Numbers crunched

Ignore me. Former Labor Senator John Black’s number crunching in yesterday’s (print only) Courier Mail is far beyond my powers of sophistication, being based on modelling from marketing company Australian Development Strategies. Apologies to Black for the following acts of larceny, and also for the removal of his Monty Python references. Let’s start with his view of the big picture:

We start in 1998, when the good old boys from One Nation came along. Peter Beattie lucked out, and won office with just 39 per cent, losing about 4 per cent of his unskilled blue-collar vote to One Nation. That election saw One Nation poll 22.7 per cent and split the conservative base as it stole rusted-on National Party farmers (about 11 per cent) and Liberal Party small business types (about 7 per cent). De facto first-past-the-post voting created carnage in Opposition ranks.
In 2001, the One Nation vote collapsed, with the combined One Nation/City Country Alliance vote falling by 11.6 per cent, while Labor’s vote rose by about 10 per cent, and the Liberals also fell again – by about 2 per cent. Most observers reasonably concluded that Labor picked up some Liberal voters and the vast majority of disaffected One Nation voters. However, the ADS demographic profiling, which compared Labor’s 2004 state and federal votes, shows Labor’s inflated state majority in 2001 and 2004 coming from former Liberal voters, not One Nation. These Liberals – in outer urban marginals – flooded across to state Labor in 2001 and stayed in 2004, with One Nation voters taking their place in the Liberal pile.

The profiling also shows that Queensland One Nation voters taken from state Labor in 1998 – older, poorer, unskilled, blue-collar workers – have returned to federal Labor candidates in Queensland, but not to state Labor, preferring the state Liberals. Presumably state Labor’s market research confirms this analysis, which explains the recent parliamentary stunts from Beattie about the Howard Government’s new industrial relations

ADS research shows the new Smart State Labor voters are indeed smarter, which contains a large element of risk for Beattie. When compared with Queensland’s federal Labor voters, they are more likely to be female, younger, better educated, well paid, white-collar workers, typically para-professionals, personal assistants and receptionists. If and when these Liberals decide the state Liberals are worth voting for, as they did with Campbell Newman at the Brisbane mayoral ballot, the electoral tide will leave Team Beattie well and truly stranded.

The only part of this that sounded jarring to these ears was the notion that an identifiable demographic group moved from Labor to One Nation to Liberal at the four state elections between 1995 and 2004, given that the Liberals went from 14 seats to five in that time. But the primary vote figures do show that the Liberal vote was not spectacularly lower in 2004 (18.5 per cent) than 1995 (22.7 per cent), and that the real disaster over that time has been the Nationals’ progress from 26.2 per cent to 17.0 per cent. Closer examination reveals Labor’s safest seats, home to the blue-collar One Nation demographic referred to by Black, were not caught up in the tidal shift towards Labor from 1995 to 2004.

The following scatter plot shows the change in Labor’s vote between 1995 and 2004 in seats contested by the Liberals in 2004, with the Y axis marking Labor’s primary vote from 1995. Seats where independents accounted for more than 10 per cent in either election are excluded, and no allowance is made for redistributions. The Pearson correlation is -0.43, which I understand to be pretty significant as these things go.

No similar pattern can be discerned if we view seats contested by the Nationals, until we introduce a bit of colour coding to mark the region of each electorate. If you draw a nice fine line between 5.7 and 6.1 on the X axis, you get a zone of Labor underachievement to the left which includes all 10 rural and remote electorates and only six from elsewhere. The 11 city, town and coastal seats to the right stand as testimony to the decline of the Nationals’ saleability in urban Queensland, and all should be left to the Liberals in future. Note that the urban outlier that swung away from Labor, the seat of Logan, is the safest Labor seat in the sample, which may be regarded as consistent with the trend identified in the first chart.

Having drawn the big picture, John Black moves on to the business at hand:

Which leads us to the real state Labor votes in Chatsworth and Redcliffe. One of the benefits of demographic profiling in the Smart State is that researchers can benchmark the party votes in each state seat, thereby measuring each candidate’s personal vote … In Chatsworth, the personal vote for former treasurer Terry Mackenroth was a respectable 5.7 per cent, while the personal vote for Ray Hollis in Redcliffe was 3.9 per cent. This means that the Chatsworth primary starting vote for Labor’s Chris Forrester, will be about 50.6 per cent, which will have to be adjusted up for a likely decline in the Green vote, but down for a new One Nation candidate. Preferences will take this starting ALP preferred vote up to about 55 per cent, so that, if Caltabiano is able to transfer his notional 6 per cent plus personal council vote in corresponding booths in Chatsworth over to his new seat, he should need an additional party-based swing of about 2 per cent to win easily. If he can’t get this party swing on election night, obviously the voters will have concluded that despite the Team Beattie track record on hospitals, roads and schools, the state Liberal Party, with Caltabiano as president, is even more lamentable.

In Redcliffe, Labor’s Lillian van Litsenburg should start with a primary vote well below 50 per cent, with an additional four candidates in the field, leaving her with a notional preferred vote starting point, of about 53 per cent. The additional minor party candidates, varying rates of vote exhaustion and a poor by-election turnout of Government voters, mean that van Litsenburg would lose to Liberal Terry Rogers, on these factors alone. This could make for a long count and not necessarily mean Redcliffe is a good pointer to the future of Team Beattie’s large majority, as the seat is also "old Labor" and politically stable. Chatsworth, on the other hand, is chock-a-block with the sort of Liberal yuppies who have been voting for Liberals John Howard and Campbell Newman – but against the Queensland state Liberals – in that ring of state seats outside Brisbane, but within the great

If the Labor primary vote in Chatsworth (after you have taken off Mackenroth’s 5.7 per cent) stays above 42 per cent tomorrow night, then Team Beattie will probably be back in George Street in 2007, whatever the spin from a Liberal machine that has elevated self-harm to an art form … However, if feedback from Chatsworth ALP doorknockers and Newspoll is any guide, Sideshow Pete is in big trouble, as voters in the Smart State seem smart enough to realise the truth isn’t all it seems in Sideshow Alley.

The Poll Bludger cannot fault Black’s analysis except to point out that he provided similar commentary before the federal election and did not do particularly well, tipping only the narrowest of Coalition majorities. I too underestimated the outcome, and my defence could be used by Black as well – we got the relativities right, but underestimated an underlying across-the-board shift to the Coalition (I like to think that the Latham-Howard election eve handshake accounted for a proportion of this). This time around, with no confusion between national and local factors, Black’s number crunching carries more weight. He has succeeded in further weakening the Poll Bludger’s confidence in his prediction of Labor victories, but nevertheless they still stand.

Late news: The Courier Mail reports Labor sources expect to lose Chatsworth but are "hopeful of holding Redcliffe in a tight contest".

Author: William Bowe

William Bowe is a Perth-based election analyst and occasional teacher of political science. His blog, The Poll Bludger, has existed in one form or another since 2004, and is one of the most heavily trafficked websites on Australian politics.