Trolling by Polling

funny_hillary_clinton_vs_donald_trump_election_postcard-ra42557157e2745d582361c25dfe3b9db_vgbaq_8byvr_630

By Mike Cronin

Now that the presidential campaign season is in full swing, we are being treated to the usual inundation of “demographic” polling results (i.e. how are “soccer moms,” Hispanics, gays, white men, etc. going to vote?)  Taken individually, as they are usually reported, the vast majority of these polls tell us nothing useful.  Piolls that tell us how a state will vote are getting closer to valuable.

Consider: This page at Real Clear Politics lists numerous polls with entries similar to this one:

Race/Topic: State X:         Poll : Qunipiac                Results  Trump 44, Clinton 38                            Spread: Trump +6

Trump is up by six points over Clinton in State X.  Sounds dire for Clinton, right?

You still might be getting “trolled.”

Regardless of how you’d like the race to go, we need to remember how presidents get elected: by winning the Electoral College vote.  The popular vote heavily influences the Electoral College, but the nature of that influence is determined by state laws. As we’ve seen as recently as 2000, it is possible to win the popular vote but lose the electoral vote and the election. In terms of polls: It means that generic surveys of this or that demographic, such as the one cited above, are almost useless by themselves for telling us how an election might go, especially in “solid” red or blue states. It makes no difference if a candidate’s poll numbers go up in a state that was already going to vote for them.  It makes a great deal of difference if a candidate’s numbers change in a battleground state with a lot of electoral votes, such as Florida or Ohio – such a shift could move electoral votes from one column to the other, perhaps enough to get to 270.

Thus, even taken in aggregate, “demographic” polls aren’t much of a barometer.  A much more instructive product would tell us how every state is currently trending, with caveats for the states (Nebraska and Maine) that are not “winner take all.”  Such a tool would give us a much better predictive assessment of the electoral college votes likely to go to each candidate.

Such a beast exists: http://www.270towin.com/. They have broken down the race several ways.  If we look at their map that “kluges” current polling data with expert forecasters’ opinions, Mrs. Clinton already has 272 votes in her camp, two more than are needed to win.  But  if we look at their chart that displays the accumulation of polling data only and omits “expert opinion,” then Clinton has about 200 electoral college votes in her camp (out of 270 needed to win), while Mr. Trump has 163.  Either situation sounds much more troublesome for him than “Trump up by six” sounds for Clinton, doesn’t it?

If “demographic” polls vice “electoral vote” polls do little to predict the outcome of the Electoral College race, why publish them? At least two possibilities come to mind:

  1. There’s nothing like stirring the pot in order to keep you tuned in and watching advertisements.
  2. To shape voters’ behavior in some way favorable to whomever commissioned the poll. Example one: Clinton’s current lead only translates to victory on Election Day if enough voters actually go to polling places and pull levers.  If you want Clinton to win, maybe you paint her as losing ground in the polls in order to generate a hint of doubt. Maybe that will motivate folks to go vote that might have stayed home if they felt comfortable she was going to win. Example two: If you want Trump to win, you might commission such a poll in order to generate enthusiasm by painting him as an underdog coming from behind and pulling ahead – a narrative that always does well in America.

Manifestly, Donald Trump still has an uphill battle.  In his best case scenario, he has to take 107 more votes and keep Clinton from gaining, while in his worst case he needs to take at least three of Clinton’s existing votes away!

Despite being much more instructive than run-of-the-mill poll reporting, even such tools as the “270towin” charts are not infallible, nor is the sentiment recorded today going to be the same on election day.  Clinton’s recent bout with “pneumonia,” clumsy messaging regarding her overall health, and “deplorable” commentary on Trump supporters certainly helped Trump’s polling tick up a bit, but did it affect the electoral vote trend?  Time will tell – and we still have to get through the “October Surprise.”

While such and event or revelation may yet upend the race, there is sure to be a battle over the next few weeks for the remaining available electoral votes. Poll results that don’t tell you how that aspect of the election is going are probably not worth your consideration.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s