With everything surrounding the Brett Kavanaugh “Salem Witch Trials” and more and more Republican held seats “coming online” (parlance for competitive) the electoral landscape for the majority party is looking decidedly dark. Texas and West Virginia, of all places, are featuring competitive races (TX Senate and WV-3)!
All this is driven by two things, the conventional narrative based on first term President’s midterm electoral history and the polls. But, in an era where cell phones are becoming more prevalent, landlines less so, and more Americans becoming disgusted with the political process it brings into question the credibility or at least ability of polling to be accurate.
Polling has arguably already suffered from these phenomenons. For example, specifically referencing midterms here, in 2010 RCP’s generic ballot average had Republicans winning nationally by 9.4 points. They won by 6.8 points. In individual races the polls were wildly off most notably in the Nevada Senate race. So, the polls were tilted towards Democrats.
In 2014, the reverse happened. Polling showed tight races across the country in key Senate and gubernatorial races and the House was neck and neck. Yet, in the end, Republicans overperformed the average Senate poll by 4 points (in dark blue states no less) and the generic ballot by over three points.
In 2012 and 2016 the polls were off yet in different ways. In 2012, state polls showed Obama with small but consistent leads in key swing states while national polls showed the race a dead heat. Obama eventually won by almost four points nationally and the state polls were pretty spot on (to a point or two on average).
The see-saw 2016 election was the opposite. The national polls were pretty close to accurate (within a point or two on average) but the state polls were widely off. The most striking example would be Wisconsin (with Pennsylvania and Michigan close seconds) where not a single poll after the primary showed Trump ahead. Yet, Trump won all three states (albeit by narrow margins).
In essence, the polls were biased toward copying one another. There are a variety of reasons why this could be the case; faulty turnout assumptions, certain voters making it through filters, wrong weighting. The list goes on. But if this happens it can have cascading consequences.
Perhaps the most notable example came from the Virginia Senate contest in 2014. Most polling had Senator Mark Warner ahead by 8-10 points (or more), but on the eve of the election a pollster, Norfolk University, found the race a dead heat. Due to the plethora of polling showing the race a blowout the pollster did not release the poll. Warner won by less than a point and trailed for most of the night until NoVA started reporting.
In this the danger of a self-fulfilling prophecy can be seen. Polls are run, created and analyzed by human beings who can let their own biases or worries about credibility influence their thinking. As a result, polls countering what 10 other polls say or the conventional narrative may never see the light of day.
In the months after the 2016 election, the polling industry (and media) reflected on what they could have done differently. A lot of hand wringing and discussion later, not much has changed. Oh sure, there was some discussion about focusing more on anecdotal, narrative driven evidence, and the media and pollsters not attempting to solicit the opinions of rural voters but for the most part it is business as usual.
All which leads to the possibility we could see something similar in 2018. If the polls are off by a couple points systematically (which usually is the case if they miss) it probably is good news for Republicans. Since Democrats enjoy such a wide lead it is more likely the polls are biased towards them (just as the polls were biased to Republicans in 2010).
Where this could benefit Republicans the most is fairly obvious. Close races in Missouri and North Dakota would go their way. Montana and Indiana would be dead-heats. Florida would be a toss-up and because West Virginia is so rural it is hard not seeing it being close.
That being said, the polls could still actually be biased toward Republicans and Democrats could do even better. I doubt that and here is why. Right now, electoral polls say Republicans are in a world of hurt. But, the polling measuring fundamentals; Presidential job approval, partisan enthusiasm, views on the economy, all point to a dead heat. For example, despite Brett Kavanaugh’s Supreme Court fiasco the President’s approval is going up after initially dipping.
Or take a new Pew survey. The survey finds both parties bases are galvanized with Democrats more-so (67% to 59%). It is not surprising Pew, which uses only a registered voter filter finds Democrats massively ahead on the generic ballot (52%-42%). This is not say Republicans won’t suffer in a couple of months but rather perhaps not as badly as initially thought if their base is actually activating.
This might be thin hope for Republicans to stand on but it is a hope nontheless. If Republicans can activate their base and polls are missing this it is possible they can hold the House and easily hold the Senate. We’ll see.