Friday, 10 May 2013

Long Posts and Small Fields

Small Fields:

One of the advantages of state elections from a reporter's point of view is that the number of voters is smaller, which means results are known sooner. With only three seats to count, this applies doubly in the TasLegCo.

My past predictive record is just above 90%. I only recorded 2 official predictions in the previous post:

  • Jim Wilkinson will retain Nelson
  • Vanessa Goodwin will retain Pembroke

You might expect us to get around 1.8 of those predictions right. The good news is that we got better than that, and so far have a 100% accuracy rating for Tasmanian state election predictions. This highlights the risk of working with small sample sizes, which I will no doubt end up doing at some point in the future of this blog. Meanwhile, our overall accuracy is at 90.3: up 0.2 percentage points.

I also made five unofficial predictions:
  • Leonie Hiscutt to win Montgomery
  • Cheryl Fuller to place second in Montgomery
  • Kevin Morgan to place last in Montgomery
  • Allison Ritchie to place second in Pembroke
  • Tom Baxter to place second in Nelson
Four of these were correct, including the first and most relevant prediction on that list. So had I gone out on a limb and predicted a Hiscutt victory, then I would have 100% off of three predictions (and raised our full record to 84/95 or 90.4%). However I was wary of that prediction and, as it turned out, rightly so.

Long Posts:

Normally I would skip over the correct predictions and focus on where I went wrong. As I didn't actually go wrong on any of the significant predictions, I'm going to look at why I was justified is playing cautiously with the Montgomery prediction.

Firstly, there was no reliable polling for Montgomery, so I was basing my results on generalisations and past results. Secondly, no-one in Montgomery had the kind of obvious advantage as in the other two seats. These combined make any prediction shaky to begin with. However, I set out my lengthy reasoning last week, beginning with the following lines:

I'm going to call Mr Morgan as the first to drop out, with most of his support flowing to Ms Fuller. The three way battle is hard to predict.

Four scenarios followed - two won by Mr Vincent, one by Ms Hiscutt and one tossup between Ms Hiscutt and Ms Fuller. None of these scenarios unfolded, because the one unofficial prediction I did not get right last week was that Mr Morgan would drop out first.

Ed Vincent polled the lowest, with only 9.73% of the primary vote (ignoring informal votes), presumably as he and Ms Hiscutt had a very similar voter base, and Ms Hiscutt got the bulk of them. This highlights the problems with first past the post polling that I have been discussing over the last few weeks; similarly placed candidates can undercut each other and let a less popular contender come out on top.

Tasmania, though, being a decently constructed system, is a preferential polling system. As such, Mr Vincent's supporters' votes were transferred to their next preferences. Surprisingly Ms Hiscutt received the least of these, while the most ideologically opposite candidate (Mr Morgan) received the most. This may represent a preference for Independents in the Legislative Council (the other independent, Ms Fuller, didn't do too badly out of the preferences.)

Whatever the reason, the fact remains that my predictions were based on a flawed premise. True, Ms Hiscutt was a strong contender either way. However, given that my predictive methodology went astray in the first round, it is understandable, I feel, that I was not confident enough to put my prediction for Montgomery outside the zone of a tossup.

Coming Up:

Tuesday next week is the second Tuesday in May, and therefore Budget Night. That may or may not steer my content for next week. We'll all have to wait and see.

No comments:

Post a Comment