The ABS released some 2006 Census data today. I was hoping to use this to discover how many adult citizens there are/were (i.e., the size of the eligible electorate), but unlike in previous censuses, they are only reporting the total number of citizens (adults plus children); see the thumbnail below. Maybe this will become available in the Community Profiles, due for “First Release” in August 2007.

I suppose I could get the data by cross-tabulating the age and citizenship variables when the data when it is released, but that seems an awful lot of work to learn just one number: the number of Australian citizens aged 18 or over (and hence obliged to enrol and turnout to vote). Why is such a basic fact about the country not available from ABS or AEC for that matter…?

Any non-Australian readers, start here. The report is here.

A proposition: its always political, always, and if it isn’t, it had better be. Thats what being leader of a political party is about. So lets get past that from the get-go.

And as my sometime co-author Mumble says, it might or might not be good policy (like Mumble, I don’t know). But I do know that the conditions of Aboriginal Australians are beyond scandalous (and have been for a long, long time), and that “something has to be done”. So something is being done. I heard an interview with Noel Pearson on PM last night. Pearson was just magnificent (as he has been in a series of media appearances I’ve seen online in the last couple of months; example); in the PM interview he basically told all the critics to bugger off. Pearson’s responses to the interviewer powerfully demonstrated that since this policy intervention is widely understood as being about the immediate protection of children, it then carries such overwhelming moral suasion that any opposition will be swept aside (at least for the time being); see also Noel’s “Your Say” in the Weekend Australian. And so we’re only left to ask the question “why now?”.

I’m not as cynical as some on this. First, I just don’t see this issue as a big vote-switcher. Big-stick Federal intervention will generate criticism on the left of various sorts (eg., see these reactions from the Greens, the Democrats; ACT Chief Minister Jon Stanhope has called the proposals racist, inter alia, et alia), but the people sympathetic to these criticisms aren’t likely vote switchers. Recalling Parliament, passing the legislation, and the implementation of the policy: these events will provide a huge circuit-breaker on the dominant political narrative of the last few months (Labor’s huge lead in the polls, industrial relations, responses to climate-change), and the Prime Minister gets to be very, well, “prime-ministerial” (Richard Farmer’s Crikey contribution on this aspect of it struck me as spot on, underscoring the power of incumbency, the ability to abruptly change the terms of political debate).

The interesting political question is does an aggressive Federal policy intervention into Aboriginal affairs switch votes in the middle? Will this policy intervention make anyone who has voted Liberal in the last couple of cycles, but is saying they will vote Labor this time around, come back to the Libs? And, in particular, these are people in marginal electoral divisions like, say, Parramatta, Lindsay, Eden-Monaro, Bonner, Moreton, Kingston, Bass… I doubt it. Frankly, this could well be an issue that improves Coalition support in hitherto safe Liberal seats in leafy suburbs (insert “doctors’ wives” comment here): e.g., North Sydney, Bradfield, Bennelong (!), Wentworth.

One other thought: while “it is all political”, don’t discount the possibility that the Prime Minister is also looking to the history books, given that his government remains odds-on to lose the next Federal election. Having presided over a period of great prosperity, perhaps Howard has got an eye on what future historians will say about what his government did for the most impoverished and vulnerable in Australian society (imagine superimposing two time series: Sydney property values, the ratio of white-to-Aboriginal life expectancies, 1996-2007). The roll-out of the policy as Australia comes under some international attention with APEC will probably also play well, all things considered. In addition, as Howard came to this decision, while he is perhaps first and foremost a political leader, I wonder also about the personal underpinnings: voices from inside his household (Jeanette?); his references to the “passion” of his Cabinet and the relevant minister, Mal Brough, on this issue; memories of an Earlwood Methodist childhood… as he contemplates/confronts the end of his tenure as prime-minister?

My paper with Peter Brent on voter enrollment in Australia just appeared as part of a series of papers published by the Democratic Audit of Australia, a project run out of the Australian National University. Our story is fairly simple: the Australian Electoral Commission has become more proficient at expunging voters from the roll while at the same time changes in legislation have made it tougher to get on and stay on the roll. I reproduce Figure 1 of the paper here (thumbnail below, click for larger version), which shows the relationship between Australia’s ERP (estimated resident population) and the size of the electoral roll. We normalized the two series at 100 around the time of the 2004 election, and it is pretty clear that recent enrollment numbers are lagging behind where they ought to be given the historical relationship with ERP.

Caveat/Gripe: ERP is hardly perfect as a (relative) baseline for assessing under/over-enrollment. That said, unless the age composition of the citizenry and/or the number of immigrants have dramatically changed since the 2004 election, the normalization of enrolments to ERP should have decent over-time validity. What we’d really like is to know the number of eligible voters (basically, adult citizens), but the Australian Bureau of Statistics doesn’t publish such a time series, although the AEC does appear to request and obtain such a number from the ABS when AEC conducts internal audits of how well Continuous Roll Update (CRU) is faring. ABS does report the number of adult citizens with each Census, and we’re expecting a June 2006 number next week with the release of the 2006 Census, so expect a quick, minor update of the paper.

Re this figure: vertical lines indicate the timing of state and Federal elections.

There is a thread doing the rounds (from the PM himself, and echoed by various media commentators) that we can expect Labor’s lead in the polls to remain there until the election is formally announced. Perhaps. For the record, I dug out a graph that accompanies my article on polling in the run-up to the 2004 election (“Pooling the Polls Over an Election Campaign”, Australian Journal of Political Science, 2005 V40(4):499-517). Full details are in the article, but basically the story is this: to the extent there was any perceptible movement in the published media polls in 2004, it was the hint of a movement towards the government around the time that the election was announced.

The graph (thumbnail below; click for a larger version) shows the estimated daily trend for Coalition 2PP over time (orange line), with the gray region corresponding to a day-by-day 95% confidence interval (the “margin-of-error”). One of the “tricks” in the article is that the daily track is constrained to run through the actual Coalition 2PP result observed on Election Day. Individual polls are show with the numbers in circles, and, with the benefit of hindsight (i.e., doing the statistical modeling after the election result is known) we can see that Nielsen and Galaxy were closer to the actual result in 2004 than Newspoll, Morgan, or the on-line polling effort promoted by ninemsn (more about that later). Collectively, Nielsen/Galaxy/Newspoll (all phone polls) did not appear to systematically over/under-estimate Coalition 2PP. But what is striking is that with the benefit of hindsight we can see that lots of the polls in the run-up to the 2004 lay outside of where we the actual trend was, plus margin of error (i.e., lots of circles not in the gray region, and plenty below the gray region).

Re-running this model both in real time this go around and ex post will be a very interesting exercise.

Safari 3 breaks Sogudi, or vice-versa (with Sogudi enabled, I can’t type into the Safari 3 Google search bar, nor can I call up URLs via direct entry into the address bar). Disabling Sogudi got back manual URL input, but the Google search window remained dead. Sob. I reverted back to Safari 2 on my office machine, and have clobbered Sogudi on my laptop, just to see how I fare in either case. The Sogudi developer is aware of the issue, so here’s hoping for a speedy resolution.

In a momentary lull in the polling, the Galaxy poll for Queensland has attracted a ton of press coverage (see Mumble for some links). Labor on 52% 2PP in Queensland (down 3) has been given all kinds of meaning re a trend back towards the government: we simply can’t say without more data.

Here is yet another interpretation: suppose Labor did win 52% of the 2PP in Queensland. According to AEC records, that would constitute the best 2PP result for Labor on the books (at least since 1949, when 2PP started to be tabulated/estimated). The best Labor has ever done in QLD is 50.7% (1961, Killen “the magnificent”, and all that), and even in the big Labor wins of 1983 and 1972 Labor won just 49.5% 2PP in QLD both times (versus a national 2PP of 53.2% in 1983 and 52.7% in 1972). That is, if Labor is north of 50% 2PP in QLD, I’d reckon its over (depending on what you think of the homestate effect for Rudd, but try talking to Bill Hayden about that).

Back on June 3, 2007, Suzanne Gibson from ABC’s Radio National contacted me, inter alia:

Simon

[...]

I’m currently researching a radio program for ABC Radio National on political star recruits. I’m not sure how closely you follow events here in Australia but the ALP in particular has recruited a range of well known faces for the up coming Federal Election.

I’m just wondering if anyone has done much work on whether celebrity or “star” candidates really make any difference to the way people vote?

Is that something you’ve done any work on?

My reply appears “below the fold”.
(more…)

The Australian Electoral Commission has a new look to their web site. I give it much better style points than the previous site, but it seems that they’ve made it harder to to dig down to get historical enrolment data. Only the most recent monthly release is shown here. They used to have monthly data by electoral division going all the way back to mid 2002 on the old site (which I have safely archived, and have used to make pictures like the following, click on the thumbnail)…