The Poll Bludger made an observation that intrigued me. While the Internet psephologists (Peter Brent at Mumble, William Bowe at the Poll Bludger, Charles Richardson at Crikey, and me) all predicted a landslide to Labor, the newspaper and territory pundits generally did not. This got me thinking, what was it that gave it away (for me at least).
We will begin with my general rules of thumb.
First, as a general principle of Australian politics, oppositions do not win; governments lose. Governments lose when the economy turns sour, or when ministers are found to be corrupt, grossly incompetent or excessively arrogant. None of these traits were particularly evident beyond the usual run-of-the-mill glitches and mistakes. Martin had steered a moderately conservative Labor government that had not alienated itself from the mortgage belt voters, nor the business sector. The economy was strong. Unemployment was down. The budgets had been balanced. The party did not appear to be internally fractured or at war with itself. When I scoured the Territory papers, I could not detect an overwhelming mood for change. (For the record, the mechanism at work here is a conservative electorate that follows the maxim: if it ain’t broke don’t fix it).
Second — another general principle — governments which win their first term by the barest majority, and provided they have not contravened the first principle above, often go on to a win a more substantial majority in their second term. This was particularly likely to be the case in the Northern Territory, where Labor had been in opposition for so long because it had been successfully demonised by the CLP. In government, Labor was able to bust the CLP mythology. The electorate having kissed the frog discovered it was a prince. The bogey man was no longer.
The third principle, which may well be a corollary of the first, is the theory of electoral cycles. Once elected, governments tend to serve at least two and often three or more terms. A one term government is a rarity in Australian politics. It would have been a heroic achievement in the circumstances for the CLP to wrest government from Labor after only one term.
These principles were supported by the hard data.
My fourth factor was the wisdom of crowds. The betting market opened with Labor on $1.33 and the Country Liberals on $3.00, giving implied win probabilities of 69 and 31 per cent. Within a week of the election the market had Labor on $1.20 (77 per cent) and the CLP on $4.00 (22 per cent). It was on the basis of the betting data that I made my first prediction on 8 June 2005. I took the market to be factoring in more than the bare one seat majority Labor had prior to the recent election. My first guesstimate on the basis of this data was that Labor would win between 15 and 17 seats, up on the 13 it had held.
I was not convinced by the argument that Territorians were betting on the CLP while big interstate money was going on Labor. After all, money is money. And I assumed all of the punters were making informed bets from their perspective.
Fifth, the Newspoll of the 12 Darwin seats conducted some 10 days before the poll, suggested a swing to the government of 9.5 per cent. I acknowledged the following caveats with this poll.
However some caveats need to be noted at this point. First, the poll results came from a survey of 1200 people in 12 electorates in Darwin. This Darwin swing may not be extrapolated to the rest of the Territory. Second, the NT electorates are very small - around 4500 voters each. This means the ‘law of large numbers’ is less applicable, and consequently random results are more likely. Third, Nelson and Braitling are held by independents and may not be as affected by Territory-wide swings to or from Labor. Fourth, and we all know this, swings are never uniform - even within the 12 surveyed electorates.
Nonetheless at this point two things were evident. If (as I observed here) the election was to be won or lost in the northern suburbs of Darwin, short of a cataclysmic disaster, Labor would win it. And, while the swing in the country seats might not be as big as the Darwin swing, it was highly unlikely to be a swing against the government. Consequently, on the basis of the Newspoll, and mapping it against both Antony Green’s pendulum and my own, I decided to increase my prediction from 15-17 seats to 16-18 seats on 11 June 2005.
At drinks on the night before the election, a few of my friends got stuck into me for being overly optimistic. I was told, never would Labor command such a majority in the Territory. While the resulting nervousness is reflected in my post at 7am on Saturday 18 June 2005, I decided to stick with my prediction of 16-18 seats. Both the principles and the data seemed firm.
And as it turns out, I was right: a big swing to the government. But at a potential 19 seats, it looks like a bigger swing than even the Internet psephologists (self included) had anticipated.