Friday, August 25, 2006
The problem with pollsters, #182
If you take a look at the latest from SES, not only are we supplied with numbers telling us how the political parties are doing, we're provided with commentary telling us why those numbers are the way they are. While one can argue that the numbers are scientific, the commentary certainly doesn't appear to be. Supporting data is nowhere to be found. Even the data that is provided shows no movement in numbers beyond the margin of error. So, in effect, the commentary is describing changes that might not be happening at all.
For those who think SES gets a lot of leeway because they've hit the mark in the last two elections, they haven't changed their margin of error to zero. Even they know they got lucky.
This ties into what I believe is the biggest problem with polls in a democracy. They end up creating the political narrative, instead of forming one small and sometimes error-prone part of it.