Tuesday, 24 March 2009

Lies, lies and damned......

It's often said that you can make statistics say anything, and it's fairly true. The Advertiser's ludicrous daily "VOTE LINE" gives people an almost random question (yesterday's was 'Do computer games dull the imagination?') 91% voted yes and 9 % No.
There is a certain sense of irony about this since said paper is flogging CD roms of various games at the moment in order to increase paper sales.
But the totally ludicrous thing is that the paper reports there were only 20 callers. Now, these figures therefore must at some point or other be wrong. Because if there were only 20 callers then each call repesents 5 percentage points. Therefore if 18 people voted YES it would be 90%, or 17 would be 85%, 19 would be 95....No way can 91% happen.
Even allowing for stuff ups, and dither dathering (and because the phone system only allows you to answer Yes or No I can't see how that could be the case) all percentages would still have to bemultiples of 5.
Of course the real issue is that the sample is nonsense...it is too small and is self-selected.
The result is therefore meaningless on this count also.

What is perhaps as disconcerting is the question for today which is "Do you support the Liberal Party's plan to rebuild the Royal Adelaide Hospital?"
The qualitative difference in importance is striking.
Now I will guarantee that there will be 10 times more votes in the sample at least, which may signify that the result is more meaningful. But of course it will be fuelled on both sides by the political interests, and the process of self-selection (that is you decide yourself if you want to ring up) which means that essentially the self-interested and the politically-inspired are likely to respond vigorously (no way of checking if they ring twice, or more even)
Truth is it's a bit of a waste of time!

PS. There were over 600 calls about the RAH question...I sunsequently penned a letter to the paper
Anyone who has been inside the RAH surely cannot believe that more than 80%
of South Australians prefer that rabbit warren to be redeveloped in preference
to a state of the art new hospital.

Some, of course, may; and
clearly the Opposition is using this as a point of difference in the lead up to
next year's election.

Yesterday's phone poll would have us believe
that more than three quarters of South Australians think that it's better to
rework the tired old buildings; , but it makes us suspect (what we already
know) that phone-in polls are subject to blatant flooding by interested
parties and thus statistically irrelevant.

No comments: