When pundits talk about the 2018 midterms, two dominant storylines often emerge: the Democratic gains in the House of Representatives, and the disappointment the party felt at being unable to capture the Senate. The losses in the Senate stung particularly hard for the Democratic Party. The GOP expanded their majority in the chamber despite Trump occupying the White House, a historical anomaly in American politics. But why? Most analysts have pointed to the deeply divisive and highly publicized nomination process of Brett Kavanaugh, which became a political flash point when allegations of sexual misconduct him were made public on September 16th (he was confirmed on October 6th). But, despite the conventional wisdom, virtually no piece of available data points to this being the case.
In order to dismantle the conventional wisdom, it first needs to be established what that wisdom is. Surrounding Kavanaugh, the two broad claims that the piece seeks to address are: First, that Kavanaugh hurt Democrats overall. Second, that Kavanaugh hurt Democrats in the Senate races in particular. Both are wrong, but both are worth addressing separately. To start, here’s a look at the national environment to see if the hearings had an impact.
Did the Kavanaugh Hearings Hurt Democrats by Moving Persuadable Voters Against Them?
This is the first and easiest question to answer. If the nomination process hurt Democrats because it looked bad with swing voters, there would be a substantial drop in the polls for Democrats around the time the hearing occurred. This is largely not the case, and any drop was brief and quickly recovered. In 538’s generic ballot average, Democrats were up 9.1% the day the allegations broke. Democrats ended the campaign with an 8.7% point lead. In the Real Clear Politics average, the allegations broke when Democrats had an 8.3% lead, they ended the campaign with a 7.3% one. For both, averages, it is worth noting a small dip right before the actual confirmation vote. Two days before, on October 4th, Democrat’s lead according to 538 had dropped to 7.7%, and around the same time it dropped to 6.6% on the Real Clear Politics average. But both of these drops are not nearly large enough to be considered conclusive evidence, not even mentioning the fact that Democrat’s polling average bounced back up before Election Day. On top of that, Democrats maintained a smaller lead for almost the entire summer, so it is hard to earnestly conclude that a small dip from nearly their highest lead of the cycle was due to voters consciously punishing them.
Did the Hearings hurt Dems by Motivating Unlikely to vote Conservatives?
We can show fairly convincingly that the nomination process did not move the polls against Democrats in a significant way. But, perhaps the process made conservatives feel attacked, and motivated low propensity voters to go to the polls. If this were the case, we could pollsters having a difficult time accounting for these voters. However, there is no evidence that this occurred. Democrats performed almost exactly in line with the generic ballot average on Election Day.
Did the Hearings Polarize the Environment Around the 2016 Result?
The 2018 midterm results were highly correlated with the 2016 presidential election results. While many expected the environment to be polarized, the degree to which it was surprised many observers. Was Kavanaugh partly responsible for this? If he was, then it could be said that he hurt Dems in the Senate, where the battleground was primarily in red states. However, there is no evidence for this. The 2018 midterms were unusual in the sense that observers were lucky enough to get a stream of quality battle ground district polls, from the New York Times. These 96 polls, completed in real time, can give clues as to how the national environment changed over the last two months of the campaign. Below is a scatterplot of every live poll the New York Times completed. The X-axis represents the date the poll was finished, and the Y-axis represents the absolute difference in the poll’s result vs. the 2016 presidential margin in that district.
As can be seen, the polls barely got more reflective of 2016 overtime, and their results did not correlate with 2016 moreso during or after the nomination process than they did before.
Did the Kavanaugh Hurt Democratic Senate Candidates?
Now, the largest and most obvious question must be addressed: did the hearings make Democrats perform worse in the Senate races than they otherwise would have? Unfortunately, time travel is not a reality yet, and some key data points that would have made good comparisons are missing. North Dakota, West Virginia, Missouri, Montana, and Indiana all elect their governors in presidential years, for example. Despite this, there are still useful comparisons to be made. The first thing that must be done is look at polls of key Senate races in red states to see if Democratic candidates lost their crossover appeal. Below is a graph of every Senate poll that was completed in the states of West Virginia, North Dakota, Tennessee, Montana, Missouri, and Indiana between September 6th and Election Day.
As is the case with the House polls, there is virtually no trend towards a more polarized environment in the polls of Senate battlegrounds in red states. If the conventional wisdom is correct there would be a pattern of Democratic Senate candidates being unable to outperform 2016. According to available polling, it cannot be proved that Democratic Senate candidates lost any of their crossover appeal as a result of the hearings.
Another claim that needs to be addressed is the idea that Kavanaugh gave voters a reason to keep the GOP in power in the Senate, while Dems made a convincing case for voters to put them in power in the House. If this hypothesis were true, in aggregate, Democrats in the Senate would have performed worse nationwide than their house counterparts. However, the opposite is true. Demcoratic Senate candidates outperformed Hillary Clinton’s margin by an average of 10.8% in their respective states, while House candidates only managed an average 5.25% over-performance.
Why Does the Narrative Persist?
With it being clear that none of the available statistical evidence can back up the Kavanaugh effect, it is worth asking why it became conventional wisdom so quickly. There are multiple reasons, but perhaps the biggest is the historical anomaly of the results. Four Democratic incumbents lost their seats. For a year with a Republican in the White House, this is highly unusual. Observers would be forgiven for thinking this was an indictment of Democratic Senate candidates. But it’s not. Democratic losses occurred because the playing field was in predominantly red states. Looking at simply the number of incumbent losses is a mistake, because it implicitly assumes that the electorate operates as an omnipotent hive mind that rewards parties that do the “right” thing with wins and punishes ones that do the “wrong” thing with losses. Democrats running for Senate had more crossover appeal than their counterparts in the House, but they simply started in too big a hole.
The second reason this narrative persists is because of the media circus that surrounded the hearings. It’s not surprising that many of the same outlets and pundits that frantically covered the hearing would believe that it had an impact on the election only a month after its conclusion. It also makes for easy, explainable analysis. It reduces results down to a single vote, a single story. It makes politics more exciting than the likely simple reality that the modern American electorate is highly polarized and inelastic, with factors like incumbency and voting record mattering less than ever.
But the numbers for Kavanaugh just aren’t there. The polls over the last two months of the campaign, both for the House and Senate, do not show the country becoming more polarized as a result of the hearings. The actual results do not show any evidence that Democratic Senate candidates performed worse relative to 2016 than their House counterparts. In fact, the data shows the opposite. Voters liked the Democrats running for Senate more than they liked the ones running for the House. The lesson to take from the Kavanaugh hearings is that much of the procedural fighting that happens within the beltway does not have much salience with voters. It’s important to remember going into the 2020 cycle, which is sure to have its fair share of Kavanaugh-esque analysis. Observers watching the cycle should pay attention to the polls, take an honest look at the result, and try not let themselves use the result to confirm prior beliefs.