I have updated my info page to reflect the fact that I am now a member of Young Nationals. Regard that information as you will.
So it’s fairly appropriate that I post about this topic. Of course, politicians of all stripes have used lies, damned lies and statistics. The whole debacle about what an “average” teacher gets paid is a fairly recent example. (I did find out what Kate Gainsford defines as an “average” teacher. I found an interview on RNZ Morning Report that she rejects the idea of an “average” teacher entirely). There have been more stories in the media about the PPTA strikes, but as usual, they lack the detail and appreciation of the context of the industrial action.
It is more than about pay but that’s what the stories centre around. On that note however, I will admit that my earlier prediction of there at least being three strikes before a circa 2.5 percent increase offer was off. It was only one strike and I don’t think the increase was as big as that.
What was rather disappointing to me was the butchering of the term “statistically significant” which has a specific meaning. But it’s not surprising. Assisting students in understanding it is something I’ve had to do for hundreds of times as a tutor.
In question four of October 21 question time Bill English provides poor context and explanation of a statistically insignificant finding and David Cunliffe doesn’t let the technical meaning of term get in the way of trying to score political points.
Bill English said:
…The decline in median weekly income is $9, which Statistics New Zealand points out is not statistically significant. It could have dropped for any number of reasons, such as more children in the population, drops in interest rates—
A bit later David Cunliffe asked:
Does the Minister of Finance consider an income drop of $9 a week to be statistically significant to a family who, as Veda Advantage recently reported, are among the tens of thousands of New Zealanders who are finding they just cannot afford to pay their bills?
The word significance in statistics has a different meaning it has in ordinary language of meaningful or important. It relates to how strong the evidence is for making a conclusion based on the data available. In this context, there wasn’t enough evidence to conclude that there was an actual drop in the median income.
Bill English instead could have said something along the lines of Statistics NZ could not find conclusive evidence of a drop in median incomes. The reasons for a drop in his answer have nothing to do with “statistical significance”. A correct explanation in this case would have been far more suitable for him politically. He could have truthfully said “there is no evidence of a drop in median income”!
Despite lies, damned lies and statistics, Statistics NZ is a very trusted department. The Minister of Statistics in the past several years of government of both parties has been asked the fewest oral parliamentary questions. Hardly even patsy ones!
David Cunliffe outright butchers the term, using the words “statistically significant” well outside of its technical meaning, when he really meant the ordinary meaning of the word. He shouldn’t have used the $9 figure, but if he had to he should have asked whether Bill English considered “a median $9 to be unimportant” for the average family. (There’s that politically charged word “average” again…)
Posted by ivorytowerkiwi