Landline vs Cellphone polling
March 18th, 2013 at 7:00 am by David FarrarAndrew blogs at Grumpollie:
The problem with calling cell phones doesn’t really lie in the cost of calls. For a polling company, calling a cell phone doesn’t cost that much more than calling a landline. The problem is the complexity and cost of employing dual sampling frames when the proportion of cell phone users without a landline is still very low. 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 we assume (hypothetically) that 6% of eligible voters have cell phones and no landline, that means that 94% of the people you call on a cell phone will not be eligible to take part (again, because they are already covered by the landline sample frame). This is where the cost would really begin to build up – all those interviewer hours required just to screen people out (eek!).
It is worth recalling the stat – you’d have to call almost 17 cell phones to find 1 person who doesn’t have a landline.
At the moment such a small proportion of New Zealanders have a cell phone with no landline that party support would need to be dramatically different among those people for this particular type of non-coverage to influence the poll results for party vote (eg, support for Labour among cell phone only voters may need to be twice what it is among landline voters for the party vote result to shift by more than, say, the margin of error).
Also a very useful point to recall when people raise the cellphone issue.
Let’s say National is at 48% amongst landline users and only 40% amongst cellphone only users. If you poll landline users only then you get National at 48%. If you poll both landline and cellphone only users you would have National at 47.5%. Not a huge difference.
Also worth noting almost all polling companies weight by age so any differences between landline and cellphone users which are due to different age profiles get compensated for anyway.
When the proportion of people with cell phones and no landline is considerably larger than it is today (like it is in some other countries), then it will definitely make sense to employ a dual sampling frame approach.
The results of the census will be interesting to see how the proportions have changed.
Tags: cellphones, Polls
March 18th, 2013 at 7:28 am
Look forward to you polling Cellphone users only, as a cross check of your wild assumptions.
Vote:March 18th, 2013 at 7:34 am
The differences are overstated. Its not about whether you have landline or mobile or both, its whether there is an opportunity to reach the sample. For example the chance of someone having their mobile with him/her vs being at home and therefore available to answer are much higher.
Vote:Another complexity, we have a landline but only use it for outbound calls and messaging. Message is permanently on to avoid nuisance sales / poll calls. Friends leave a message. So we are classed as available by landline but in effect are not to pollsters
March 18th, 2013 at 7:53 am
Hey DPF, as a pollster yourself, what is your undertstanding of the impact of number of people per household?
I can’t ever remember being asked by a telephone pollster how many eligible voters there are in the household, let alone asking to survey more than one person in the house – yet surely this makes a difference.
I would speculate that households with more eligible voters would lean more left. This is based on a few stereotypes. Pacific Islanders and Maori families often live in larger households with more extended family under one roof – both groups are known to favour Labour in polling. Similarly, university students and teens and twenty somethings who are flatting quite often have 3 or 4 eligible voters under one roof sharing one landline. By comparison, older voters who are typically more conservative often live alone – giving them a voter to landline ratio of 1:1.
I agree with hannity above – it’d be interesting to do a few polls of just cellphones (where the ratio of voters to lines will always be 1:1) and see how that changes the results.
[DPF: Some polling companies weight by household size but not all. The most common weights are by gender, age and area. Sometimes also income, ethnicity and household size.
It is worth noting that in my experience the differences between weighted and non-weighted data is relatively minor - again less than the margin of error. It is only when different demographics have massively different responses that they impact much.]
Vote:March 18th, 2013 at 8:33 am
Whew, another inequality. (Bells inequality?) I know, change the rules for voting.
Vote:March 18th, 2013 at 8:35 am
The poor old activist greens who can’t afford a landline but can afford a smart phone, tablet and laptop.
Vote:March 18th, 2013 at 8:37 am
Hi Richard29
You are correct. Polls that select one person per household will have a bias in favour of people who live in smaller households.
Some polls do weight to adjust for household size.
Cheers
Vote:Andrew
March 18th, 2013 at 8:40 am
Pardonez moi … if I err sil vous plait but! … I dont recall? a census question on cell phone ownwership. Or do we find out this information? (from other census sources)
BTW, the census was an opportunity lost for a huge number of (optional?) “referenda” type questions to be asked (and moderated), for example the performance of your local council, support or not for a four year electoral term, support or not for Charter Schools … see what I mean?. Stay off the hot buttons of politics and religion though … they could never be “moderated”.
I know … I know … it is such a problematic proposition and could go on and on but! … I ask you to consider, how often is it that we get the chance to find out some indicative responses from such a huge sampling population.
After all we’ve paid for the “stamp” … IT access, surveyors etc already. Such a waste of public money huh?
[DPF: The census does ask about telecommunications devices - in the household portion.
An interesting idea about using the census to do referenda also. Would get a larger turnout for referenda but I guess the concern is fewer people doing the census if it is asking for views, not just factual information]
Vote:March 18th, 2013 at 8:46 am
lazza, I’ve never heard that suggested before, but I think it is a great idea. Could have asked about support for gay marriage etc, even if it was just an optional section at the end of the census form.
Vote:March 18th, 2013 at 9:01 am
What are the proportion of registered voters / likely voters who have a landline but not a cellphone? If the problem is not the two sets of calls but the dual sample frame, why not make cellphone owners the primary sample frame instead? It’d be interesting to see how the numbers look on that – perhaps 15% of people have a landline but no cellphone. But how many of them are likely voters?
Vote:March 18th, 2013 at 9:08 am
Why?
The stat was entirely made up:
If we assume (hypothetically) that 6% of eligible voters have cell phones and no landline
Vote:March 18th, 2013 at 9:09 am
perhaps 15% of people have a landline but no cellphone. But how many of them are likely voters?
Probably a higher percentage than the other way ’round. I imagine it skews old, and I believe old people are more likely to vote.
Vote:March 18th, 2013 at 9:29 am
I’ve been seriously considering dumping my landline connection .
We have 2 cells already, and the free minutes on offer now mean that we can probably dump our landline.
It is only recently that this ‘tipping’ point has been reached for me. The only issue for me now is that some people do not like calling cellphones as they do not get free calls.
Vote:March 18th, 2013 at 9:33 am
I have no land line and a cellphone. I also live overseas, but I still vote. I am unreachable by pollsters in New Zealand.
It would be interesting to have a survey of overseas voters to see how they compare to the general population. I suspect that there would be less support for the greens.
Vote:March 18th, 2013 at 9:39 am
Given that there are now more mobile phones than landline phones in New Zealand [citation needed]
And that a mobile phone is generally used by a single person, not an indeterminate number in a household
And mobile phones are more accessible to their users (who’s near your landline during say work hours?)
Wouldn’t it be better to poll only mobile phones to get a cleaner sample?
Vote:March 18th, 2013 at 9:41 am
gazzmaniac
We do that every three years, eh?
Vote:March 18th, 2013 at 9:43 am
Overseas voters cast special votes. The Greens do well on special votes, enough to pick up seats in over half* of the MMP elections they have contested. This suggests that overseas voters may support the Greens more than internal voters.
*I say “over half” because I don’t have details for all elections to hand. I know that at the 1999, 2008 and 2011 elections this was the case.
Vote:March 18th, 2013 at 10:03 am
My view was that overseas voters skewed to the extremes – more Green voters, more ACT voters. I’m not sure if it’s because the kind of people who get up and go overseas also tend to have strong opinions, if it’s because the young and the well off go overseas, or if it’s just distance from the day to day business of politics in NZ that tends to polarise people.
Vote:March 18th, 2013 at 10:40 am
Graeme – you might find that domestic special voters cause that too, because of kids away at uni but registered at the parents’ address.
Vote:March 18th, 2013 at 10:49 am
I am not sure the assumption that a higher proportion of poorer people do not have landlines is correct.
Vote:A lot of wealthier people I know do not have them because they are prepared to pay for expensive mobile plans (or work pays).
March 18th, 2013 at 10:58 am
Very predictable trend. Once landlines are done away with, CEll ph charges will skyrocket. Just like fuel, food, power, rents and soon water
Vote:March 18th, 2013 at 11:29 am
Absolutely.
My inclusion of (and italicising of) the word ‘suggests’ was intended as a note to the limits of the data upon which I was basing my tentative conclusion.
Vote:March 18th, 2013 at 12:56 pm
Sorry to burst that bubble of yours DPF, but that ain’t necessarily so.
I know plenty of people who don’t have landlines. It’s a bigger percentage than you think.
Vote:March 18th, 2013 at 1:00 pm
@ Rick,
http://media.tumblr.com/tumblr_m44lciKr5z1qcmg7q.gif
Vote:March 18th, 2013 at 2:22 pm
I wonder if the Census results will actually help you at all, DPF. The relevant question is wonderfully ambiguous:
If I have a cellphone but no land line, I still have a telephone (just it’s a mobile one, not a fixed line one), so I could tick both (but don’t know if I’m supposed to).
Vote:March 18th, 2013 at 10:09 pm
Out of interest, how do pollers get paid for this information?
Do newspapers have to buy it? Surely you can’t copyright facts – so couldn’t a paper just publish the results without paying for them? Who is paying for it?
I understand that companies will pay for market information useful to them, but who really cares enough to pay money for political polls? Or are political polls just a way (a ‘loss leader’) to get your company well-known so you get businesses interested in your services?
Vote:March 19th, 2013 at 2:37 am
Hmmmm, we have four in a household with six mobiles, and eight computers, combined income $240,000 plus, but no landline. We ditched that ladt year, what with wifi and skype and whatapp. We are two labour, one green and a national voter – not that DPF cares, sincehe seems primarily concerned with denying we even exist, along with every other inconvenient truth his handlers in the beehive dislike.
Vote:March 19th, 2013 at 2:44 am
Here is another prlem with trying to survey mobiles – caller ID. A large majority of people I know don’t answer calls from unknown numbers. They will listen to any voice mail left, or screen via a text, but not just answer any call – especially ones that have CLI blocked, for those just scream telmarketing/polling.
Vote: