Everything seems to be going the Democrats’ way on their march for House control in November. Republicans, including House Speaker Paul Ryan, are retiring left and right. Democrats continue to overperform in special elections. Nonpartisan handicappers, including CNN, continue to move more races in the Democrats’ direction.
And yet, the Democrats’ position on the generic congressional ballot seems to have worsened since the beginning of the year. Just this month, four “gold standard” pollsters (i.e., live interview surveys that call cell-phones and are transparent about their data) show an average lead for the Democrats of just five percentage points on the generic congressional ballot. That’s down considerably from 14 points in December among gold standard polls.
So what is going on, and should Democrats be worried?
There certainly has been a decline in the Democratic position since the end of last year, though it’s not entirely clear that the changing numbers should shift our perceptions of what will occur in November.
The Democratic lead on the generic congressional ballot in gold standard polling over the last two months is among the worst it has been this cycle. They were up by just five points this month. In March, Democrats were ahead by seven points in seven gold standard polls. Before that, there hadn’t been two consecutive months when Democrats held a generic congressional ballot lead in gold standard polling of seven points or less since March and April 2017. That suggests that what we’re seeing is real movement and not statistical noise.
The seven-point mark is important because it’s around the break-even point for House control. Because of the way the district lines are drawn, Democrats need to win the national House vote by a substantial margin to win enough seats for a majority. While we cannot be sure of the exact advantage Democrats will need in the national House vote to take the chamber in November, a seat-by-seat analysis reveals that a seven-point lead in the national House vote is probably what Democrats need to take control. Fall below that line, and Republicans are more likely to hang on.
But while the gold standard polling has shown a massive shift towards the Republicans, other polls don’t show anywhere near that change. This month the non-gold standard polls have Democrats ahead by eight percentage points on the generic congressional ballot. That’s right near the nine points they were up in December in non-gold standard polls.
Many of these non-gold standard polls weight by variables such as party identification that stabilize results. The volatility of the gold standard polls may be because more Republicans are answering polls than in December for some reason, which is a good reason to weight by party identification. Still, these non-demographic variables such as party identification are not necessarily consistent (e.g., a person can decide tomorrow that he or she doesn’t feel like a Democrat anymore, so you’re weighting up Democrats when you shouldn’t be). This is why most gold standard polls don’t weight by party identification.
Perhaps an even bigger reason not to make too much of a change on the generic ballot is history tells us that sizable shifts at this point may not mean that much come November. I collected generic ballot data from the last 20 midterms (since 1938) at this point in the cycle. Polling at this time in the campaign is telling of the November result, but only to a point.
We expect two trends to occur in voter behavior between now and November based upon previous campaigns. The first is that there is a reversion towards a tied result. That is, big leads tend to become smaller. The second is that the president’s party tends to do worse in the actual result than the generic ballot suggests at this point. These two forces sometimes compete against each other, such as this year, given Democrats held a big lead but are also the opposition party.
Past campaigns suggest that a 14 point lead on the generic ballot at this point for the opposition party like the Democrats held in December forecasts about a 9.8 point win in November. A five point lead, however, translates to a 6.4 win for the opposition in November. That’s a difference of just a little over 3.4 points in the forecast final result, even though the polls differed by nine points.
Forecasting the Democrats to win by 6.4 points versus 9.8 points is an important difference if those were the final results. Projecting a November result from polling at this point, however, has a wide margin of error associated with it. A 6.4 point margin forecast versus a 9.8 point margin forecast based off the generic ballot are not significantly different projections statistically at this time.
The generic ballot still points to a national environment that is going to be strongly Democratic in November. That’s in line with the special elections and individual House race ratings. Whether that translates into Democrats falling just short or exceeding the bar necessary to gain control of the House is simply unknowable at this time.