Senate Predictions: Plotting Minor Party Shifts
Yesterday I showed how to plot Senate outcomes in terms of two variables: The ALP's proportion of the major party vote, and the total major party vote (or inversely, the minor party vote). The former, as explained, can be estimated from House of Reps. primary votes. However, Senate polling of minor parties is notoriously unreliable, and the total minor party vote far less predictable. The best bet here is to devise some rules of thumb from past elections, and see where that gets us.
The last time a Liberal government was voted out in 1983, the thrust of movements was directly from Labor to Liberal. This is seen best in the small shifts in SA, Vic, and NSW. There was variation though, in a strong Labor shift in WA, a strong anti-Liberal shift in Qld, and in recalitrant Tasmania a small shift from Labor to Liberal.
Voters reacted unfavourably to a second poll in 1984, with a strong anti-Labor vote. However, in a pattern that will shortly be familiar, this was not pro-Liberal, with no increase in their quota (except in Qld which was slightly pro). Tasmania was different again, shifting strongly pro-Labor to make up for the previous election. South Australia was neutral on ALP proportions, but shifted to minor party votes.
There was little movement in 1987. The four states that have tracked each other to date (NSW, Vic, SA and WA) all shifted slightly to Liberal at the expense of the minor parties. Qld went slightly Labor. Tasmania is again different, mostly because of the strong swing to Brian Harridine when he was running.
The end-game for Labor started in 1990, when the states most strongly affected by recession took their revenge. Vic, SA and WA all voted strongly anti-Labor, but not pro-Liberal. Qld and NSW shifted towards the minor parties (mostly Democrat) without changing their voting proportion. Tasmania continued to move upwards (it started at a low base), effectively shifting their vote to he Liberals.
1993 was a mixed bag. Strong pro-Labor votes in NSW and Vic cost Hewson the election. A general shift away from minor parties resulted in the highest major party vote recorded in most states, and strong Liberal votes in WA, SA and QLD. Tasmania was again different.
There is no clearer demonstration of the baseball bats awaiting Paul Keating than this. Huge swings away from Labor in all states, but little movement towards the Liberals, except in Tasmania that has finally come into line. Interestingly too, while the major party spread is quite wide, the minor party vote, the minor party vote is fairly consistent at around 19-21%.
One Nation and GST combined in 1998 to produced the strongest anti-government push of all the elections. While all states showed a very slight movement to Labor, the smaller (and generally more volatile states (WA, Qld, Tas and SA) all increased the minor party vote by 6% or more. Qld recording the highest seen at 29% (with a 2-2-1-1 Senate result to match).
2001 showed a small correction to the minor party vote in Qld, WA and SA, but interestingly not elsewhere. In Vic, NSW and Tas the movement was away from Labor (or perhaps, from Labor to Green, and Democrat and One Nation to Liberal).
Whereas 2001 was good for the Liberals, 2004 was astonishing. Large movements to them in WA, Vic and Tas, while SA and NSW saw shifts away from the minor parties (mostly Democrats). The strength of the Green vote (at the expense of the ALP) has left the electorate both strongly Liberal, and roughly where the historical minor party vote sits.
To 2007 then and the polls predict is for a large swing to the Labor party in the primary vote - that is, from left to right on the graph. Traditionally, shifts away from the government are also accompanied by a minor party shift (Pearson's -0.593, Sig. 0.000, R-Sq. 0.351).
Putting it into a regression, the 95% confidence interval for the minor party shift is -0.991 to -0.507. Unsurprisingly, given the limited data, the shift loses significance at the state level, being both more variable, and more effected by anomalies. Nevertheless, for an 8% swing a 4-8% swing to the minor parties is a reasonable estimate at first glance.
Whether it is a reasonable estimate depends on the level of anti-government sentiment and the strength of minor party candidates. On the first, there isn't either a strong sense that country is going badly nor a significant level of dissatisfaction with the Howard government.
Similarly, for the minor parties, Morgan polling indicates that their level of support will be roughly similar to the last election. It is hard to believe that Democrats could do worse than in 2004, but their campaign has been weak. Their vote has a strong negative correlation (-0.429) with government swings, meaning they would (normally) pick up a few percent. However, it is hard to see them picking up more than a couple of percent, and are likely to be in the 2.5-4% range.
Family First should improve, having a higher profile than before, and may benefit from any swings against the Liberals or a declining One Nation. An extra percent is not unlikely.
The Greens, by far the largest of the minor parties are an unknown. They have a weak negative correlation with the ALP vote (-0.323) which might translate into a small drop. This is from limited data however, and will likely be offset by gains elsewhere as their profile (and their young voting base) continues to grow. A percent or two increase would not surprise, but nor would a stagnant vote.
All considered. An increase in the minor party vote seems reasonable, but with the exception of South Australia (where Mr X is a massive unknown) the likely swing will be on the low side (60-70% of the ALP swing) rather than the high (80-90%). What this translates into on a state by state basis we will look at in the next few days.
20th November, 2007 12:59:11