Pollsters have had their tail-ends handed to them for several election cycles now and it’s not too difficult to figure out why.
To understand why, let’s start with the obvious. Who is paying for the poll? Much like many other “studies,” the folks paying the bill expect a certain outcome. The folks being paid intend to deliver that outcome. It’s business – nothing new here. This falls into the “assume the conclusion” category – wherein you know what the desired outcome is and you try to backfill the data to support that outcome.
Somewhat less obvious might be the structure of the poll, the questions asked, and the available answers.
Disclosure: I tried to answer a few political polls many years ago and gave up. I refuse to answer them now. Politicians that want to know my opinion on any given issue are free to read the letters I’ve sent them; the pollsters can just guess.
The reason I gave up on answering polls was the structure of the questions and the available answers, and the obvious intent behind both. The polls I’ve been asked to answer were not designed to evaluate what I thought on the topic, they were designed to support a predetermined ideological position.
“Are you satisfied with current federal firearms laws? Yes/No.”
What if I answer “No”? Does that mean I want more restrictive laws, or does it mean I think the current laws go too far and are overly restrictive? The pollster is usually looking for evidence that the current state is not sufficiently restrictive – so a “No” answer will likely be logged as a vote of support for more restrictive laws. The fact is, a “No” answer may mean the exact opposite. Are there follow on questions to clarify this? Generally no, in my experience anyway.
“Do you support restrictions on abortion? Yes/No.”
What if I answer “Yes”? Does that mean I favor a total ban under any and all circumstances? What restrictions are we talking about exactly? Any provisions for rape and incest? What timeframe are we talking about here? More restrictions than are currently in place?
Point is the pollster asks for your position on a topic but makes no effort to understand WHY you hold that position. No nuance – and precious little context.
“Would you vote for Mitt Romney for president?” Depends, what’s the alternative, who’s he running against?
“Do you think assault weapons should be available to civilians?” Which civilians? Criminals are civilians. I don’t favor selling a semi-automatic rifle to criminals, but I have no problem with a law-abiding citizen buying one.
Polls and pollsters are going to continue to get it wrong. Whether this is due to the bias and desired outcome of whoever it is paying for the poll, or the bias of the folks structuring the poll, the fact remains that until they peel the onion back a few layers and understand the “why” behind the answers, they will miss the mark.
Polls and polling should be, logically, a straightforward exercise in root cause analysis. The first rule of root cause analysis is to ask the right question. If you don’t ask the right question, your odds of getting the right answer are nil.