Leadership polling: a comprehensive rundown

Yesterday both Galaxy and Newspoll released polling on leadership ratings of a Rudd/Gillard head-to-head, while Morgan asked the same with a phone poll late last week. Looking at the three questions, Galaxy asked:

Which of the two Labor politicians, Kevin Rudd or Julia Gillard, do you believe would be the best choice to be leader of the Labor Party?

Newspoll asked:

Thinking now of the leadership of the federal parliamentary Labor Party, which one of the following do you think would be the best candidate to lead the Labor Party?

While Morgan asked:

Thinking of Mr. Rudd and Ms. Gillard. In your opinion, who would make the better Prime Minister — Mr. Rudd or Ms. Gillard?

All three questions produced similar results:

While that’s marginally interesting and all, it doesn’t really provide us with any temporal context. Fortuitously, Morgan covered this leadership issue further with it’s semi-regular phone poll that not only asks this head to head question, but asks additional questions which look deeper into leadership perceptions by providing more response options, covering not only the Labor party, but the Liberal Party as well.

As Morgan runs this poll a few times a year, it gives us a longer term perspective on how leadership perceptions are changing over time.

First up, the headline figures for each Party since December 2008. I’ve knocked out a few of the smaller guppies that generate small single figure responses (like Jenny Macklin for Labor and Christopher Pyne for the Libs) and threw them in with the “Others/Can’t say” category – but you can see the full list over here if you’re particularly tragic.

The big ‘story’ coming out of these polls has been how Gillard has closed in on Rudd, with there being only an 8 point separation between the two now on the leadership question. We can see the convergence with a quick chart (click to expand):

However, if we look closer and split the numbers into voting intention cross-tabs, it’s not quite what it seems.

What is happening underneath the headline figures is that Labor voters have shown only a slight decline in their support for Rudd while Gillard had a slight jump among Labor voters in December and has kept that approximate level of support since.

Where the majority of the Rudd/Gillard convergence has come from is with Liberal Party voters. As the Liberal Party vote has increased over the last few months, the proportion of Liberal Party voters making up the headline figures has increased pretty substantially. As we can see, it has been Liberal Party voters moving away from Rudd and slightly towards Gillard that has driven the majority of the change in those headline figures.

If we look at how the Rudd/Gillard “Better Leader” numbers have changed since the October 2009 ALP polling peak, it tells us the bulk of the story:

Liberal voters have deserted Rudd and moved to Gillard at over twice the rate of ALP voters during the period – which, when combined with their increased weight in the headline numbers, has driven most of the change in those headline numbers.

The big question is how much of that change on the part of Liberal voters is just the rusted ons getting cranky (which we’ve seen occurring in nearly every other poll result – the polarisation of issues along partisan voting lines) and how much is caused by soft Liberal voters taking that view?

Moving now to the Liberal Party leader results and running a chart on the headline figures (from the first set of Morgan tables at the top of the post), we get (click to expand):

Here we’re seeing a three way convergence at the headline level, with only 2 points separating the three politicians! Again, it becomes more interesting when we break in down by voting intention cross-tabs:

Charting these buggers we get (click to expand):

It’s interesting that Turnbull is almost the perfect opposite of Abbott in terms of supporter bases, while Hockey wedges right in the middle of the two. More interesting however is the way Abbott’s support takes a massive jump in December (the Morgan poll in December 2009 was taken the few days following his ascension). It suggests that a substantial chunk of support for a given party leader simply derives from the fact that they happen to be the leader. Abbott’s ratings were low before his ascension, yet rocketed up to 42 immediately on taking the leadership mantle.

It’s worth mentioning that we’d expect to see that same leadership effect for any party.