In the latest polling for the Iowa caucus, both Barack Obama and Mike Huckabee are moving ahead of the pack and challenging the long-standing front-runners in their respective parties. This sounds really interesting but is it hype or reality?
I think that polling is not the “science” that it used to be. Consider the obstacles to the modern day pollster.
First, most polling is done by telephone. It is the cheapest way to have two-way conversations with potential voters. From voter data such as that available from your county clerk, a pollster can know how often that you have voted in past elections and want your party affiliation is. This information coupled with the demographics of your neighborhood can give them a good idea of who you voted for in the past. There are many questions that they ask in the course of a poll to verify their guesses by comparing actual answers with their predictions.
Part of the challenge to pollsters is to get a statistical cross-section of the electorate. Their model assumptions about voters are important to the accuracy of the result that they get. Have you ever wondered how 1,200 people are polled and in a headline the next day we are told with certainty what 300 million Americans think about presidential popularity or Hillary’s latest healthcare idea?
The biggest challenge to this model is technology. A large percentage of people under age thirty have only cellular phones and not a landline. These folks are not listed in Ma Bell’s directory. People also have IP telephony and other means of communicating that are not listed. This coupled with traditional unlisted phone numbers on landlines makes contacting a cross-section of voters even more difficult. The only thing that is saving their models of the electorate is that many in this age group are not high propensity voters. Older, high propensity voters tend to have traditional phones.
Another challenge is who is actually answering the phone? Most times that a telemarketer calls my phone, we hang-up on them. Due to the number of polls done in modern political campaigns and the small population of a state like Iowa, I think many folks are burned-out answering their phones too. Are the results showing shifts in the candidate preferences of Iowa voters accurate or are the number on non-responding folks causing results to be weighted differently to keep the pollsters in business?
Most people made up their minds months ago and it would take a big event to change their preference. Neither Obama nor Huckabee can point to such an event. (Yes, some might argue that Hillary and driver licenses for illegals qualify but we will have to see.)
If given my assumption about the lack of a big event, the logical place to go to explain any shift in voter preferences is that the “undecided” are finally settling on their choice. My problem with this is two-fold. First we must ask is this “undecided” voter who up ‘til now has not been paying attention really firm in their commitment to a candidate and second, will they actually go to the polls and vote?
Lastly, going into the holidays will detract from campaigning. This is a season for family and the sacred not secular. This will result in committed people voting and those with little interested getting distracted and not making the extra effort to vote. I doubt anyone knows how to model this because it has never happened before. The timing of Iowa is an unquantifiable variable that should not be minimized.
I could go on but I think I have made my point. The polls are entertaining but should not be trusted. There are too many variables that make Iowa difficult to call prior to January third.
