Another questionable poll

There are many very questionable polls run online. There is no way of knowing how inaccurate or slanted or rigged they are.

Here is some strong and generally justified criticism of some polling two election campaigns ago.

Colin Craig and his dodgy polling, Ctd

Colin Craig is at it again, using dodgy polling to justify his own agenda.

…That all sounds interesting until, yet again, it turns out that it is not a representative sample and there is a margin of error of 6.9%.

Colin’s strategy is flawed.

Here is another flawed polling strategy.

POLL: Will National get your 2017 Party Vote?

Whaleoil is running a  regular poll to see where those who supported National during the 2014 election are likely to go this year.   We aim to detect any shifts throughout the year leading up to September.

If you did not vote for National in 2014, and you do intend to do so for 2017, then this poll isn’t for you.  We are looking to measure the “restlessness” among 2014 National supporters.

But it is a self selecting poll there is no control over who responds to the poll, so no way of knowing who respondents voted for in 2014.

Normally running political polls on partisan blogs is pointless.

Not just normally.

Asking our audience who they will vote for and assuming that is a fair representation of the nation would be beyond stupid.

But in this case Whaleoil is somewhat representative of the most committed and interested in politics that normally would support the centre right of politics.

There’s no way of knowing that with any accuracy.

As such, measuring a shift in our audience will be significant as an indicator of what National supporters are considering at large.

A rough indicator at best. The poll:


A decent number of responses. The poll results:


Oddly the number of responses has dropped by a about 20%. Results from all polls this year:

Feb Mar Apr May Jun Jul Aug A Aug B Sep
Other 5 5 4 5 5 4 3 3 3
WILL vote National again 53 55 48 52 57 55 68 76 78
MIGHT vote National again 23 21 20 17 15 16 13 8 7
WON’T vote National again 19 19 28 26 23 25 16 13 12

This was from a 1900+ voter sample*.  The voters are Whaleoil readers.

It isn’t a sample. “A sample survey is a process for collecting data on a sample of observations which are selected from the population of interest using a probability-based sampledesign. In sample surveys, certain methods are often used to improve the precision and control the costs of survey data collection.”

It is self selecting  survey with no way of knowing anything about the quality of those who responded.

Based on this non-scientific but strongly indicative Whaleoil poll, National will get 39-41% of the party vote in September, depending on the proportion of wasted votes being distributed from parties such as TOP and the Conservatives which will not make it to parliament.

It isn’t strongly indicative, it is roughly indicative at best. There is no indication what the assertion “National will get 39-41% of the party vote in September” is based on.

Even if the survey could have limited respondents to people who voted for National in 2014 this only tries to measure the number of people who intend to vote National again, or who intend to not vote National.

It has deliberately tried to exclude anyone who voted Conservatives, ACT, UF, NZ First, Labour, Greens or the Internet Party last election and who may now be considering switching to National.

It tries to exclude people who could have voted last election but didn’t. It tries to exclude first time voters.

It may (or may not) be a coincidence that for months Cameron Slater has been claiming that National support will drop and they are likely to go as low as the thirties, and that is what this suspect poll analysis somehow arrives at.

The survey and the analysis/assertions should be taken with a big bucket of salt. It could be as shonky as Craig’s self promoting polls last campaign.

1 Comment

  1. patupaiarehe

     /  September 5, 2017