More on Dodgy Polls, Ctd
Yesterday I blogged about Colin Craig’s dodgy poll where he talks about 89% of people in Helensville wanting consultation. (Quite apart from the fact that the Conservative Party can’t spell Helensville.)
I contacted Research First to ask about the exact questioning and other pertinent details of Colin Craig’s commissioned poll.
I did this because of the history of their dodgy polls. Commendably their CEO Roger Larkins returned my call and pointed to the full poll results at the Conservative Party website. It appears they have learned a bit from the last foray into manipulating public opinion through dodgy polls and most questions are vastly improved.
However the report tells us something that Colin Craig failed to tell us. Lying by omission, if you will.
You see the polling was not done in just Helensville, rather it was done across two electorates, Helensville and Rodney. Further it was a survey of just 355 respondents…with a margin of error of 5.2%.
To make matters worse the report does not split the 355 respondents by electorate so we have no way of telling how many were in each electorate. Despite demographic breakdown of other details this was omitted. I suspect that is because the vast majotiy of the 355 respondents were actually in Rodney electorate.
If we assume a 50/50 split across electorate then that also makes a major impact on the margin of error, blowing it out to 7.5%.
Colin Craig was intellectually and statistically dishonest to suggest the poll result was just for Helensville electorate. I blame him and not the polling company for this, for even though Research First omitted the demographic split they still highlighted it in the title of the poll.
I am sure David Farrar will look at other questions and details in more detail but for now I want to focus on Question 8. This question is seriously dodgy.
The details are as follows:
By loading up the possible answers in such a manner they were actually forcing a positive outcome better than 66%. It was such a stupid question anyway because if you ask almost everyone if they want to be consulted on something they will almost always say yes. But to ask it and have two positive answers and one negative answer then the pollster, presumably under instruction from their client, have forced respondents to answer in the way the client wanted.
It is dishonest and makes the results for that question statistically irrelevant. The error is further compounded by the dishonest way that Colin Craig presented the results by saying that 89% of Helensville voters want to be consulted. That is actually a lie. The poll was conducted over Rodney and Helensville electorates and no split in results was given, in no way can he take the total figure and apply it just to Helensville.
Colin Craig continues to use dodgy polling results to support dodgy contentions. This is really amateur hour stuff politically and shows that Colin Craig hasn’t learned a thing from his bullshit polling in Rodney and Epsom prior to the last election. He may have a lot of money to burn being stupid but he is a real slow learner.







