David Hargreaves says the harsh reality of what will pass as Census 2018 information demonstrates the urgent need to bin the whole thing and bring forward the next census

David Hargreaves says the harsh reality of what will pass as Census 2018 information demonstrates the urgent need to bin the whole thing and bring forward the next census

By David Hargreaves

Well, we knew it was going to be bad. Now we know how bad. And it's BAD.

The confirmation by Statistics New Zealand on Monday that long-delayed information from last year's ill-starred Census will be rolled out from September onwards comes with the very unpleasant confirmation that try as they might the good folk at Stats haven't been able to sufficiently paper over the gaps in some key information - notably that around iwi affiliation. So there will be no release of that information. 

Let's put that into context. Some of the brightest brains in Stats NZ have been working for months to find a way of using the data they have from the census, plus other available data, but haven't in key areas been able to patch it together well enough  to release information they think is reliable. That's an appalling situation, frankly.

What's really surprised me in all this is the absence of any meaningful input either from the Government or the opposition. 

The opposition's relative silence on this one (maybe it's because it isn't about tax?) is perhaps understandable given that the census, the great dog's breakfast of it all, was put together on National's watch under the previous administration.

The Coalition Government's reticence is rather more perplexing.

After all, you might have thought this would be a free pass at some National-bagging. A casual drive-by salvo along the lines of "they tried to penny pinch and this is the result".

Be that as it may, now that we are truly confronted with how bad this census outcome has been, the Government really needs to get into gear.

Now, right now, the Government should be urgently putting in place a new census as soon as is practically possible. It would appear unlikely it could logistically be done before 2021. Well, okay. Let's have one in 2021.

And I would go further. The Government should come out and say that it rejects the information in this census and it will not use it. That's extreme. But I think this is a big enough shambles to warrant extreme reaction.

I hesitate to say this, but is the Government's relative silence on the issue down to ignorance of just how crucial this information is in terms of forming key planks of decisions on things such as infrastructure?

As I said in an earlier opine on this subject, having ship-shape statistics is not some 'nice to have' fanciful thing. It's absolutely vital.

Patch-up jobs never work. Something as big as a census has to be done right first time. And if it isn't, then no amount of tinkering is going to fix it, not to the point where you can truly trust the results. 

And sure, Stats NZ will defend the final outcome because what else could they do at this stage?

But the fact is, someone somewhere - probably in this Government - should have said a long time ago to Stats NZ that this 2018 census was not going to work.

Well, I say again that this is what needs to be said now.

Pull the plug. Put the whole thing in the bin. 

Yes, some would consider it an outrageous waste of money, but I think the money's already been wasted. We've got results we can't trust. 

Nobody hates waste of taxpayers' money more than me.

But think how much MORE money could be wasted if key infrastructure spending decisions are made based on information that proves not to be, to use the favourite corporate word, "robust".

I say the Government needs to get on to this now. 

It's a mess and it will need some quick affirmative action for us to set about as a country remedying it.

We need to swallow the foul-tasting medicine on this one.

And we also need to say: "Never again."

We welcome your help to improve our coverage of this issue. Any examples or experiences to relate? Any links to other news, data or research to shed more light on this? Any insight or views on what might happen next or what should happen next? Any errors to correct?

We welcome your comments below. If you are not already registered, please register to comment.

Remember we welcome robust, respectful and insightful debate. We don't welcome abusive or defamatory comments and will de-register those repeatedly making such comments. Our current comment policy is here.

18 Comments

And perhaps a little word in Gubmint's ear about the failed processes that led to them Gaps:

  • Reliance upon on-line completion
  • Insufficient user-proofing to detect and flag issues such as skipped questions or constant selection of Enumeration Zero choices
  • Relative lack of appreciation for and support of non-on-line customers
  • Lack of manual follow-up especially in remote/tribal/off-the-grid areas

There's still something to be said for Census Enumerators charged with getting every household in their area, attended in person, questionnaires distributed and collected, and followed up as appropriate. If nothing else, it's a great Gubmint Employment scheme....

Perhaps the rise of IT capacity and knowledge has led to a lack of skills on the ground. You know - How to organise the beating of the feet and the doorknock.

I still never understand the smoothing/managing/estimating/making-up of data in a Census by StatsNZ. The Census should not have any missing data to smooth. It is THE record, with which all other datasets are smoothed too.

As suggested, it needs a complete redo immediately. But current form suggests they will not manage it well, if at all.

Census data is always augmented. In 2013, 4.7% of the data was donated from other similar records. This is normal practice to produce a usable data set.

This time round, they have used data from other government data sources to fill the gaps - arguably a much better way to get a fuller dataset.

This census is also a better count of the population than the previous one.

Given that the 2018 Census cost taxpayers about $120million, the outcome is shattering. A monumental failure of a flagship exercise.

I don't buy the argument that the current shambles is due to under-funding. That could be argued about virtually any Government activity - yet plenty of initiatives still achieve acceptable outcomes. All public servants know they have to do the best they can with the limited resources made available. Clearly, on this occasion, Statistics NZ didn't do well enough.

Neither do I buy the argument that "on-line was to blame". Recall that the 2013 Census was conducted partially on-line - which was seen as being a trial/pilot run. Stats NZ should have learned from its 2013 trial and also from on-line censuses conducted overseas - including Australia. Potential issues with response rates, the importance of putting sufficient enumerators in the field etc should have been foreseen, thought through carefully and properly sorted before the census date of Tuesday, 6 March 2018........

Stats NZ has a very talented bunch of professional statisticians, I've been told by people in the know. Why, then, were their abilities not harnessed and used to avert the crisis that's eventuated? I can't help wondering whether there's been a degree of ineptitude/incompetence within the higher-level management of Stats NZ - as Noncents alludes to in the final sentence of his post above.

The social cost of the 2018 Census debacle will be enormous - extending far beyond the $120million cost of administering it. The real cost of the failure will be measured in terms of compromised policy-making and decision-making by government. In short, thousands of public policy analysts, advisors, economists, demographers, social statisticians etc across the Public Service and wider public sector won't be able to work as effectively as they have in the past - because they will be hampered by census data deficiencies. Thus, Government Ministers and Cabinet will be provided with worse quality information in deciding how to allocate funds and, ultimately, the public will suffer as a consequence of this. And right at the time when government wants to focus on the "wellbeing" of people as a new budgetary approach.

The census fiasco will also impact negatively on work carried out across the NGO and private sectors, which will also suffer from data deficiency.

Whatever, whether the next census is held in 2023 (or sooner [2021] as many independent experts believe it should be) the government and the public need assuring that it will be run by a fully competent senior management within Statistics NZ. Let's hope that reassurance will be coming soon.......

TTP

I particularly like how people definitively say there is a housing shortage, rather than a bubble, when we don't have reliable stats.

From the data released today you can piece together an occupancy rate - there were 4,700,000 counted in census with an estimated 58,000 people not counted living in 1,870,000 properties counted plus 0.8% undercount. This works out as 4,758,000/1,885,000 = 2.52 people/dwelling.

These are not final numbers yet but I would not expect it to vary much.

And in 1998 we had a population of 3,815,000 and 1,440,000 private dwellings which =2.65 people per dwelling.

Some might argue that less people per dwelling contributes to a housing shortage (renters vs owner occupiers etc), but it simply means we’ve gone from 377 dwellings per 1000 people in 1998 to 396 per 1000 in 2018. Shortage?

There's a shortage of properties that people want to live in. Many people are living garages or other run-down properties with poor maintenance and would rather not be.

Well, we really have the perfect situation to carry out a meaningful census: a very small population, a common language, a very high literacy rates, good organisational and communication infrastructure, and supposedly a very good sense of national identity, loyalty and pride.
Try carrying out a census in China which would miss out on each of these criteria.
There would be very few countries where conditions were anywhere as good for completing a meaningful census. Yet we failed - were our expectations too high regarding reliability of data?

It isn't easy. ""Domesday Book is a manuscript record of the 'Great Survey' of much of England and parts of Wales completed in 1086 by order of King William the Conqueror"" The next attempt was 1873.

The Anglo-Saxon Chronicle states: ""Then, at the midwinter [1085], was the king in Gloucester with his council ... . After this had the king a large meeting, and very deep consultation with his council, about this land; how it was occupied, and by what sort of men. Then sent he his men over all England into each shire; commissioning them to find out 'How many hundreds of hides were in the shire, what land the king himself had, and what stock upon the land; or, what dues he ought to have by the year from the shire.' ""

Anything achieved in England in mid-winter on foot is impressive.

Spending another $100m+ on a census in 2021 to count Iwi membership seems a little excessive. The main uses for census are population-based and this census is an improvement on the previous one.

Yip. This seems to be a kneejerk reaction that some data wasn't collected therefore it is assumed that this is vital data that we must have.

.. so , wotcha saying David , is that we spent a lot of money on the last census and it resulted in a shambles , and is incomplete ....

Therefore we ought to bite the bullet in 2021 , and spend a heap more on getting it right , and produce a complete shambles ?

Given the census has the ability to deliver some rather solid data points, I wonder if it wasn't sabotaged via neglect by National so they couldn't be held to account over issues like homelessness, actual inbound immigration, foreign home ownership, that sort of thing. Of course National hadn't even planned who to form a coalition with in order to actually be in government and need to defend what was bound to be some pretty damning statistics....

That is so cynical, I'm surprised I didn't think of it myself

No way are we scrapping the data from the 2018, that's ridiculous and absurd. It's not that long to wait for another few months to clarify the data, the first release of the data will be on 23 September this year.

It's really important from an economic point of view to see what happened from 2013 to 2018. We all know that National which is responsible for the so called online 'dogs breakfast' census if the first place.
And they would love nothing more than to cover up the damage they did over that time frame, when National were at the height of blatantly selling off NZ to overseas non-resident buyers. We need to see how much damage has been done?! How many empty homes are there? And what is the real economic situation we're in?

We simply can't wait for another few more years for yet another costly census.

There were many. There were too many to count. They will never find everyone. Not everyone wants to be found. But I reckon that between IRD, Winz & a couple of other key govt databases, they could get to within plus or minus 1%. My guess is they're doing something similar right now.