jackman.stanford.edu/blog
bannerImage

“Rudd would offer Labor electoral gold”

Sunday December 3, 2006

Filed under: Australian Politics,statistics — jackman @ 9:12 am

The Sydney Morning Herald’s political editor Peter Hartcher has a story titled “Rudd would offer party electoral gold“. A Labor Party led by Rudd would win 48% of first preference votes, according to a poll conducted by ACNielsen for the Herald; under Beazley the corresponding vote share is 41%. The same poll says Labor’s share of the two-party preferred vote under Rudd would be the same as that under Beazley, 56 percent. I have a couple of reactions.

First, to get a sense of the political stakes, consider the following five “best ever” actual election results for Labor, at least in the “modern” post-1949 period, for which two-party preferred results are available and more or less meaningful. I present both two-party preferred and first preference results:

 

Year

Labor Two-Party Preferred (%)

Year

Labor First Preferences (%)

1.

1983

53.2

1954

50.1

2.

1972

52.7

1972

49.6

3.

1984

51.8

1983

49.5

4.

1974

51.7

1974

49.3

5.

1993

51.4

1961

48.0

The point here is that if Labor won 56% of the two-party preferred vote, such a result would constitute Labor’s greatest electoral victory ever, and by a long way. In fact, it would be the 2nd largest 2PP victory in post-1949 Australian electoral history, surpassed only by the 56.9% of the 2PP won by the Coalition in 1966, and on a par with the Coalition’s win in 1975. If Labor won 48% of first preferences, that would be less epic, since Labor has won at least that much first preference vote share on 5 other occasions (1954, 1972, 1983, 1974, and 1961); nonetheless, the last time Labor won that much first preference vote share was 1983 (49.5%), closely followed by 1984 (47.6%).

All this leads me to think that the poll numbers are unrealistic. Does anyone really think the Coalition is “on the nose” with the electorate to the extent that we’re looking at 1983-type result? Now, to give the poll its due: I am sure the question they ask is “if an election were held tomorrow…”, which invites respondents to engage is some measure of projection, if not whimsey. The poll is also subject to uncertainty due to the vagaries of random sampling, but according to Hartcher the poll has a “margin of error” of “2.6 per cent” (i.e., 2.6 percentage points). See the graph below, with the bell-curve corresponding to the uncertainty around the poll estimate (given its stated sample size), the red region indicating the 95% confidence interval around the estimate of 56%; the tick marks indicate historical election results, with some of Labor’s better outings labeled, along with the result from 2004. The poll’s estimate of 56% implies that ACNielsen (and by extension, Hartcher and the Herald) believe that if an election were held tomorrow with Rudd as leader, the 2PP result for Labor would surpass the record 1983 result (53.2%) with probability .86. Again, this strikes me as implausible.

Second, the part of the poll result that I do find more plausible is the differential between Rudd and Beazley: there the poll is reporting a 7 point difference in first preference vote share. That also strikes me as being on the high side, but not as egregiously biased as the estimates of Labor 2PP vote share. But, curiously, even when Labor under Rudd outpolls Labor under Beazley 48 to 41 on first preferences, we wind up at the same two-party preferred number of 56. If true, that is also a remarkable result. Think about it. Is it the case that Rudd pulls in first preferences from both the left (e.g., Green voters who otherwise direct preferences to Labor at high rates) and from more centrist voters who would otherwise vote for the Coalition? Seems odd. Anyway, to say the least, it would be helpful to see a complete breakdown of vote shares (and estimated/assumed preference flows) under the two leadership scenarios.

So call me skeptical…

Comments (12)

Glenn Milne

Friday December 1, 2006

Filed under: Australian Politics — jackman @ 1:50 pm

How could I not link to this? Nothing like tonk in a tux…

The holiday season in Australia coincides with the start of summer, and people in all walks of life — but especially politics and the media? — take to the piss with gusto at various end-of-year events. A few years ago with it was Andrew Bartlett going troppo in the Senate chamber when a Liberal party senator threatened to do him for a prank involving stealing booze from a Christmas party; now its Glenn Milne going apeshit over Crikey’s Stephen Mayne getting podium time at the Walkleys.

It would seem that some people just don’t their limitations re piss and warmer weather in Australia. That said, we’ll see what next week brings for yours truly?

As my Dad would say, “If thats what the piss does to you, you should give it away…”

Comments Off

Powered by WordPress

Bad Behavior has blocked 3998 access attempts in the last 7 days.