Senate Predictions: The Grid
In the week leading up to the election, I want to focus on the fun part: trying to predict the Senate. It is hard for two reasons. Minor party preference deals make even the most outrageous outcome a dim possibility, and even the most sensible looking prediction, less likely than it first appears. And the easier, but still difficult issue of predicting the major party vote, whose constitution has the biggest effect on the final result.
To that end, and with the help of some useful Senate vote archives I have graphed the changes in the two key variables that decide the Senate campaign:
1) The ALP Proportion of the Major party vote. This is different to a two-party preferred, because in the Senate the primary vote of the major parties may be (almost) all they have.
Senate primaries tend to be slightly lower than their House of Reps. equivalents because of fewer parties, and the Democrats successful repositioning of the Upper House as one of review. However, with the exception of Tasmania and ACT, the ALP proportion of the major party vote tends to be the same in both houses - that is, people switching to minor parties tend to come from both sides. With a few minor adjustments we can therefore use the primary votes recorded in the polls to predict Senate outcomes.
2) The Total Major Party Vote. This varies around 80% which just happens to be where the greatest uncertainty in predictions lies. Nevertheless, there are a few rules (to be outlined in a forthcoming post, that should give some guidance.
By plotting these two values, it is possible to show clearly the true state of the Senate. To help though, some guidelines on the points that matter can also be plotted.
The most important two lines are the solid blue and red lines curving diagonally across the grid. They represent the point where Labor (red) and Liberal (blue) pass 3 quotas in their own right. This splits the graph into four areas.
The upper triangle - Marked 3 Lib 3 Lab, as both parties have 3 quotas. For obvious reasons, this is the least interesting area of the grid. Fortunately only two recent (since 1980) Senate elections have points in there, one before there were 6 seats (WA: 1987) and one shortly after (Vic: 1993).
The Side diamonds - Of recent times, (particularly 2004 when every Senate race was in the left-hand diamond on the grid), the area marked "3 Lib 2 Lab" has predominated. Here, the Liberals have 3 seats already, and are therefore feeding other parties their surplus. Labor's position between the two solid red lines indicates the strength of their surplus. Toward the upper-right is strong, and vice versa. This zone has its inverse on the Labor side, though (for this election) the Greens will be heavily favoured in any contest here, as the ALP surplus will move directly to them.
There is one leftover seat in this area, and therefore, two quotas of votes. What that means, is that at pivotal points in the count, certain vote percentages are needed for the smaller of the two major parties to continue.
To come 3rd (or better) of the last 4, a candidate needs 50% of a quota (marked as the lighter of the dotted lines). Coming third is of no help unless that fourth place candidate preferences them, so most times a major party in this area remains vulnerable to the minor party vote. Just how vulnerable we will get to.
To come 2nd (or better) of the last 3, a candidate needs 66% of a quota (marked with the heavier dotted line). It is possible to come second with less than this - even going as low as 50% - but as the major parties tend not to receive preferences until late in the count, they must get near this line. Obviously, they then need to swallow the preferences of the third candidate, but that is a more complex problem.
The Lower Triangle - This is actually a diamond too, but no Senate race has ever got close to leaving it. As in the previous areas, the dotted lines show the 50% and 66% marks for a quota. Unlike those, they are not a useful guide. towards the eventual outcome As there are not two quotas, but three remaining, guaranteeing third place (of four) means getting 75% of a quota (marked by a dashed line), though as little as 50% may suffice to eventually win. In the past, 75% has usually got a major party a seat - it means that they should surpass the minor parties that preference them. However, a number of three-cornered scenarios exist where Labor, the Greens and the Liberals all surpass 75%, and the final two seats are essentially a lottery.
Individual minor parties are not represented on the grid, however, a rough guide to what they need to do can be gained from looking at the three horizontal green lines. The bottom line represents the level of total minor party support required for a minor party to win with 50% of minor party preferences. The middle line 66%, and the top, 85%. Note that, in the side diamonds, the minor parties are supplemented by Liberal or Labor surplus, and the total minor party vote required drops.
In the lower triangle, it should be possible (for a given level of minor party and ALP support) to examine preference flows to see whether a minor party is able to a) pass one or more of the major parties and b) put together a complete quota. In the next post - tomorrow - I'll look at what level that minor party vote share is likely to be.
20th November, 2007 02:51:02