Four polling analysis problems for the Democrats

So you talk to liberals and they are all pretty sure that they will beat Trump in 2020. When asked why, they invariably suggest that all of the candidates are beating Trump in the national head to head horse race polls, and that some of these polls show their top candidates up by as much as eight or nine points. Here is the most recent RCP general election polling averages:

• Biden + 5.4% over Trump
• Bloomberg +5.2% over Trump
• Sanders +3.7% over Trump
• Warren + 1.8% over Trump
• Klobuchar +1.8% over Trump
• Buttigieg -+0.6% over Trump

So those are not exactly as impressive for the Democrats as some of the outlier polls would have you believe.  Sure, all of the candidates are polling at least slightly ahead of the President, but does that really tell us anything? After all Trump spent most of 2016 trailing Clinton and Sanders by double digits. But let's move past the intangible ideas and into the historically consistent and tangibly relevant polling realities that we need to look at.

First, these are registered voter polls, not likely voter polls. If history has shown us anything, all adult polls are the most favorable for Democrats, followed by registered voter polls, with likely voter polls generally being the least favorable. From year to year this changes... but as a general rule when pollsters start to flip from registered voters to likely voters, their numbers tend to move towards the Republicans by at least a percentage or two. Taking this into account, the current numbers might start to look like Biden/Bloomberg leading by about four percent, Sanders leading by about two percent, Warren and Klobuchar approximately running even and Buttigieg down by a percent or two.

Secondly, there is the phenomenon of the last week polling consolidation. This is where the outliers come back to the rest of the pack and pollsters settle on the final number that they ultimately expect to be judged on. It generally starts to happen about a week to ten days before the election and for some reason seems to take the Nate Silvers and other media pundits by surprise every time. But it always happens. In 2016, Hillary had a polling average anywhere between five and seven points heading down the final month, only to see it shrink down to a little over three percent in the RCP averages at time of election. I am not expecting that pollsters will try to make anyone believe that the Democratic Candidate is going to win by twelve to fifteen points as some pollsters did in 2016. But I fully expect to see the Democrat probably leading, and then we will see that lead shrink (or go away) in the last twilight of the campaign. A five point lead might turn into a two point lead, and a two or three point lead might turn into a virtual tie. A virtual tie might turn into a Trump lead.

Thirdly, there still seems to be a small, but tangible bias in media polling that favors Democrats. In 2016, the final RCP averages had Hillary up by just over three points, and she won by two points. That is remarkably close, but still the polling was favorable to Clinton. Had she won the popular vote by three percent rather than two, she would probably be President. If you look at Nate Silver's predictions (based on polling averages), you can go back as far as you want and never once find Silver predicting a Republican to win a state wide race, who hasn't won. Every single incorrect projection Silver has made has been made projecting Democrats doing better than they do. Now, Silver does his own polling manipulation, but the reality is that if you simply rely on the polls as the pollsters provide them, you will over project Democrats every single time.

Lastly, the Democrat will likely need to win the popular vote by AT LEAST the same two percent that Hillary won. With the polarization between the coastal elites and fly over country at heightened levels, and given the unpopularity of impeachment in the battleground states, Trump could conceivably lose the popular vote by as much as three or four percent, and still win the electoral college. Even if that is a bit of a stretch, it's certainly not a stretch to suggest that a national poll showing the Democrat up by a couple of points, does not reflect any assurances that they will win the election. In fact, if I was looking at building a model today, I would certainly offer that the "actual" point of parity between the two candidates might be at the point where the Democrat candidate leads by somewhere between two and three percent. A Democrat winning by three percent or higher is favored, while Trump down by less than two percent (tied or ahead) is likely going to win. Now much of this will depend on the Candidate and the manner in which the campaign goes, but there is a very good chance that we will see the third such instance of a split between the popular vote and the electoral college vote (and let the protesting begin).

So if you look at all of these factors, suddenly you shouldn't feel too comfortable with a five point polling lead. Keeping in mind that these are just historical polling factors, and will have nothing to do with what Trump does to whichever Democrat is left standing. The President will have more money, better infrastructure, and the power of the bully pulpit this time around. He will be relentless in his drive to bring down the Democratic nominee by any means required. He ultimately may have already taken down Slow Joe Biden, even before the first Caucus. Watch for him to do what he did to the GOP field in 2016 as time moves forward. One by one until he has his desired opponent?