Newspoll — on trend or undercooked?

This fortnight’s Newspoll via The Oz comes in with the primaries running 41 (down 2) / 40 (up 2) to Labor, washing out into a two party preferred of 53/47 the same way, a two point drop for the ALP since last fortnight. This comes from sample of 1148 giving an MoE that maxes out around the 2.9% mark.

Rather than behave like the other two polls this cycle, Morgan Phone Poll and Essential Report which both had rises to Labor, Newspoll has decided to throw a spanner in the works and break right away from the collective.

Since mid April, the ALP has been riding a rather nasty looking vote slide – as we can see from the Pollytrend chart on the top right in the sidebar. The polling has been lumpier than the trend suggests, with chunks of support moving Labor’s way for a cycle – like the immediate post budget polling as a whole for instance – before retreating back to the Coalition.

It’s that lumpiness which the Pollytrend tries to cut through, but it’s also that lumpiness which should give us reason to ponder the results from this cycle.

We’ve had two polls showing a Labor rise, while Newspoll has a fall – the question is whether it is Morgan and Essential that have overcooked the results, or whether Newspoll has undercooked them this round.

If we take a squiz at the All Pollster results for the last 100 days:(click to expand)

We can clearly see all of the pollsters moving together over any arbitrary period here, but what stands out is the way the pollsters this round have the Coalition primary vote somewhere within the 3 point spread between 37 and 40, while the Labor spread is 7 points – between 41 and 48.

There’s a lot more variation in the ALP primary vote. Greens interplay can explain some of it, but there’s still a fair bit of volatility happening there. You can see how the minor party vote size affects the ALP TPP numbers by a simple chart that plots each ALP TPP result against the non-major party“Others” primary vote for every poll since January 2008. We’ll run a regression line through it to point out the obvious.

For every one point increase in the Others primary vote, the ALP two party preferred drops by, on average, half a point. That might seem a little counter intuitive since Labor gets larger preference flows from the broad “Others” than the Coalition enjoys – but what it highlights is the way a large chunk of voters move readily between being ALP voters and minor party voters, yet when they move from the ALP to the “others” column, preference allocations leak away slightly from Labor and toward the Coalition – hence the relationship in the polling. The more variation in the “Others” vote we get, the more variation we should expect in the headline two party preferred results as a consequence. It’s also worth mentioning that Newspoll had a relatively high non-major party vote this poll of 19.

If Newspoll has the Labor primary undercooked by a couple of points, then the next Newspoll (assuming nothing much changes over the next two weeks) should give us something around the 55 TPP mark – which would suggest that the trend against Labor has petered out.

The alternative is that the Essential and Morgan Phone polls (and it’s worth mentioning that Morgan had small samples this cycle) have overcooked the ALP primary, meaning that next cycle we would expect them to come in with TPP’s around the 53/54 mark – which would suggest the trend against the ALP is still continuing.

I personally think Newspoll has undercooked the Labor vote by a few points and the next poll will be back around the 55 TPP mark – but I could well be wrong and the ALP will need to have a good hard think about what they’re doing, while Turnbull will undoubtedly keep on doing what he has been – now with the added bonus of being free from the shackles of the Member for Higgins.

Meanwhile, on the other metrics, Turnbull has done well in continuing to pull his net satisfaction ratings up out of negative territory – the usual charts come in like this: