Wednesday, September 14, 2005

Dodgy Polls

As they say there are lies, damn lies, and statistics. This election campaign has proved that beyond a doubt. Some of you may have been curious about the poll results in the Sunday Star Times last weekend, and other of you may have wondered why the National Business Review hasn't run one for a while.

Well the good folks at Molesworth and Featherston have found out what's been going on. Since their material requires a subscription I can't link to it but I'll reproduce it here (apologies to M&F if this treads on any creative toes).

There has been no National Business Review poll since early in the campaign and there are conflicting stories why. One report is of a fall-out between UMR-Insight (also Labour's pollster) and the weekly business rag. Another suggests the pollsters have refused to do one out of the monthly sequence because they are too busy with their other work. A third rumour is that a poll last week was "pulled" for no apparent
reason. Sad, because it has been a reliable poll over the years and would have been a useful addition to the campaign period.

Meanwhile a Sunday Star Times poll completed mid last week, which we hear had National on 42, Labour 40, the Greens 6 and New Zealand First barely 5, was dumped. ( It was referred to in political editor Helen Bain’s column comment.) It seems the paper's not-so-savvy management wanted something more up-to-date to test the reaction to Don Brash’s ‘I misled the public’ admission over the Brethren pamphlets. So they commissioned a much smaller 500 person poll on a Friday night – the statistician in us blanches at the biases that would contain – which showed National out to a big lead. Even stranger they failed to give the numbers of their more reliable poll.

They weren't alone.

The Herald on Sunday also did a snap poll with an even smaller sample - 400 – but at least with the saving grace that it was over two days; Friday and Saturday. Unlike the SST poll it showed a big lead for Labour.

What was that thing about accuracy in reporting again?

And one other thing,. Sometime I posted this little item on the pornography problems besetting the police force. Having looked at some of the recent referrals and links to my blog I see my suspicions about the title have been confirmed. So for all of you dirty little monkeys who are turning up looking for hot naked copper sex, I'm sorry but you'll have to find your gratification elsewhere.

[A Friday addition]
It seems the NBR has torn s trip of Molesworth and Featherston about the speculation made above. Here is the response from NBR's Nick Bryant (again courtesy of M&F).

How bloody irresponsible of you to note three rumours about NBR's polls but not seek to clarify which, if any, was right. NBR last had a political poll three weeks ago. We weren't due for a political poll until the week after the election. When I realised this, also nearly three weeks ago, I contacted Stephen Mills to request UMR do a special poll for publication this Friday, Sept 16. I offered to pay whatever it took to get the poll done but Stephen informed me that he simply didn't have the resources to do any extra quantitative research. The UMR poll is a bloody good poll. However, we understand Stephen's position and his being unable to do this poll hasn't affected our relationship in the slightest. What you've suggested about
NBR in your two other scenarios offends me and I think you've done yourself a disservice by disseminating them.

No comments: