New polls are starting to come out a bit quicker now, three from Roy Morgan Reasearch (1, 2, 3), and the first post-election One News-Colmar Brunton poll. I guess I will have to go back to semi-regular once-per-month or so updates.
The updated polling averages now have National on 48.2% +/- 1.5%, Labour on 29.3% +/- 1.3%, the Greens on 13.0% +/- 0.9% and NZF on 4.8% +/- 0.6%. Changes from the previous polling update are not statistically significant!
As usual, the graphs below summarise the polling averages for the party vote after the new poll. The horizontal axes represent the date, starting 60 days before the 2008 NZ General Election, and finishing on the present day. The solid lines with grey error bands show the moving averages of the party vote for each party, and circles show individual polls with the vertical lines representing the raw statistical errors.
A couple of points to note. The Greens have been holding steady at about 13% +/- 1% since a month or so before the election – about 2% higher than they actually polled. National are holding steady at about 48% +/- 1%, also slightly higher than they recorded at the last election. Labour, on the other hand, look to have had a bit of a rise, peak, and fall since the election, but it’s not necessarily statistically significant. Occam’s razor says that they are polling at 28% +/- 1%, where they have been since a couple of months or so before the election.
The other point to note is the talk on Twitter and the blogs about the credibility of the recent Morgan Poll results. National, for example, went from 48.5% two polls ago down to 44.0% (a drop of 4.5%) and then back up to 49.5% (a rise of 5.5%). On the face of it this looks a bit strange, but keep in mind that the error on National’s polling is about 2%, and relative to this the fluctuations aren’t huge. On top of that, previous post-election polls had them at 47.0%, 46.0%, 45.5%, 45.5%, all well within a reasonable margin of error. Taken alone these previous results show, if anything, too little fluctuation. But whenever you deal with political polls, or any numbers with an element of randomness, you should expect periods of little fluctuation followed by some large swings. That’s just how randomness works. Given the small sample sizes on the Morgan Polls of ~800 respondents (cf. 1000 for Colmar Brunton) they seem pretty credible.