One of the good things about
doing political commentary is that you can recycle your stories:
similar circumstances keep recurring, and most observers failed to
learn the lesson the first time. Eighteen months ago the ALP
was consumed by the leadership issue. Here’s what I wrote then
(Crikey, 27 January 2005):

Opinion polls published over the last week or so … consistently show:

(a) that Beazley is the preferred leadership candidate of both Labor voters and people in general, and

(b)
that when paired against either John Howard or Peter Costello as
preferred prime minister, Beazley does better than any of his
leadership rivals, although he still trails Howard by a considerable
margin.

The interesting thing about these results is that
neither matters one iota in an election. Both approval ratings and
preferred prime minister (or premier) numbers have a notoriously bad
record at predicting actual election results. What matters is which
party people vote for, not what they think about the leaders.

This
year we’re talking about the Liberal leadership, but otherwise nothing
has changed. The polls still produce results that are less meaningful
than they seem, and none of them ask the question that matters: what
effect a leadership change would have on people’s voting behaviour.
First Morgan, then AC Nielsen, and today Newspoll.

Dennis Shanahan, commenting on today’s poll,
says “if it were a Costello-Beazley contest, the ALP’s chances would be
much better.” That may well be true, but we can’t really tell from the
poll. We do know that a significant number of those who tell the
pollsters they prefer Howard to Beazley also say they are planning to
vote Labor. If that group includes the people who prefer Howard to
Beazley but Beazley to Costello, then the latter preference is
presumably not going to change their voting intention.

It
wouldn’t be difficult for one of the polls to ask people which party
they would vote for if Costello were Liberal leader, and it would be
much more informative.