Marginals
factor is marginal
Peter
Brent
In
the face of stubbornly lopsided opinion polls, federal
election-watchers determined to construct that nail-biting finish
have recently latched onto a tantalising narrative.
It
is that while the published national surveys show the Labor
opposition a mile ahead, the individual seat battle may well be
different. Specifically, that the Howard government might survive
this year’s election with significantly less than 50 percent of
the two party preferred vote.
There
are certainly precedents for minority vote wins – five since the
introduction of preferential voting in 1918, most recently in
1998.
In
each case the incumbent won, which tells us something both about
governments’ ability to target seats, and the types of voters
who tend to live in the marginals - risk
-averse
and difficult to shift.
So
the ALP will probably need more than 50% support to win. But how
much more?
A
related exercise focuses on the Mackerras pendulum, according to
which Labor needs a uniform swing of 4.8 percent to get a majority
in the House of Representatives. The polls currently show almost
double that, but if the government does well in the marginal
seats, the story goes, 4.8 might not be enough.
The
pendulum is a wonderful visual aid, but its predictive record is
mixed. It worked well in 2004 but not at the two previous outings.
Yes,
the ALP might need a significantly larger swing than 4.8 percent
to get over the line. On the other hand, they might need
significantly less.
To
suggest something
"might"
happen, and not back it with decent reasoning, is simply wishful
thinking. And there is no reason to believe 4.8 percent will not
be enough for Labor.
Yes,
governments can generally extract maximum value from the marginals.
But after three re-elections, the current margins reflect this.
That
is, Labor’s vote-seat disadvantage is already built into the
"required
swing".
At the last election they got 47.3 percent, which added to 4.8
percent comes to 52.1 percent.
That’s
a high number, and to suggest it understates Labor’s task
stretches credibility.
If
anything, it is too high. It is simply based on the way votes
happened to fall one Saturday in 2004. Each election is different.
For
example, if we had elections every six years instead of three, and
so were basing today’s pendulum on the 2001 result (but still
adjusting for subsequent redistributions), we would instead be
saying today that Labor needs only 50.7 percent to win in 2007.
Which
is not to suggest this is their
"real"
target, but instead to show how arbitrary these on-paper required
swings are.
So
what will Labor need to win? That’s in the lap of the voters. It
could be anything from, say, 49 percent to 53 percent. But a stab
in the dark might be that 50 probably won’t get them there, but
51 should.
That’s
just a guess, though.
Peter
Brent is publisher of mumble.com.au and a member of the ANU’s
Democratic Audit
= premium | Logout 
