Sunday, March 27, 2016

Push polling, but in an entirely different context

There is an interesting political strategy I learned about recently, called "push polling". Push polling is a polling technique designed not to get an opinion. The goal is to change the opinion held by the recipient of the poll. It's usually done by phone because the voice can be used to set the tone of the conversation much better than an online poll or one conducted by email.

A leading question is one where the recipient is pointed in the direction of the desired answer. Push polling is a poll with a series of questions leading the recipient to answer the questions in a certain way so as to reinforce a political agenda. Instead of asking questions that seek an unbiased opinion, the questions lead one to think and change their mind. It's very subtle and there are some very interesting articles on the subject that any search engine can find.

I've been the recipient of a push poll. The poll was designed to turn my opinion against the candidate of my choice. Once I saw what was happening, I doubled down on my answers and made sure that the caller knew that I knew what was happening. Back then, I didn't know what it was called, but I did know what was happening. I let the caller know in no uncertain terms that I was not going to change my mind. Click!

Polling can be used in another way to reinforce opinion or outcomes in an election. For months now, we've been treated to polls showing that Hilary Clinton has the lead in support from potential voters, over Bernie Sanders. For awhile, this seemed to have held true. But then some people started to notice a divergence from the polls to the outcomes. 

The divergence was not subtle. In his article, Poll-Defying Pattern Predicts Sanders Victory, Jonathan Greenburg at the Huffington Post has noticed some very interesting trends in polling. Greenburg has identified huge discrepancies in polling vs the outcome of at least 4 elections. See the chart below:


In one state, Minnesota, the divergence from the outcome predicted by the poll was 57%. That is one giant yawning chasm of error on the part of the mainstream press. In those four contests, the average divergence was 35% and Sanders was the winner.

In Utah and Idaho, Sanders won by 79% of the vote. Yet, no major news outlet did any polling to predict the outcome. They did go out of their way to poll Arizona though. Check out the list at RealClearPolitics.com (RCP). I scoured their list of polls and only found one poll on Clinton vs Sanders and that poll had Sanders up by 8%, and that was only for Utah. The discrepancy was 51%. The news outlet that actually polled Utah was the Deseret News, a local news source.

There were no polls for Idaho.

This would seem like a fluke, but then there was a vote held March 26th in Washington, Alaska and Hawaii. Washington is a big state, but alas, there were no polls on that state at RCP. Perhaps that's because Sanders won Washington in a landslide with 72% of the vote. Sanders won Hawaii with 70% of the vote. He won Alaska with 81% of the vote.

There were no polls for any of those three states at RCP. If you know of one from anywhere else, I'd like to see it. I'd guess that they are polling states that favor Sanders, but they're not publishing the results. 

Washington is a giant among those states. With more than 100 delegates at stake, why were no polls published to predict the outcome? It has a population of 7 million, more than the 6 million in Arizona, but no one felt any obligation to conduct a poll in Washington and publish it? 

It would appear then, that the mainstream media is attempting to use the polls to set and support their narrative: Clinton is the inevitable nominee. 

This is push polling, to the extreme. The mainstream media seems only willing to publish polls where Clinton is the likely winner or at least has a chance of winning. I guess this is what we can expect when 90% of the media is owned by 6 parent corporations. I wonder if we can predict a Sanders landslide in a state that has had no polls published before the vote.
Post a Comment