Quote:
Originally Posted by transplant99
Yeah...really noticeable and i wonder why that is? Obviously they are polling different areas or different segments of the population.
IFF is good at explaining thi stuff...maybe he will chime in.
|
Well, thanks for that. As a disclaimer, I tend to know more about polling and politics in the US, but I can offer some explanations for effects like that, most of which are similar to what Resolute posted. Also, I'm no expert on this stuff--this is just what I've gleaned from spending way too much time prognosticating on the internet.
My understanding is this: the methodology of doing a poll is a bit more complicated than people realize--it's not simply a matter of calling up 800 random people, interrupting their dinner and asking a bunch of questions. Pollsters have demographic "targets" that they try to meet--so many young people, so many members of each party (at least in the US they do this--it would make less sense in Canada), so many seniors, etc. etc.
If you don't do this, you wind up with a great deal of statistical variation among polls--you could get vastly different results from week to week, and the reason might be something way outside your control--like, for instance,
Matlock was on, so seniors didn't answer the phone, and you had too many young people skewing the sample.
But each pollster does this a little differently, and as a result you can get a "house effect" where a pollster seems to consistently show a slightly higher or lower percentage for a particular party. This isn't the same as "bias" because it's not produced deliberately, it's just an effect of the methodology they use, and it's only "wrong" as compared to the mean findings of other pollsters.
What it means in practice is that one should try to compare apples to apples. If Ipsos-Reid has the Tories at 35% on April 15th, and Decima has them at 28% on April 25th, that may indicate a house effect from one or both, and not a "trend." If two Ipsos-Reid polls a week apart show the same trend, that is a bit stronger (though still not absolutely determinative).
Hope that helps. I'm sure there are tons of people around who know way more about this stuff than I do--I'm not even a statistician, just a political junkie.