From left field: polls apart - quantitative surveys and predictions 2001

Ever since the election, we’ve seen a continuing debate about which pollster called the outcome. In other words, people are wondering how Morgan was so far off the mark, and Morgan is desperately trying to suggest that people changed their minds between the time he polled them, and the time they voted.

We chose not to pick the election on the basis of generic polls. Instead, the preferred approach focused predominantly on the demographic foundations of the two parties’ support and the distribution of the swings across the 150 electorates.

On this basis, From Left Field was prepared to go beyond making a generic call – which was all the attitudinal pollsters can do – and go one better. That is, From Left Field looked at the likely results in individual seats and tried to pick the outcome by seats.

During the course of the campaign, From Left Field made a number of ‘predictions’ about the likely outcome, and which seats we would see change hands. These predictions varied minimally from one to the other, typically in response to changes in campaign dynamics, but the general picture was the same.

Three weeks out from the election, From Left Field called the result as a ‘status quo’ with some reshuffling of the deck chairs. This view did not change during the campaign. Using demographic analysis, rather than attitudinal polling, we called the election almost to a seat.

From Left Field’s predicted result with a week to go was: Coalition 79, Labor 69 and Independents 2. We weren’t far out, with the final result being Coalition 80, Labor 67 and Independents 3. Modesty forbids me from bragging too much, so I’ll let the figures speak for themselves.

In making these projections, the demographic-based analysis got Ballarat, Dickson, McArthur, Parramatta and Canning dead right. The Nationals holding Hinkler was also anticipated, as was the status quo results in Longman, Moreton and Petrie.

From Left Field was, however, of two minds over Dobell and Makin, expecting a very tight contest. Earlier on, the expectation was for Dobell to fall to the Coalition and Makin to remain unchanged. By the last week, the pick was Dobell for Labor (just) and Makin for Labor (just). Just wrong on both those counts. On the flip side, the early money was on Labor losing Bass, but by the last week the pick was for no change, which proved to be right.

We also were wrong on Ryan, largely because we had over-estimated the extent to which Labor’s gains among the blue-collar workers at the time of March by-election would stick, and over-estimated the impact of Labor’s gains among the well-educated ‘intelligentsia’. This last group did not swing to Labor at the time of the by-election, but was moving Labor’s way by November. But no where near enough to compensate for the fall in Labor’s vote from the middle-mortgage payers.

Actions speak louder than words

I had previously argued that Labor’s defeat/Coalition victory cannot solely be explained by reference to the Tampa issue. In fact, what we witnessed in the result was a consequence of a long-run secular narrowing of Labor’s support base and a significant re-alignment of party support.

Labor’s claim is that Howard’s victory is tarnished by Tampa (no dispute on this front from me, but frankly neither side can claim much moral high ground here – remember the ‘shoulder to shoulder’ stuff?) and that the election was lost because the race card was played.

This simplistic appraisal shouldn’t be taken too seriously. Labor’s new leadership certainly understands that the reality is much more complex. Simon Crean’s first act as leader was to visit Sydney’s western suburbs, which speaks volumes about what the leadership actually thinks went wrong.