Monday, 17 September 2007

Howard v McKew not all it seems?

I've commented before that I think the results of opinion polls over the last couple of months are based more on expectation than reality.

Ninemsn's online poll question this morning was
Are you going to vote for Kevin Rudd's Labor?

This question was only up for an hour or so before it was changed to
Would you vote for Maxine McKew over John Howard?

Now, conventional wisdom is that it's a close race with election analysts such as Malcolm Mackerras predicting both a Labor win and McKew to unseat Howard. The closeness of the betting in the seat of Bennelong is shown in the following markets:
Sportingbet   Howard $1.75  McKew $1.95
Portlandbet Howard $1.80 McKew $1.85
Betfair Howard $1.94 McKew $2.00
However, the results from this morning and late this afternoon aren't telling the same story.



The Yes:No ratio this morning was 0.815 and this afternoon is 0.810 so it's not as if any vote stuffing is going on.

In the 2004 election John Howard received 54.33%. The result above is showing him with 55.26% of the vote.

Two points are clear:

1. The election is far from over.
2. Howard is not a shot duck in his electorate.


Jonathan Lowe said...

no offense, but unless almost all of the 85,000 people in the seat of Bennelong voted online and only them, then your arguement holds true....but you are trying to read too much into the stats, because this is obviously not true, and your arguement about this seat does not hold.

Jack Lacton said...


My experience with these polls is that they either reflect on the issue itself or, in the case of political questions, on national voter preferences.

If this is not reflecting on Bennelong then it's dividing along party lines.

Neither of those positions accord with current wisdom, which is why I'm suggesting that Morgan et al are reflecting voter expectation rather than reality.