A dog's life: Aussie Blue Heeler might chase Labor's Ground Hog away

We knew that Simon Crean had trouble with his public profile when "Anthony9" from Kurrajong said "Crean is a creep" (Rtf 89 kb). He wasn't talking about last Monday's contest between Beazley and Crean - this was 17 months ago during the last federal election. The Labor Party had been running negative advertising based on the proposition that if John Howard was not going to serve a full term, then that election was really a contest between Costello and Beazley. The group we were interviewing were soft-centre voters.

What these voters told us was that not only did they prefer Howard to Beazley but they preferred (a) Costello to both Beazley and Crean (despite leaked Liberal Party polling saying Costello was unelectable), and (b) Beazley to Crean emphatically. This was confirmed by other groups, no matter what their predominant voting intention.

A negative public view of Crean as against Beazley seemed to also be confirmed by the polls which were published (in the case of Newspoll and Morgan) or alluded to (in the case of internal ALP polling) in the run-up to last week's leadership contest. Does this mean that the Crean and Beazley head-butt was the "minor premiership" with neither Labor leader competitive against Costello, let alone Howard?

Only if you believe the professional political commentariat and the pollsters, and if you catch a sarcastic edge in my phrasing, it is because I don't. I can't comment on the internal ALP polling because I haven't seen it but as a tool of political prediction the questions asked by Newspoll and Morgans were close to useless. They centred around which leader interviewees "preferred" or even worse which leader might make Labor "more likely to win an election".

The first type of question misses the point. If Simon and Kim were pitching competitive nights out on the town with the boys, which one you prefer to lead the party (and I'm not talking political party here) might be significant. But it isn't when trying to work out what might happen in an election. That is because electors rarely vote along personality lines in an election. This question is out of context. The correct question to ask is "if X leads the Labor Party are you more or less likely to vote Labor". That way voters weigh up their leadership preferences in the context of the other issues that decide their vote (and the fact that they dislike all politicians almost equally).

When we asked the better question (Excel 35 kb) we found that 28 per cent of our sample thought they were more likely to vote Labor with Beazley as leader, while 32 per cent were less likely. Partly, this was a function of the over-representation of Greens voters in the poll, but not wholly. When we asked voters what would make them hesitate before voting for Crean Labor or Beazley Labor only 8 per cent (Beazley) to 9 per cent (Crean) nominated personality issues. A further 15 per cent said that they thought Crean was ineffective while 6 per cent thought Beazley ineffective. The rest concentrated on things to do with policy and with the Labor Party itself.

The pollsters should have been on to this one in 2003. Their polling of the popularities of the leaders might have had Beazley well in front but both Newspoll and Morgan were showing Labor under Crean with a vote at least as good as the last election. Past polls also suggest that leadership changes fix nothing. Did Kerry Chikarovski really do any better than Peter Collins would have (NSW 1999), or Robert Doyle than Denis Napthine (Vic 2002)? Our research into the last NSW election showed that one problem the Liberal Party had was that it had changed leader too often.

The second question - "who is more likely to win" - is an irrelevancy. The question asks voters to predict what other voters will do. This measures the voters' expectations, which is a legitimate thing to measure, but it is only as good at predicting who is more likely to win as voters' expertise in predicting other people's behaviour. This is generally not very good. Most of us have trouble predicting what we ourselves will do, let alone our neighbours - that's why there is a good living for psychiatrists. The whole point of public-opinion polling is not to rely on this sort of speculation but to build up a profile of what voters will actually do by asking them individually about their own intentions. In fact, those of us who understand political polling often poll for expectations as a way of designing a strategy to frustrate the result most voters predict!

So, if personality is not the major issue for the Labor Party, what is? Refugee issues and national security are important but not in the way it is discussed in the media. Labor didn't lose the last Federal election because most people voted against its refugee policy but because of perceptions connected to the way it was formulated; and they certainly didn't lose because the government used the "children overboard" controversy to inflame the debate. The reason they lost was two-fold.

In the first place, voters trusted them less than the government. As a result, while voters warmed to Labor's domestic agenda they had reservations as to whether they would deliver. This distrust was from voters on both sides of the spectrum and arose primarily because Labor initially had a poorly articulated and variable refugee policy, and were subsequently seen to mimic the government in an attempt to minimise the possibility of fall-out. Left-wing voters thought that Labor was too close to the government and didn't trust them to deliver what they promised. This amplified a feeling of betrayal that they have had ever since Hawke's election in 1983 when he and Keating embraced economic rationalism, despite having won on a campaign against it. Right-wing voters by contrast distrusted the ALP because they thought it would give in to the left and not stick to Howard's refugee stance.

Second, the Labor Party spent too much of its time talking about issues that were on the government's agenda. In a way "children overboard" did do for the ALP. Morgan's polling (look at the graph) had the ALP coming back strongly in the last week of the election. This was probably a reflection of the failure of the government's domestic-policy agenda and the poor reception of its campaign launch. This was all undone when Labor decided to spend a large part of the last week delving into rumours that no children had actually been thrown overboard. The general public don't care whether children were thrown overboard; they do care how many refugees come to Australia. By raising the issue the ALP distracted media attention from its advantage on domestic issues at the same time as it raised doubts in electors' minds about its own trustworthiness by talking about issues in the area where it was most suspect.

So, how does the ALP go about winning the next election? Here are a few bullet points.