Pollytrend:Did the budget change the polling trends?

OK sports fans – as promised we’ll have a bit of a squiz at these post-budget polls collectively to see if there was a trend change – and to do this we’ll use our Pollytrend model. For those that don’t have a clue what I’m talking about, just click here for an explanation.

The reason we use the Pollytrend approach is because of its flexibility – we can effectively tune how aggressively we want it to look for a change in polling trends, which will be a handy thing come an election campaign when we can ramp it right up. The more aggressive we make the locally weighted polynomial regression that defines our Pollytrend line, the larger is the probability that the algorithm will start treating polling noise as real polling movement. Conversely, the less aggressively we tune the algorithm, the less likely is the chance of it picking up noise and the smoother the trend line will be as a result, but at the expense of our trend picking up any smaller polling movements which may not be noise, but may in fact be real. Pollytrend is a trade off between certainty and accuracy.

Let’s take a squiz at the ALP two party preferred vote for every poll this year, and where we’ve got three trend lines running through it. Firstly, the Pollytrend as it was this time last week (the black “pre-budget” line), secondly the Pollytrend as it is currently (the red, “post-budget” line) and finally a third line where we ramp up the aggressiveness of the algorithm (which, statistically, uses a smaller bandwidth and seeks out cubic regressions trends rather than the quadratic trends we use for our usual Pollytrend line).

(click to expand)

As we can see – using our ordinary Pollytrend regression, the downward trend that started mid-April has simply been pushed out to last weekend. The magnitude of the trend didn’t change, nor its direction – but the slope of the trend, the speed of the trend, has changed. It became a slower,  gradual trend.

However, when we ramp up our Pollytrend regression and force it to look for shorter term trends, it finds an ever so slight upward movement in trend terms as a result of the last round of post-budget polling.

If we take the longer view – which we should – there is a slight trend happening away from Labor that started in mid-April and which is continuing through to the present.

If we take the shorter view to answer the question “did the budget change the trend?”, then we can say that it did indeed change the trend to the best of our knowledge over that short term period – it halted the pre-existing downward trend that Labor was experiencing.

But we cannot yet tell whether the short term trend change that we’ve found with our aggressive Pollytrend will continue and force the longer term voter trend to change. Only time will tell us that. But it does put yesterday’s Fairfax hysterics in it’s place.

The other thing that’s worth having a squiz at is how the dynamics of the approval/satisfaction ratings of Rudd and Turnbull have been behaving recently. If we look at how the results have changed since the end of March/beginning of April period through to today – we see a remarkable polling consensus. Again, just click these babies to expand them:

That data supports our Pollytrend analysis, where government approval ratings generally and historically correlate with voteshare. As Rudd’s approval has slipped, so too has the ALP vote over that period. However, with Turnbull literally standing still on all of the approval and disapproval measures, there is nothing really driving the Coalition vote in terms of a buyers market for swinging voters – it’s still all about Rudd.

That will only get an Opposition party  so far with voteshare – and certainly nowhere near enough to form government if history is any judge.