Comments : Leave a Comment »
Tags: Cell phones, Landlines, non-coverage, Response rates
Categories : Methodology
Landline survey population non-coverage
The 2013 Census tells us that 85.5% of NZ households have a landline telephone. This means that landline RDD (random digit dialing) sample frame non-coverage is up 6.1 percentage points, from 8.4% of households in 2006 to 14.5% of households in 2013.
This isn’t the percentage of individuals not covered by landline sampling frames. It’s the percentage of households.
Between 1% and 2% of households will not have access to either a cell phone or landline, so this means the proportion of cell phone only households has increased from somewhere around 6% in 2006 to somewhere around 12-13% in 2013.
Other than door-to-door surveys (which can take a couple of months to carry out and process), there’s still no other single sample frame that provides the coverage of a landline sample frame.
- Simply adding cells to the sample frame won’t increase coverage by much, because you’re after people who have a cell but not a landline (all the others are already covered by your landline sample frame).
- The Electoral Roll can be a useful sample frame, except the process of matching names/addresses to phone numbers reduces coverage down to around the 40% mark. Also, the phone records used for tele-matching can be based on home and vehicle ownership databases, so they introduce a socio-economic bias to your sample.
My view about conducting a good phone survey
This is something I’ve been working up to blogging about, because I know it will annoy a few people (it already has). I think this is partly because my views are seen as naive by some – not those of a ‘seasoned pollster’. Oh well, I’ll just have to live with that. Someone obviously believes I know what I’m doing. I hope they’re right!
My litmus test for blogging is ‘would I say this to a client?’ I would definitely say this.
Read the rest of this entry »
Comments : Leave a Comment »
Categories : Uncategorized
The discussion over at Dim-Post inspired me to have a play with the New Zealand Election Study (NES) data.
Each wave surveys a fairly large sample of voters and a small sample of non-voters. So I was having fun, and I started to wonder what would happen of all the non-voters with a party preference had got out and voted on Election Day. There are a bunch of caveats to this analysis, including the small sample size and how representative the sample of non-voters was. BUT, if we assume for a moment that the data were broadly representative, then inspiring all non-voters to get out and vote wouldn’t have had a massive impact on the 2011 result.
I ran the same analysis on the 2008 data. A higher voter turnout would have made a bigger difference in 2008. The result for National would have been lower by around 4 points and the result for Labour would have been higher by the same amount (based on results for just 15
12 non-voters with a party preference)
There were a few odd things in the 2008 data – I couldn’t get the validated voter/non-voter result to match the actual voter turnout, and there seemed to be some sort of weighting issue (which I haven’t explored yet). If anyone has time to check that out and replicate the above, I’d appreciate it.
Comments : 1 Comment »
Categories : Real polls, Self-selection
There has been a bit of commentary about whether Len Brown should resign following his confirmation that he had an affair. Although there has been a very unscientific poll on this topic, at this point there have been no random polls released to show what the general public think. I’d say there’s a fair chance one may pop up soon (please note that at this point I have no direct or indirect knowledge of the existence of such a poll). In the meantime, I did some digging.
Back in September 2006, TVNZ commissioned a ONE News Colmar Brunton poll to find out whether eligible New Zealand voters thought Don Brash should resign following an alleged affair. The question wording and results were as follows:
Don Brash is taking time off from Parliament to spend time with his family after accusations that he has been having an affair. Do you think that he should resign as leader of the National Party as a result of this?
- Yes – 23%
- No – 70% (84% among National Party supporters and 66% among Labour Party supporters)
- Don’t know – 8%
Based on the results of the above poll I’d say that, to a point, New Zealanders will generally differentiate a politician’s ability to do their job from what they get up to in their private life. The situation for Len Brown is a bit different though, as his affair can be interpreted to be an abuse of his position.
I suggest taking the unscientific Herald poll with a pinch of salt at this point, as self-selecting polls can be quite misleading as a measure of public opinion.
The Herald commissioned DigiPoll to carry out a proper random poll of 500 Aucklanders on Wednesday and Thursday. Some of the key results were:
- 51% think Len Brown should remain mayor and 40% think he should stand down (assume the rest are unsure). Note how different this result is to the one from the unscientific poll, where 57% said he should stand down.
- Among those who voted for Len Brown in the election, just under 70% said they would vote for him again if a new election were held.
- Based on the poll result, the Herald says Len Brown would win a new election on 34% support, with John Palino gaining 20% and other candidates gaining 32%.
Comments : 1 Comment »
Tags: Census, sampling
Categories : Sampling
The essence of my previous post was that the 2013 Census usually resident population count was lower than Stats NZ’s estimated population count by around 230,000. That seemed to me to be quite a substantial difference, but there are a number of reasons for this:
- As Thomas Lumley at StatsChat pointed out, the estimate and the Census cover slightly different populations. The estimated population count is for all those who live in New Zealand, including those on overseas trips (holidays, business trips, etc). The Census usually resident population count includes all those who live in New Zealand and are present in New Zealand on Census night (ie, it excludes people who completed the form on Census night who usually live outside New Zealand, as well as New Zealand residents who just happened to be out of the country on Census night).
- According to Stats NZ, around 205,000 NZers took trips overseas in August 2013. According to MED the average trip takes around 20 days (credit given again to Thomas Lumley for pointing me to these stats). Thomas suggests temporarily overseas NZ residents may account for around half of the difference between the Stats NZ estimate and 2013 Census count. This leaves around 115,000.
- I’ve been in touch with Stats NZ. They carry out a post-enumeration survey to check the accuracy of coverage and response to each census. The results for the 2013 Census will be available in March 2014. However the 2006 post-enumeration survey estimated that the 2006 Census undercounted by around 2%. If the same is true in 2013, then this would be an undercount of around
84,840 people (correction for bad math) 86,570 people.
Stats NZ base their estimates on available census data, and births, deaths, and migration since the previous census. It’s clearly a complex calculation – and as far as I can see only around
30,160 28,430 people are unaccounted for. This figure is less than 1% of the 2013 Census night count – fairly impressive at the total population level!
Next I plan to look at Stats NZ’s sub-national population projections, and to compare those to Census counts. Those projections are even more complex, so it’ll be interesting to see how they stack up.
Comments : Leave a Comment »
Tags: Normal distribution
Categories : Interpretation, Sampling
Okay, so my blog is quickly becoming a copy of whatever gets posted on StatsChat.
This video is a nice way to explain the normal distribution, why it's useful, and why sample size is important in surveys. It's an animation by Shuyi Chiou, published in the New York Times, which explains the implications of the Central Limit Theorem.
If I can't come up with more of my own content soon, I'll just redirect hits to StatsChat