Do you believe in Polls?

[maplesyrup]

There are several different polling firms in Canada and their results often conflict with each other. Do you think these polling firms are connected to one or another of the political parties, and if so, which polling firms are connected, or supportive of, which political party?

My understanding is that these polling companies right now are polling every day.

Could it be that when their polling results are better for the political party they support, those are the polling figures they release to the public?

I think maybe that's the way it works, eh?

[maplesyrup]

Now we have another poll out today, quite a contrast from one out a couple of days ago, by another polling firm. This one a Compas Poll for National Post/Global shows:

Bloc 11%

Cons 31%

Libs 39%

NDP 17%

[August1991]

The Conservatives are now virtually neck-and-neck in Ontario at 39% to the Liberals' 42%. Only in Toronto do the Liberals hold a strong lead.

National Post Compas Poll

I think this is too much of a jump for the Tories. The margin of error for Ontario is probably +/- 6% or so so this is likelthe top end.

On the other hand, it makes sense that rural Ontario is beginning to pay attention. And they well may feel comfortable with Harper.

[maplesyrup]

Does anyone know how these polling firms operate?

Do they choose the best data possible for the political party they are closest to, in their daily polling, and publicize it?

It kinda makes sense to me, as the intention it seems with publicizing these polls, is to manipulate voters.

[August1991]

Do they choose the best data possible for the political party they are closest to, in their daily polling, and publicize it?

Would you choose a doctor who always told you good news even if it was wrong?

The newspapers hire the pollsters. The newspapers don't want spin (unless it will sell more papers or generate more advertising revenue).

Rather, a sample of 1000 people, if wisely chosen, should give a good measure of the whole electorate.

A preoblem in Canada is that we don't really have one election. We've got more or less three: Quebec/West/Ontario + Atlantic. They should really have samples iof 1000 in each.

The results is that this Compas poll of 1500 has about 600 or so in Ontario. The Ontario predictions are not "precise". Another poll in a day might show the Tories/Liberals elsewhere.

[CanadaRocks]

Could it be that when their polling results are better for the political party they support, those are the polling figures they release to the public?

Nice point that polls can be biased to favor a predetermined outcome.

Professor Linda Gordon once told me that "everyone has a bias and the only way to overcome it is to recognize it is present."

Of course, Professor Rajput once pointed out that any poll or survey can have its results skewed while maintaining the appearance of true data.

What I gathered from these folks, is that one should always question, holding true to our personal beliefs yet keeping a mind open enough to update it with new information.

Trust a poll? Unlikely. Consider it? Absolutely.