A few weeks ago I said that I had developed a Bayesian model for primary vote shares. At the time, I was seeking (but had not attained) an integrated solution that ensured the primary votes shares across the four party groups (Coalition, Labor, Green and Other) summed to one (or one hundred per cent).
The publication of a Galaxy poll in today's Murdoch tabloids allows me to explain how I have updated my Bayesian models to handle primary vote predictions from the polling agencies.We will begin with the primary vote estimate. I have rewritten my TPP JAGS model for this purpose.
Today I want to look at the house bias revealed by a 365 day span localised regression (LOESS). I will restrict my analysis to the most regular polling series: Essential, Morgan face-to-face, Newspoll and Nielsen. And, just in case you missed it in the heading, I will focus on the polls of primary voting intention.We will start with the major parties. There is an interesting juxtaposition here.
A couple of days ago I examined my treatment of the Morgan face-to-face poll. For the next few months at least, I plan to leave the residual face-to-face polls within the weekly Bayesian aggregation. However, I will exclude them from the sum-to-zero constraint in the Bayesian model.