The problem with opinion polls

Colmar Brunton released a poll result last night.

The results were quite different to the recent Reid Research poll and were remarkably stable to the last Colmar Brunton poll which was conducted last November.

The Reid Research Poll had National at 43.3% and Labour on 42.5%.  The Greens were on 5.6% and NZ First was in the danger zone at 3.6%.

The Colmar Brunton poll had National remaining steady on its November 2019 result at 46%, Labour was up 2 to 41% and the Greens were down 2 to 5%.  NZ First was struggling on 3%.

The different results allowed the related media entities to use remarkably different ways to describe what is happening.  Clearly both cannot be correct.

Jacinda’s preferred PM rating shot up a remarkable 6% to 42%.  Bridges inched up to 11%.  Over three quarters of his party’s supporters do not see him as being viable.  How embarrassing.

My personal view is that current polling does not properly represent the electorate as a whole.  It misses poorer people.  But unfortunately poorer people are less likely to vote. 

So the polling tends to approximate the eventual result but not through design. The inbuilt bias approximates the bias caused by turnout.  Which is why turnout and the ground game of each party will be so important this year.

Powered by WPtouch Mobile Suite for WordPress