Morgan comes a day early this week (good thing too, let’s hope it’s a new 2010 feature) with a face to face poll showing the primaries running 47.5 (up 2) / 35.5 (down 1.5) to Labor, washing out into a two party preferred of either 58.5/41.5 (up 1.5 to Labor) if we use the preference flows from the 2007 election, or 60/40 ( up 3 to Labor) if we let respondents allocate their own preferences – but more on that in a tick. The Greens are on 10 (steady) while the broad Others are sitting on 7 (down 0.5).

This comes from a two week sample of 1819, giving us an MoE that maxes out around the 2.3% mark.

Before we get to preference allocation issues – this is what both Morgan phone polls and Morgan face-to-face polls look like for the ALP two party preferred since the last election (where preferences were distributed on the basis of the 2007 election result)


One of the bigger questions around the traps as we approach the election is whether or not preferences will flow at the same ratio that they flowed at the last election – especially with every pollster showing a higher Greens vote compared to their 2007 election result.

We have two pollsters that measure the two party preferred in two separate ways allowing for a comparison – Nielsen and Morgan. Nielsen show no difference between their respondent allocated two party preferred (where respondents get to tell pollsters which party would get their preferences) and the 2007 preference flow two party preferred (where primary votes are distributed in the same ratio as they were at the 2007 election).

Morgan, on the other hand, does exhibit a very slight difference between the two results – less than half a percent – but a difference none the less.

If we look at all Morgan Polls and take the difference between respondent allocated ALP TPP and the 2007 election allocated TPP, this is what we get:


When respondents get to allocate their own preferences, the two party preferred ends up, on average, 0.4% higher for the ALP than if preferences are distributed by the pollster on the basis of the 2007 election results.

It not only occurs with Morgan face-to-face polls, but with their phone polls as well. If break down the difference between the preference allocations by polling type, we get:


The face-to-face polls show a slightly higher ALP two party preferred than the phone polls, both in headline terms and in terms of the size of the difference between the TPP allocation methods.

We’ll have to keep an eye on this as we approach the election, especially if Nielsen starts to show the same phenomenon.

If we start seeing a significant difference in the two party preferred between respondent allocated and 07 election allocated results, then pollsters like Newspoll which allocate prefs using only the 07 election method, might end up undercooking their polls by a point for the ALP.