Census, Surveys, and Cellphones…

.

.

Three polls in February (2013)  yielded two distinctly different results.

.

	Roy Morgan 13 Feb 2013	TVNZ/Colmar Brunton 17 Feb 2013	Fairfax/Ipsos Poll 20 Feb 2013
Right bloc:
National	44% (-2%)	49% (+5%)	44.9% (-1.3)
Maori Party	0.5% (-1%)	1% (n/c)	1.3% (-.01%)
ACT NZ	0.5% (n/c)	.01% (-0.5%)	.04% (+.04%)
United Future	0% (n/c)	.02% (-0.3%)	.01% (-.01%)
Left bloc:
Labour	34.5% (+3%)	33% (-2%)	36.3% (+1.9%)
Greens	13.5% (n/c)	11% (-2%)	10.7% (+.02%)
Mana Party	0.5% (n/c)	1% (n/c)	1.4% (+.08%)
Other:
NZ First	4% (-1.5%)	4% (n/c)	2.8% (-1%)
Conservative Party	2% (+1.5%)	1% (n/c)	1.6% (.02%)

n/c = no change

At the previous election, National was consistantly polling high – in the low-to-mid 50s. Their election result was actually 47.31%, several percentage points out from polling figures from Roy Morgan, Colmar Brunton, and Media Research,

.

	Roy Morgan 24 Nov 2011	TVNZ/Colmar Brunton 24 Nov 2011*	Fairfax/Media Research 23 Nov 2011	2011 Election results	Closest Polling result
Right bloc:
National	49.5%	50%	54%	47.31%	Roy Morgan
Maori Party	1%	2.0%	1.1%	1.43%	Media Research
ACT NZ	1.5%	1.7%	0.7%	1.07%	Media Research
United Future	0.5%	0.1%	0.1%	0.6%	Roy Morgan
Left bloc:
Labour	23.5%	28%	26%	27.48%	Colmar Brunton
Greens	14.5%	10%	12%	11.06%	Media Research
Mana Party	0.5%	1.0%	1.1%	1.08%	Media Research
Other:
NZ First	6.5%	4.2%	4%	6.59%	Roy Morgan
Conservative Party	n/r	2.4%	n/r	2.65%	Colmar Brunton

n/r = no result provided

As the chart above shows, polling figures were all over the place. Roy Morgan called it closest for the result for National – but under-reported for Labour.

Colmar Brunton called it closest for Labour.

Whilst at the same time, Media Research was way out for National – 6.69 percentage points off the mark – over twice the margin of error (3.1%) for that poll.  (see:  Gap closes as election looms – poll)

None of the polling companies were consistent in their results and all over-polled National. Only one pollster over-polled for Labour.

Part of the problem are anecdotal  stories that many low income families, students, transients, etc, no longer rely on landlines and use only cellphones. Polling companies do not call cellphones – only landlines. (A low-income family living not far from us fits this demographic group perfectly; no landline; cellphones only. The sole-parent head of the household votes Labour.)

This year’s census has an interesting question; Question 17,

.

.

The question asks the respondent to “mark as many spaces as you need to show which of these are available here in this dwelling”.

What will prove interesting is not whether or not “a cellphone/mobile” is marked – but how many households will mark “a telephone”.

This will finally give us a clearer understanding what percentage of households do not have a landline.

If the numbers of households without a landline are significant (+/- 10%), then polling companies will either have to adjust their polling techniques – or be rendered useless. Without factoring in cellphone-only households, polling companies risk becoming an expensive ‘parlour game’ with little value.

One option is to return to the days of door-knocking pollsters. It’s an expensive option, but may be more reliable than phoning people on landlines.

This blogger keenly anticipates Statistics NZ releasing poll results on Question 17. It’s impact on politics in this country may be greater than we realise.

.

*

Previous blogposts

Three recent polls

.

.

= fs =