"The only poll that counts is to one cast on election day." That is the proverbial comment among political strategists because polls can be misleading.
I am one of a number of people who are troubled by the accuracy, especially when it comes to push button polling. That's the poll where an automated voice asks you to participate and you respond to questions by pushing a button on your phone pad.
Like many people, my husband finds these polls annoying and intrusive. If I am home, he hands the phone to me. Otherwise he hangs up. I go through the poll, mainly because I want to know who is doing the polling, what questions are they asking, how are they asking the questions, and what can I learn from their way of collecting this data.
I am troubled because it seems like only the curious and the furious participate. You either are passionate about a particular candidate or issue, or like me, you want to know what is going on. So you know that many people deselect themselves because you hear the stories about hanging up on polling calls.
The second question that troubles me relates to who is in the sample size. I don't know many people who get polls on their cell phones. With the decline in landlines, at least in this state, you have to wonder if your most active voters are ever contacted.
Here is a case in point. During the 2008 Presidential campaign, I rode from Colorado Springs to Pueblo on a bus with then Vice-Presidential nominee Joe Biden, then Senator Ken Salazar, and former Governor Roy Romer. National polls had come out showing Senator McCain leading Barack Obama, even in my state. That made no sense to me so I politely argued about the validity of the polls. Biden called his pollster from the back of the bus who then engaged in a discussion with me. So here were some of my questions. Did the polls include cell phone owners (who less and less have landlines)? Did they include newly registered voters (like the 250,000 we had just registered)? Did they go to infrequent voters who might come out in this election? The answer was no, no and no. The results in Colorado on election day were a 9 point victory by Obama over McCain.
Then how is the sample weighted. Coloradopols.com pointed out recently that the current Reuters/Ipsos poll, which showed Ken Buck in a lead over Michael Bennet, consisted of 48% Republicans, 40% Democrats, and 8% Unaffiliated voters. Are they assuming that few Unaffiliateds will vote this fall? They still represent over 30% of the registered voters. And there sure isn't that much difference between the number of Democrats and Republicans in the state.
The recent Democratic primary certainly points to how the public polling was off. Even the private polling didn't show the spread. So while polling might help campaigns identify categories of people to target in your race, or places in the state where you need to concentrate Get-Out-The-Vote, they may be sending the wrong message to the political pundits. What you can count on is the poll cast on election day.