A Tale of Two Polls

We are constantly hearing about poll results and what they mean. So often, the actual question is not discussed and the results are spun to get as much media attention as possible.

For instance, if I were to run a poll in January asking "do you find that your interest in ice cream has declined in the last 6 months" one can easily surmise that the majority of people would say yes. After all, who has as much interest in ice cream in January as opposed to July?

But, if we were to make a headline out of the results, then you would probably see something like "Canadian Support for Ice Cream Industry on the Decline"

Now, suppose the question had simply been, do you still like ice cream? We can easily assume the outcome would be quite different.

Where am I going with this? (I wouldn't blame you for asking) I recently came across two different polls about our involvement in Afghanistan.

The first was a Canadian Press-Decima Research which ran in the Globe and Mail under the headline "Support for Afghan intervention waning: poll". The question that was asked, from which they came to this conclusion, was whether the number of deaths in Afghanistan was acceptable. Say what?

Who is going to say yes to that question (surprisingly 25% actually did) as no death is acceptable. Many who read my writings know that I am an ardent supporter of our troops, and their mission in Afghanistan, but no death should ever be acceptable. (I will refrain from commenting on how many soldiers died due to ancient equipment while the Liberals were in power, oops, too late) If we found their deaths acceptable then why do we bother to even give them guns to defend themselves?

On the flip side, I also came across a poll run by the Dundas Star News. Their question was simply "Do you support Canada's role in Afghanistan beyond 2009?" The results are quite a bit different. As I type this, 89% say yes.

Of course, the Dundas Star News poll is unofficial and could be swayed by multiple votes by the same people but we also have to look at the way the major pollsters get their results as well (something that is usually glossed over)

First off, the survey sample is 1,000 people. I often question how 1,000 people can stand for the opinion of 30+ million but I am not a statistician. What I question are the people who are giving the answer. The majority of these types of polls are telephone surveys. Now, think about it, how many people actually;
A. are home when they call
B. given the preponderance of call display, actually answer, and
C. if they do answer, have the time to go through the survey which consists of at least 5 mins of questions just to ensure you are eligible to answer the questions (for the few times I accidentally answered the phone I told them my wife works for a marketing company because that instantly disqualifies you)

So basic demographics, and psychology, tells you immediately that they are not getting a true sampling of Canadians. They are missing most of the people who work routine hours and have kids in soccer, dance, hockey, etc and anyone who actually has a life. What they do get are a lot of people who have an axe to grind because they love to tell people what they think.

There is also the question of cell phones. More and more people are forsaking landlines and going to strictly cell phones. This is increasingly true in the major urban centres. Due to the cost involved, pollsters rarely do cell phone surveys and they would most likely get very belligerent answers. (I know I would be ticked having some polling company costing me money on my cell)

So, the next time you read the results of a poll, ask yourself, what kind of person actually answered these questions, and take the results with a very large grain of salt.