[LINK] "Why The Pollsters Got BC Wrong"

[rfmcdonald]

Wonkman's recent post examining the rationale for the recent bad polling in British Columbia's election, and why there might not be that much of an incentive to fix it, is worth reading.

BC is perhaps the least homogeneous province in the country. The economic, social and demogaphic conditions in one community can be wildly different from those of a neighbouring community, to say nothing of the stark differences between a big city on the Island and a small town or rural area in the northern interior. In a province like Prince Edward Island, you can pick 50 random numbers out of the phone book and be pretty sure you’ve got a balanced sample; in British Columbia, you have to consider dozens of different factors and weightings (which means polling hundreds and thousands of different people!) just to get basic data.

And that’s a problem.

Because of that diversity, BC is expensive and difficult to poll. So are Ontario and Quebec, but Ontario and Quebec both have way, way, way more seats than BC, making their results more critical to the national picture.

This obviously bodes poorly. It means polling agencies have weak operations in a critical province, and—in situations like this—the lack of resources is plain and obvious.

But is that the wrong decision?

The bottom line is that Ontario and Quebec do matter more than BC when it comes to federal elections. Given the choice between throwing resources into a province with 107 seats or one with only 36, you’d need a very compelling reason to opt for the smaller province.