Let’s talk about Election Polls

Regardless of political leanings the headlines this week are full of stories about how Clinton’s lead in the polls has “vanished,” to quote Right Wing mouthpiece Breitbart.com or this Politico story spinning a story about Trump falling “back in love with the polls” after the race tightens back up. We even see Nate Silver’s FiveThirtyEight website announcing that “The Odds of an Electoral College Split Are Increasing.” Either way, what they’re reporting on is the very common fact that polls tend to tighten up right before a presidential election and, more to the point, that said tightening up turns out to have virtually ZERO bearing on the final results. Below I’m posting the Real Clear Politics graph of the NATIONAL AVERAGE of presidential polls for 6 months ending as of yesterday, October 31. As you can see, the national average of polls is showing Trump’s numbers rising from a low in the middle of October but Clinton’s staying relatively steady. And she still leads by 2.2 percentage points across all the major national polls. Now scroll down.


The next slide I’m going to post is from the LAST presidential election. Anyone remember that one? Obama vs. Romney. Obama won handily with 51.1% of the popular vote compared to Romney’s 47.2% and trounced Romney in the Electoral College with 332 electoral votes to Romney’s 206.

Remember, also, that in the lead-up to the election the media was busily writing stories about the tightening of the race and that things were totally up in the air. Here’s a graph of the 6 month period leading right up to that election.



Just DAYS before the election Obama was only ahead in the national polling average by .7 percent. Essentially nothing. Yet he beat Romney by almost 4% of the popular vote and by a huge margin of electors. Hmm? What’s wrong with the polls?

Now let’s look at something even more telling. The following chart takes the previous 2012 Obama vs. Romney chart and overlays it on top of Bush vs. Kerry polling from the last months of the 2004 election. Now, I don’t know about the rest of you but I remember going to bed the night of the Bush-Kerry election with it STILL too close to call and not learning Bush had won until the next morning. This chart, ending November 2, 2004, shows Bush leading Kerry by 1.5 percentage points. Much MORE than the 0.7 percentage points Obama is shown leading Romney. So if you were to believe polling numbers that means Bush should have trounced Kerry or Obama should have barely squeaked out a victory over Romney. Neither one is true. Bush beat Kerry but only by 1.4 percent of the popular vote (which is close to the polling for once) and only by a thin margin of 35 Electoral College Votes.


What does all this mean?

It means that polling is notoriously imprecise but that it makes for neat graphs that are easy to talk about.

It also means that the media needs a horse race, right up until the end. Even in the case of an election like Obama vs. McCain, where the polls were showing a huge spread that turned out to be true, the media was calling it like it was down to the wire. They need that for the same reason the networks want the Superbowl to be close, not a massive blow out. Nobody wants the fans to check out early. There’s too much money riding on it.

In presidentially elections polling is generally predictive. The person who is ahead in the polls generally wins. But getting all wrapped up in the last minute wiggles of the graph is just silly. That’s the media trying to wring dramatic tension out of some math.

I think it’s going to be a Hillary Landslide but that doesn’t mean I’m going to get complacent. Everyone still needs to do what they can do to support their candidates, up and down the ballot. And most of all, VOTE.


One thought on “Let’s talk about Election Polls

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 )

Facebook photo

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

Connecting to %s