Junk in, junk out

The Press reports:

Online polls have offered few clues to the early favourite in Christchurch's mayoral race.

Incumbent and outgoing Labour MP Lianne Dalziel are locked in a head-to-head battle to lead a new-look city council from October.

Three unscientific surveys have given conflicting pointers to ratepayers' preferred candidate.

Of course they have. That is because they are unscientific. How is this newsworthy? This is like saying that the astrology forecast in The Press and the differed this week, and writing a about it.

Late last week, The Press emailed a survey to 1700 readers asking them whether they would vote for Dalziel or Parker if they had to immediately cast a vote.

Some 597 readers responded and 377 (63 per cent) said they would support Dalziel, while 149 (25 per cent) said they would support Parker.

The remaining 71 readers (12 per cent) said they would support neither candidate.

A readers survey is perhaps the least worst unscientific survey. The sample is not representative, but people can only vote once and had to be pre-registered.

A press.co.nz poll with more than 11,000 votes has Parker leading on 50.5 per cent and Dalziel well back on 38.8 per cent.

This sort of poll is akin to astrology – worthless. People can vote multiple times if they want to, and candidates e-mail their supporters encouraging them to vote. These polls are entertainment, not a scientific poll.

Political forecasting website iPredict suggests Parker has a 65 per cent probability of retaining the job, and Dalziel a 37 per cent chance of claiming it.

iPredict has a good track record, but the fuel for it tends to be public scientific polls. In the absence of such, people are investing in a vacuum.

Polls in the last body election had Parker lagging well behind his opponent, , throughout most of the campaign but on election day he walked away the victor, with a nearly 17,000-vote majority.

Umm, probably relevant to mention the that occurred! All those polls were prior to the earthquake.