Polls apart – but what are we to make of political surveys when the results are so divergent?

What  to  make of the latest   opinion  polls?   Because they diverge  so far  from  each   other,  experts   say  one   must be  wrong.

Perhaps some of those pollsters who got the mood of the  Australian  electorate so  badly  out of kilter with the actual election  result have been  imported to  carry out   one or  other of  the two samples.    Or, maybe,  because  the polling  done  by  each organisation covered   different   periods, there  was  a  dramatic revision  in the political mood almost  overnight.

Even given the divergence, the pundits were virtually unanimous:  Simon Bridges  is  gone—if not  by  lunchtime,  then some time soon.

That’s, of course, what they said at  the time  of  the  last set  of  polls.

National’s leadership preoccupies these  commentators much more than the actual performance of the  government.

Over   at   Kiwiblog,  David  Farrar – who   knows  more about the science  of  polling than any  of  the  political  commentators – says  the   results in the latest two  are so   far  apart that  statistically it can’t be a margin of  error.

Let’s look at the four main parties:  Labour 42% in OneNewsColmar Brunton and 50.8%% in NewsHubReid Research;  National 44% in ONCB and 37.4% in NRR;  Greens 6% in ONCB and 6.2% in NRR;  NZ First 5% in ONCB and 2.8% in NRR.  You basically can’t reconcile these polls.  One (or both) of them seem to be outside the 95% confidence interval, ie is the 1 in 20 “rogue” result.  The only other plausible explanation is that as the ONCB poll started a few days after NRR, Labour had a massive drop in support after those first few days. But the difference in dates is unlikely to explain the massive gap”.

 Farrar further notes the polls even show the direction of change differently.  One has Labour down 6% and the other up 3.3%. National is up 4% in one and down 4% in another.

The NZ First result is also outside the margin of error. A 5% and a 2.8% result is outside the 95% confidence interval. Bottom line is that at least one of those polls is wrong. They can’t both be right.

Perhaps the most egregious bit of reporting  from  the polling  was  the headline   used  by    NewsHub   over a   question  on   National’s  move to   leak  elements of the  budget.   According to  NewsHub, Most NZers  think   National was  wrong to leak  Treasury  budget  details”,  as  if   it had  sampled    the  bulk of the population  instead  of a  randomly selected   group  of  fewer than  1000..

The pollsters  might   start  losing credibility as  fast  as  those in Australia did  if  this is  how the  results of their work  is presented.



Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.