be prevented from voting by a serious storm, reports of a disease epidemic, threats of violence, or other fairly mundane threats. Only 10% of Hard Trump voters said the same. There was one thing that could keep 38% of Hard Trump voters from the polls, though: attack by extraterrestrials. Some things just can’t be kept waiting.