At the beginning of June, it looked like the “Blue Wave” might be turning into a blue ripple. The Republicans had suddenly pulled the Democratic advantage in the generic ballot (a poll that essentially asks a national sample of voters which party they’ll vote for in the upcoming House elections) down to about three points in the RealClearPolitics average, and in late May FiveThirtyEight showed a 4-5 point advantage for the Democrats. Democrats need a roughly seven point advantage in the generic ballot on Election Day to have a 50-50 shot at taking the House, so these numbers were pretty good for the GOP.
But now the polls have moved back toward Democrats: Fox News has Democrats up by nine, Monmouth has them ahead by seven, Reuters/Ipsos puts them ahead by 10 and Rasmussen Reports (which typically produces GOP-friendly results) gives Democrats a four-point advantage.
So what happened? And is the GOP doomed again?
The (weird) answer is that maybe nothing happened—and whether you believe Trump enjoyed a bounce depends on what sort of trendlines you like to draw.
Here are the technical details of this graphic (skip if you’re not interested): This graphic shows recent generic ballot polling data collected by FiveThirtyEight. I didn’t use FiveThirtyEight’s weights, but I did decrease the weight on tracking polls to compensate for the high number of individual data points those pollsters create. I then used local regression to draw two trendlines (with different spans) through the data—a noisy (red) one and a smooth (blue) one.
The basic message here isn’t tough: If you use the red trendline, you can see a dip in the Democratic advantage within the last month (look at the far right hand side of the graphic). If you use the blue trendline, you don’t. That trendline mostly stays between six and eight points (the black lines).
The red trendline represents a more reactive way of thinking about the polls, one that trades stability for detail. It’s able to catch small ups and downs in the data, but sometimes it’s too twitchy and sees a trend where there’s just noise.
The blue trendline represents a less sensitive way of looking at the polls. It’s less reactive to changes in the data, so it’s less likely to chase noise. Smoother trendlines have drawbacks as well: If they’re too smooth they can lose detail or be too slow to react to what’s going on in the data. But in this case the smoother line doesn’t see a GOP surge in early June (the far right hand part of the graphic).
And I think the blue line is right about the most recent “shift” in the polls. Democrats have held a roughly seven point lead for most of the Trump administration, and the averages have appeared to return to roughly that level. I’m not sure that positing a short-lived GOP spike in early June helps us explain anything about public opinion or predict anything about the upcoming elections. So for now I’m OK with thinking of it as noise.
But that doesn’t mean you should rule out the red line as a method for thinking about polls. The blue line might have underreacted to the Democratic surge in December 2017 (see the big spike in both lines), and more reactive trend lines sometimes catch on to new patterns more quickly. It’s good to be able to hold both methods at the same time—to take a long, cautious view of public opinion without ignoring new information that might be the start of a trend.
These polls raise one other important question—is the Blue Wave back?
Sort of.
Democrats have a real edge in the race for House control, but the outcome is far from guaranteed. The shape of the House map makes it possible for Democrats to score a mid- to high-single-digit win and still fail to get to 218 seats. President Trump’s approval rating has also improved since December 2017 (when he hit a low of 37 percent). Those numbers may not hold up (e.g., a strong majority of Americans disapprove of family separation, and that story is still developing —plus it’s unclear what else will happen between now and November). But at this point there’s still a lot of uncertainty about which party will control the House in 2019.
