Rival psephologists debate the impact of One Nation in Western Australia

Still waiting on WA

Crikey psephologist Charles Richardson wrote the following in the 19 January subscriber email:

Westen Australian premier Geoff Gallop announced this afternoon that he would not be calling a state election this week, meaning that the widely expected date of Saturday 19 February is no longer available. Speculation now shifts to the 26th. Gallop said that state politics had been overshadowed by events at the federal level, and that more time was needed to bring the focus back to local issues.

“I’m very, very keen for this election to be held on its own terms and its own merits, on state issues,” the Australianreports him as saying. “There has been a lot happening in the first month of this year and it has made it very difficult to get the focus on state politics.”

Mr Mumble agrees with my view that when the election is held a Labor win is the most likely outcome (he says “comfortable” win, which is perhaps needlessly bold). But he helps to perpetuate what I think is an urban myth when he says that “The Gallop Government’s fluke win in 2001 was thanks to a high One Nation vote and that party’s policy of directing preferences against sitting members.”

In the Fin Review in August 2001 I described this view as “a fine illustration of political myth-making at work.” Put briefly, it just doesn’t stack up with the figures. Labor won 13 seats from the Coalition at the last election, but in 12 of those it was already ahead before One Nation preferences. Only in Bunbury did they make the difference. As I said then, “Labor would still have won the election comfortably, even if all the One Nation voters had stayed at home.”

But what about the targeting of sitting members? It had very little effect, because One Nation voters don’t follow the how-to-vote cards. (Whether because not enough of them are given out, or because its voters are just ornery, is uncertain.) The average flow of One Nation preferences against sitting members was under 55%; even in Bunbury it was only 61%.

Of course, if you just assume that the One Nation voters would all otherwise have been Coalition voters, then even half the preferences going to Labor represents a big problem for the Coalition (and correspondingly a benefit for it this time, with One Nation defunct). But the evidence is against that assumption: in the Queensland election the same year, for example, the presence or absence of a One Nation candidate made no difference to the two-party-preferred swing to Labor.

For more on WA, see the very informative wrap on the Poll Bludger site.

Antony Green disagrees

This produced the following response the next day from ABC psephologist Antony Green:

I usually agree with Charles Richardson, but I think he has missed a very important point concerning One Nation and the 2001 WA Election.

There is a tendency in Australia for analysts to look at published preference distributions and simply tally the number of seats decided on preferences. The better measure, as Charles has done, is to look at the number of seats where a party trailed on the primary votes but came from behind to win after preferences.

But even this measure misses the point that sometimes in Australia you need to take your eyes away from the after preferences count and have a look at the relative share of primary vote achieved by parties. In non-preferential electoral systems, analysts go to great effort to track the primary vote flows between parties in explaining election results. In Australia, the question is too often reduced to one of two-party preferred swing.

Sure, two-candidate rather than primary totals determines who wins seats. But Charles’s point that Labor only won one seat at the 2001 election from behind on One Nation preferences misses an important point. Labor would not have been ahead in the first place had it not been for the extraordinary loss of Coalition primary vote support to One Nation.

At the 2001 WA election, Labor’s primary vote rose only 1.4%. The Greens rose 2.5%, Liberals for Forests polled 1.6%. But the Democrats fell 2.4%, the Coalition fell a whopping 11.1% and One Nation polled 9.6% at its first election. All this translated into an 8.1% two-party swing to Labor, quite a result given how little Labor’s primary vote rose.

Yet compare this result to 1993 when Labor under Carmen Lawrence was defeated. Labor’s primary vote in 1993 was 37.1%. When you take into account that Labor did not contest three seats in 1993, that means Labor was defeated in 1993 with a higher primary vote than when it won in a landslide in 2001.

The other comparisons over eight years are that the combined Green and Democrat vote rose from 6.6% to 9.9% and the Coalition primary vote fell from 49.4% to 34.4%. Overall, the minor party vote more than doubled between these two elections, from 13.5% in 1993 to 28.3% in 2001. At modern elections, only Queensland in 1998 records a higher level of third party voting.

I believe in relation to the 2001 WA election, the more pertinent question to ask is to what extent the presence of a minor party candidate changed the way voters would otherwise have voted? The two main parties of interest in this question are One Nation and Liberals for Forests. If these candidates had not been on the ballot paper, would the result have been different? Were there voters who voted for One Nation and gave preferences to Labor, who if their had not been a One Nation candidate, would have voted for the Coalition? If this was the case, then One Nation changed the result by acting as a ‘pied piper’, leading voters across the left-right political divide with their preferences.

If you think back to the arrival of Pauline Hanson in 1996, all the talk for two years was how One Nation would deliver Labor votes to the Coalition. The theory was traditional Labor voters would be attracted by the One Nation message, vote for the party and direct preferences to the Coalition, who for two years were less strident in their attacks on Hanson.

Most surveys suggested this was a possibility. Academic surveys following the 1998 election indicated that the income and occupational profile of One Nation voters was more like that of Labor voters than the Coalition voters. From the Coalition’s behaviour through 1997 and 1998, you suspect that its internal polling was suggesting the same thing.

In one sense, the result of the 1998 Queensland election backed this theory. In the nine seats not contested by One Nation, the swing to Labor was 7.1%. In the 45 two-party contests where One Nation contested and had its preferences distributed, the swing was only 4.1%. A three percent difference is a big variation in swing for Australian elections.

This Queensland gap provides some evidence that once a Labor voter had taken the breath-taking step of abandoning their traditional party and voting One Nation, it was psychologically easier to then give preferences to the Coalition. The three percent difference above backs the idea that it was easier for a Labor voter to give a preference to the Coalition than it was for them to give a first preference to the Coalition.

However, as the Howard government realised to its horror that same weekend, this advantage was completely irrelevant. The more important point than One Nation’s preferences was that One Nation had halved the Coalition’s primary vote.

All the talk beforehand by the likes of Senator Bill O’Chee that the Coalition directing preferences to One Nation was irrelevant proved to be wrong. In 22 seats, Coalition preferences were distributed, producing a 3% swing AGAINST Labor, the exact opposite of the 54 seats where One Nation preferences were distributed.

The Howard government learnt from this lesson, very quickly compromising on the contentious Wik legislation then delayed in the Senate. It also switched to talking about tax reform, promised to direct preferences against One Nation, and started to attack Pauline Hanson’s views as being dangerous. No more talk about the veil of political correctness being lifted.

This experience needs to be remembered in analysing the 2001 Western Australia result. After the One Nation breakthrough in Queensland, the 9.6% achieved by One Nation in Western Australia is the second most dramatic arrival of a new political party in decades.

Overwhelmingly these votes came from the Coalition. Those preferences then split evenly, generating roughly two-thirds of the two-party swing against the government. Rather than talk of which seats were decided on preferences, the point of One Nation is that they took away a quarter of the Coalition’s primary vote, of which at best only half came back as preferences.

Charles is correct that only in Bunbury did Labor come from behind to win on One Nation preferences. However, in four other seats, Albany, Geraldton, Mandurah and Roleystone, Labor would not have been ahead during the distribution had the Liberal Party not lost so many votes to One Nation.

Not that I am arguing One Nation was the sole cause of the Court government’s defeat. There were clearly many factors, as I outline in my summary of the 2005 election here. Everything played a part, from the unpopularity of the Howard government, to the mortgage broking scandal in Albany and Alfred Cove, failure to declare pecuniary interests in Geraldton to even the donkey vote in Collie.

The key point in trying to pick the winner of the 2005 Western Australia election is figuring out where the 9.6% who voted for One Nation in 2001 will go this time. A key factor in the election of Geoff Gallop as Premier was the number of ex-Coalition supporters who voted One Nation and gave preferences to Labor. Without a One Nation candidate this time, just how many of those voters will decide to give their first preference to Labor?

I suspect many of them will not. At its simplest, that is the main reason (over-looking the state’s malapportioned electoral boundaries) why Labor has a tough job trying to win re-election in the west.

Charles Richardson responds

By Crikey psephologist Charles Richardson:

Antony Green, as usual, is very thorough in his analysis, and I agree with most of what he says on Crikey. His review of the 1998 Queensland election is particularly valuable. But on the point of disagreement between us I think he fails to make out his case.

He starts by saying that although Labor in 2001 was already ahead before One Nation preferences, it “would not have been ahead in the
first place had it not been for the extraordinary loss of Coalition primary vote support” – OK so far; but then he adds the words “to One
Nation”.

Agreed that many of those Coalition voters swung to One Nation, and many (a bit over half) then gave preferences to Labor. The question
is, would they have deserted the Coalition anyway if One Nation had not been there?

Antony knows that that is the key issue: as he puts it, “If these candidates had not been on the ballot paper, would the result have been
different? Were there voters who voted for One Nation and gave preferences to Labor, who if their had not been a One Nation candidate,
would have voted for the Coalition? If this was the case, then One Nation changed the result by acting as a ‘pied piper’, leading voters
across the left-right political divide with their preferences.”

But is this really what happened? We don’t know; the high One Nation vote of itself isn’t evidence. But there is some evidence from the
election held the following weekend in Queensland. There, One Nation only contested a bit under half the seats. But the swing to Labor –
about 9% – averaged around the same in the seats One Nation contested and those that it didn’t. This suggests that voters were swinging to
Labor anyway, and that whether or not One Nation was there to use as a half-way house didn’t make any fundamental difference.

I don’t pretend that that’s conclusive. For a start, Western Australia isn’t Queensland, and local factors might make an important difference.

Moreover, just the presence of One Nation in the election (with all the controversy about preference deals and differences within the Coalition) was probably hurting the non-Labor parties, regardless of whether they had a candidate in a particular seat. After all, most voters don’t know which parties are running in their seat until they turn up at the polling booth and see who’s on the ballot paper.

So I’m happy to accept that the strong One Nation showing in WA benefited Labor in 2001. I just don’t think that it was as big a bonus
as people imagine, and I doubt very much that it made the difference as to who won. But there’s no doubt that those ex-One-Nation voters will be an important factor this time.