Polling and cellphones – sorry, Frank, salvation does not that way lie (Part 1)

At the Daily Blog, Frank Macskasy is back on one of his favourite hobby horses, which is essentially that you can’t trust the polls because they don’t call people’s cellphones. Mr Macskasy lives in the hope that a vast conglomeration of the young and the poor, who either can’t afford a landline or don’t see the point in having one, will rise up on 20 September, proving the pollsters wrong and delivering a crushing victory to the Left.

The object of his derision this time round is 3News reporter Emma Jolliff for her story “Cellphones make political polling tricky“. The reasons for his scorn are twofold. Firstly, the length of time it has taken for 3News to work out there is a possible problem for polling companies, given that the census data reported on 3 December 2013 showed that 14.5% of households do not have access to a landline; and secondly, Ms Jolliff’s statement that:

“…because there is no directory of mobile phone numbers those people are essentially off the grid to pollsters.”

Mr Macskasy rightly points out that Roy Morgan call cellphones (and had in fact called Mr Bradbury’s cell once for its political poll), making a lie of the statement that cellphones are “off the grid to pollsters”.

The actual issues for pollsters, in terms of calling cellphones, are set out most elegantly by Andrew from Grumpollie, in his post from March 2013 entitled “The whole landline vs cell phone thing“. According to Andrew, the main issue is that:

If the purpose of calling cell phones is to reduce non-coverage of likely voters, then you may actually need to ‘screen out’ those you call on cell phones who also have a landline (because they are already covered by the landline sample frame).

If approximately 85.5% of households currently have a cellphone and a landline, that’s a hell of a lot of interviewer hours screening people out!

Additional issues that Andrew identifies are (to paraphrase):

  • Some people have more than one cellphone, meaning that additional weighting is required to adjust for the probability of selection.
  • A lot of cellphone numbers are out of use, but still go through to voicemail when called, meaning it’s almost impossible to determine if there’s actually an eligible person at the end of a number.
  • The small proportion of New Zealanders with a cellphone but no landline is so small that party support would need to be dramatically different among that population for non-coverage to influence the poll results for party support.

Andrew picks up that last issue again with his latest post – “Polls and cell phones… again…“. Using data from Statistics New Zealand’s 2012 Household Use of Information and Communications Technology Survey, he notes that although there is indeed a greater proportion of non-landline households in lower income groups, 47% of non-landline households have an income of more than $40,000. In fact, 10% of households without a landline earn more than $100,000.

Then combining the Statistics NZ data with stats from Census 2013, Andrew breaks all New Zealand households down by income band and whether they have landline coverage, with the following results:

  • 8% of New Zealand households earn up to $40,000 and have no landline. 30% of households are in that income bracket and do have a landline.
  • 3% of households earn between $41,000 and $70,000 and have no landline. 20% of households are in that income bracket and have a landline.
  • 4% of households earn more than $70,000 and have no landline. 35% of households are in that income bracket and have a landline.

Andrew concludes:

When you factor in the size of each non-covered income group (8%, 3%, and 4%), they would have to differ massively from each covered income group, when it comes to party support, for them to make much difference at all to a total poll result.

According to Frank Macskasy, the polls are wrong because those without landlines are overwhelmingly left wing voters. Their support will therefore remain undocumented by the polling companies; undocumented until The Only Poll That Matters, election day. However, given the spread of non-landline households across all income brackets, there doesn’t seem to be much evidence that political preference of those households differs dramatically from the rest of the population. And the difference would indeed have to be dramatic for there to be any significant impact on the poll results.

Not that that will stop Mr Macskasy from railing against the polls from now until election day…

Correction: The original post was aimed at Martyn Bradbury, thanks to my completely missing Frank Macskasy’s byline on the Daily Blog post. When someone at the Daily Blog rants about polls being inaccurate, my mind tells me I must have just been reading something by Mr Bradbury. Mr Bradbury’s name has now been replaced by Mr Macskasy’s, which doesn’t alter a thing, since Mr Macskasy’s view on the pollsters seems identical to that of Mr Bradbury…

Advertisements

5 comments

  1. These pollsters could avoid all these problems if they just increased their sample sizes. God knows it can’t be that expensive to pay students minimum wage to make a few more phone calls.

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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s