Monday, August 26, 2013

Weekly update

Three key polls since last Monday's aggregation:

  • Morgan - 47.5-52.5 in the Coalition's favour - moved half a percentage point to the Coalition
  • Newspoll - 47-53 in the Coalition's favour - moved a percentage point to Labor
  • Nielsen - 47-53 in the Coalition's favour - moved a percentage point to the Coalition

While there has been a lot in today's paper about the significance of Newspoll moving a point to Labor, I suspect it is nothing more than the usual noise we see in randomly sampled opinion polls.

Aggregating these polls gives me a result that suggests Labor's slide in the polls might be coming to an end (but before calling it, let's wait an see what next week's data shows). Collectively, the polls are suggesting a sizable win for the Coalition.

At this point in the blog, it is my normal practice to remind people that I anchor the above Bayesian aggregation with the assumption that the net bias across all of the polling houses sums to zero. You will need to come to your own view about where the actual level of collective systemic bias lies for all the pollsters. At the 2010 Election (with a different set of pollsters), the population voting intention was about one percentage point more in the Coalition's favour compared with the pollster average (see here). In light of the 2010 experience, it is arguably plausible to subtract (say) half a percentage point or more from the above aggregation to adjust for the collective systemic bias across all of the polling houses. [As an aside, you will note that Simon Jackman, who seeks to anchor his Bayesian models with respect to the outcome of past elections regularly produces an aggregated poll that tracks well below the vast majority of individual poll results].

The astute among you will have noticed that I no longer include the Essential poll in the aggregation. I have excluded it because this series does not behave like a polling series. In plain English, it does not bounce around enough.

Aggregating just Newspoll and Nielsen yields ...

1 comment:

  1. Essential did it again this week with another 50-50 albeit off quite different primaries to last week. If it was just steady that would be one thing, if it was just steady and slow to respond even that would be bearable, but now that it's off inverting the trend behaviour of every other pollster it is really difficult to deal with.

    Simon anchors his model off the 2010 election. Some other models that anchor off past election results, such as William Bowe's, get quite a different result by anchoring off a range of recent federal and state elections. Possibly the ideal approach is to use lots of data but weight for federal cf state and weight for recency.