On the Uses of Polling

Polling often gets a bad rap. Politicians are accused of using polls as a substitute for principles. I’ve worked with candidates who didn’t want to put a poll in the field because they didn’t want to know ahead of time how the election would turn out.

Both represent misunderstandings of the legitimate use of polling.

To answer the easy one first – the politicians who didn’t want to see a poll done because they didn’t want to know the future – polls are not, contrary to popular opinion, predictors.

What! You exclaim. What do you mean? Of course they are used to predict future.

Not really. Just about the only time a poll is used as predictive device is when a presidential candidate is determining what states to compete in and when a party committee determines what districts to play in. For example, the National Republican Congressional Committee might look at a poll and determine that Kathy Castor is not vulnerable in a general election therefore decide not to put any resources into that seat. That is an example of the predictive use of polling, but it is not the primary function of a poll.

There are two main types of polls – baseline and tracking polls.

If you have never worked in a paid capacity on a campaign, you have probably never seen a baseline poll. These are the polls done early in the process that are used for message development. Contrary to popular opinion, when properly used, polls are not tools for picking positions. It is used for picking what you emphasize – for example, a clear majority thinks that Sen. McCain is dead wrong in his belief that invading Iraq was the correct action. He did not say, a la Edwards, call his support for the invasion a mistake. Instead, he talks about Iraq issues where he is less at odds with the American people – his criticism of Rumsfeld (which he exaggerates to the point of untruth, but that’s another matter) and his early calls for a “surge” in Iraq. Some issues, he avoids talking about all together. McCain is fairly far out of the mainstream on abortion and while he hasn’t changed his position on the issue, he does refrain from speaking about it to swing voters.

Because a baseline poll is done early, before most, if not all, of the advertising, it is not predictor, because it is done before all the tools of the campaign have been put into motion. It doesn’t say if a candidate will win, it merely helps illuminate possible paths to victory.

The other kind of poll is a tracking poll. These are those regular polls we see coming from Gallup, Zogby and Rasmussen. They are not predictors, they aresnapshots in time. When used properly, it is not used as a magic 8 ball to guess what will happen in November. Instead, they measure movement from the baseline.

A candidate will do a baseline poll to find issues and messages that will be relevant to voters. The candidate campaigns using that information and the tracking polls (which usually will have a smaller sample size – mainly as a cost-saving measure because it does no good to spend your entire budget on polls and have nothing left for media and field) measure whether they are getting any traction. They may also be used to measure how voters react to issues that come up in the middle of a campaign.

For more information about polling, I can recommend no better site the www.Pollster.com. Another go site with an unique take on polling is www.FiveThirtyEight.com.