Things to Consider About Presidential Polling

You can’t throw a rock these days without hitting the results of some sort of presidential poll. There are all kinds of them out there. Some show Obama ahead. Some show Romney. It is a cliché, but it’s true – the only poll that counts is the one on Election Day.
(By the way, for people who think Romney’s goose is cooked with polls showing him consistently 1-3 points behind Obama in mid-September, think about this. In mid-September 1980, Jimmy Carter was ahead of Ronald Reagan by 44 to 40 points in the Gallup poll. By mid-October, Gallup had Carter up by 47% to 38%! Reagan went on to have good debate performances and won the presidency with an Electoral College vote of 489 to 49!)

So what is the value of a presidential poll? For me to keep watching polls, as a mere consumer of news, is like planting a seed and then digging it up every few days to see if it is growing yet – or any. It produces very little except frustration. For the candidate it is essential – to see if his message is resonating, to see if the other guy’s message is resonating, to see who is trending up, down, or neither. It is the ultimate metric of whether or not the campaign is succeeding in moving enough voters in the direction of one’s candidate.

If you are a poll watcher like me there are some things you need to know before you put any stock into what you read. Those things all relate to the accuracy of the polls you are watching.

1. What percentage of Republicans, Democrats, and Independents are included in a poll’s sample? In the 2010 mid-term elections, Democratic voters outnumbered Republicans by three percentage points. That gap held roughly true in 2000 and 2004. 2008 was an anomaly, when Democratic voters outnumbered Republican voters by seven percentage points. Most current, high profile presidential polls have over-sampled democratic voters significantly. CBS’s poll is a good example. In their poll, Democratic voters outnumber Republicans by a whopping 13 percentage points. Why the over-sampling? I won’t try to look into someone’s mind or heart, but it is producing a result that I believe is inaccurate by several percentage points in Obama’s favor.

2. What day of the week or time of day was the poll conducted? Polls taken just during the morning and afternoon hours of the weekday tend to over-sample Democratic voters. Polls taken just during evening hours or on weekends tend to over-sample Republicans.

3. Where in the questionnaire was the respondent asked who they support for president? The placement of that question – before or after the consideration of certain issues – can influence a wavering person’s response – at that moment.

4. Undecided voters tend to overwhelmingly vote for the challenger. In every presidential election since 1964 – except for 2004 – every percentage point of undecided voters, on Election Day, went for the challenger. If Obama goes into Election Day with the 47% or 48% he appears to have today, he will probably lose. Being ahead alone isn’t enough – for either side.

So how can the typical person possibly know the answers to questions 1, 2, or 3? The answer is they probably can’t. And that is precisely why unless one is involved in a campaign enough to be in the know, he/she shouldn’t put much stock into presidential polls.