Tuesday 25 September 2012

Lies, damned lies and opinion polls

I once had a job interview with a polling company. I'd done the serious Maths subjects at school, gone on to do some reasonably serious Maths units and uni and was generally half competent with crunching numbers. I was looking for work, so the agency I was signed up with sent me off to a polling company.

I flunked the interview totally. I was asked some fairly deep questions and my mind went totally blank. Ah well, so much for that career. About all I recall of the company is that they were in a dingy brown office at the cruddy end of North Sydney - a long hike up a big hill in the hot summer sun, and well away from the flash companies in the North Sydney CBD that employ the hot looking chicks.

During the interview, I was handed a fat blue book full of polling results and we spent quite a bit of time talking about the methods used to properly weight a survey to avoid skewing the results. For the people who interviewed me, that was a big deal. The quality of their reports hinged upon them getting that right.

According to what Paco has discovered, it's not a big deal for those doing the Democrat polling in the US.

Go and read it.

There. My duty is done.

No comments: