Another Newspoll is out. 52-48 (ALP-Coalition), two-party preferred. Very little change from the last poll, save for Abbot’s preferred PM numbers.
I just heard an interview with Newspoll’s Martin O’Shannesy from the ABC’s AM. The interview featured some discussion of what we call the poll’s “toplines”, but after that there is a lot of speculation that struck me as crying out for a little more work…
Q: “How much is due to the way that people view Kevin Rudd versus their view of the government?”
A: “Thats pretty hard to untangle because all we see is both a fall off in optimism for the government, and a fall off for of course for Kevin Rudd at the same time… I think this is probably part of the government’s problem, that there has been such a presidential, lead-from-the-front sort of style from Mr Rudd which of course has tied the government’s ratings closely to his personal ratings.”
O’Shannesy’s caution notwithstanding, the next question was
Q: “So you would put it down to Kevin Rudd’s effect, on people…”
We could of course analyze the micro-data, to see if the probability of supporting Labor is more or closely tied to evaluations of the leaders now than in the past. But that requires access to the micro-data. And it wouldn’t be the hardest thing in the world to do, to explain this simple sort of an analysis in a radio interview etc.
Later on, a good question:
Q: “You’re assuming Green preferences flow mostly to Labor as they did at the last election, but considering the disenchantment of voters with both parties now, do you need to change that assumption?”
A: “I think those assumptions do tend to be in question at the moment, especially there are so many people voting Greens and others…”
Why not ask minor party voters where they might direct their 2nd preferences? Sure, some people answer this question with some whimsey, but surely it couldn’t hurt. It would be very helpful to compare survey-estimate rates of preference flows with what we actually see in actual elections, helping us to calibrate the survey-based estimates.
I do agree that when the Green vote gets up as high as we are seeing in recent polls, the idea that those preferences flow back to Labor at the overwhelming levels you see in House of Representatives elections starts to become questionable. Its just not something we have a lot of data on; I’ll try to dig some analysis of preference flows from previous elections — the idea is examine the extent to which Green-Labor preference flows vary as a function of Green 1st preference share (one hypothesis is that the further the Greens dig into Labor vote share, the more likely you’ll hit people who actually prefer the Coalition over Labor, but are nonetheless giving their 1st pref to the Greens). On the other hand, if you are leaving the ALP to give the Greens your 1st pref, you have to wonder how an Abbot-lead LP gets your 2nd pref.
Q: “Do you think that if the [Labor] party were to change to Julia Gillard as leader … that that would change things for Labor?”
A: “I think thats like the $64,000 dollar question…”
Which led me to wonder “why didn’t you poll on it!” Maybe for $64,000 Newspoll would!
Which raises a serious point: even with the (relatively) deep pockets of News Ltd behind you, there is only so much political polling Newspoll can do. I wonder if these numbers come from 3 or 4 “political” items added to a commerical/market-research survey, or a general purpose “omnibus” survey for a collection of clients etc.? That is, perhaps its just not commerically viable for Newspoll to do, say, 15 mins of political content on n>1,000 person surveys every fortnight, with, say, split ballot randomizations for the Rudd/Gillard vs Abbot matchups etc (although one has to imagine that such a survey would attract a ton of eyeballs for The Australian).