 |
| Click to listen to Prof Paul Spoonley pn343 |
Click here.
__________
Our first attempt to produce a census with people completing their return on line seems to have been a disaster.
One might have thought @StatsNZ would had a trial go before replacing the old system when people completed hard copy returns, delivered and collected personally to their front doors. With this census you could request the personal delivery but it would appear that as many as 400,000 (see Update above, now 700,000!) people did not, or did not complete the return online. That's one person in every ten, with proportionally more omissions from remote areas, poor people and Maori.
StatsNZ is using standard "fill in" data procedures obtained from other sources to compensate for the missing data. These procedures are usually used for minor omissions: not for 400,000. The end result is likely to be of very questionable validity.
All we appear to have at present is a very large sample survey with built-in biases, hardly better than a Colmer Burton or other poll.
Why does it matter? What is the census used for?
Much government spending is based on population, who and where people live. Among its more important uses is the funding of health boards, education and schools, and the re-drawing of electoral boundaries. Under-representation in the census returns could result in Maori losing a parliamentary seat, and health boards in the Northland and the East Coast and low decile schools receiving less funding.
It's hard to know who to blame, and now not all that important, though our politicians and StatsNZ need to heed the lessons learnt from what went wrong. Now, it's more important to know what to do about it, and whether the measures being taken by the census office to compensate for what appears to have been massive under-reporting will result in a "sufficiently accurate" census.
National, for its own reasons, has talked of the need to have another census before the 2020 election.
If informed people like Massey's Professor Paul Spoonley, whose thoughts are noted below, are not persuaded the "patched up" 2018 census is adequate for important analysis and use, I'd back National's appeal for a new one that we can rely on. And that means enumerators going back to knocking on doors, until we improve taking a census totally, almost totally or partially on line. -- ACW
"So a whole lot of systems are actually breaking down because we don't have Census data available."
Mr Spoonley said he believes it was wrong to switch to an online system without testing it.
"What went wrong is we went online with a system that we really hadn't tested - so I would have kept the old system, which had people turning up at your door and saying, 'Here's the Census, here's why you need to fill it in, and I'm coming back.'"And there were a whole lot of people who were not able to access the data online, and unfortunately the numbers involved are quite significant for groups that we need to know about."
Mr Spoonley said a colleague of his had estimated that the issue could disproportionately affect Māori populations, especially in the Northland and East Coast areas.
"For some Māori it might be as low as 70 to 80%...so if we're looking at policies and looking at funding, that sort of shortfall is huge.
"It's based on a population funding model - so you need to know what your population is."He said Statistics New Zealand are "backfilling" missing data from other sources, but that leaves the serious question of whether New Zealand can have confidence that the replacement data will be correct.
"I'm still very, very sceptical."
Mr Spoonley called out Stats NZ's MacPherson, who has delayed releasing any data four times now.
"This is a public agency - it's really important to this country. It provides something of value, around a billion dollars, to this country.
"I think Liz MacPherson does need to front."
Read also
http://www.scoop.co.nz/stories/PA1904/S00058/the-census-isnt-adding-up-whats-stats-nz-hiding.htm
https://www.radionz.co.nz/national/programmes/ninetonoon/audio/2018685977/census-2018-what-went-wrong
No comments:
Post a Comment