Update on the 2021 Census collection – 13 September
In this post, Glenn updates us on the latest Census tally figures from the ABS. How is the Census response rate tracking? How do unoccupied dwellings affect this story?
As at Monday 13th September, the ABS had received 9,484,122 completed forms for the 2021 Census, which they have estimated as of 91% of Australian households. Of these, 79.7% were via online and 20.3% on paper. I predicted an 80% online response rate in a recent blog, and it’s almost on this now, but has fallen – later responses are more likely to be on paper, because Census field officers have delivered paper forms to non-responding households. If you look at last week’s update, it’s apparent that 102,000 of the 160,000 responses (64%) in the past week have been on paper.
Census response rates by state and territory
Here is the state-by-state tally. The Northern Territory continues to lag, likely because of remote indigenous areas taking a while to get forms back. Other states/territories are all over 90% returned.
|State/Territory||Household forms received (online and paper)||Household forms received (estimated %) *|
|Australian Capital Territory||174,206||93%|
|New South Wales||2,956,165||90%|
Source: Australian Bureau of Statistics, Australian Statistician urges households to action Census reminder letter 13/09/2021
How is the completion rate estimated?
Let’s unpack that number of 91% a little bit. The asterisk on the ABS media release states that the percentage of forms returned is based on the estimated number of households, which includes dwellings which are unoccupied on Census night (eg. holiday homes, houses for sale or lease, people away). As the forms are returned, and Census field officers are visiting households, the number identified as unoccupied goes up, and is removed from the denominator here.
9,484,122 households have returned a form. This is already more than 600,000 greater than the total occupied private dwellings from the 2016 Census (8,861,620 occupied private dwellings).
Of course it should be higher – our population has grown since then. But how much higher should it be? The last Census counted 9,924,844 dwellings in Australia, with 1,039,872 unoccupied on Census night. The (randomly adjusted) difference between those figures is the number of occupied dwellings, above.
On 23 August the ABS stated a percentage return of 78% on a total of 8,762,036 forms. Reverse engineering this indicates the ABS was estimating there were about 11,233,000 households out there to return a form. Today’s numbers – 9,484,122 households for 91% – indicate that they’ve revised this total down 10,422,000. So a large part of the increased percentage is due to identifying dwellings as unoccupied or invalid addresses rather than returned forms.
How many dwellings should there be? Well we don’t know for sure – that’s part of what the Census of Population and Housing is run to work out. But based on the last Census with 9,924,844 dwellings, plus just over 1.01 million building approvals in the last 5 years, it should be around 10,934,000. (Yes, some would’ve been approved prior to the last Census but not constructed, but equally recent construction which is incomplete won’t make it in this time so it balances out. You’ll also have some demolitions in there, so this could be a slight over-estimate.)
Using this number, it’s apparent that the ABS have likely already identified quite a lot of the unoccupied dwellings and removed them from the total, but probably not all of them. We can’t really assume the last Census figure of about 1,040,000 unoccupied is similar this time around. One of my Census predictions was about the unoccupied rate, but it’s hard to tell. On the one hand far less people are overseas – so more likely to be at home – but on the other, we know many short and long-term residents have left the country. For instance, there are likely to be a lot of student apartments in our cities which are unoccupied. Maybe these average out? If the rate of unoccupied is similar to last Census, that would bring the likely total occupied dwellings to around the 9.9 million mark. Meaning that the current form return rate would be around 95%–96% of occupied dwellings, once all unoccupied are factored in, rather than the 91% reported. This is comparable to the final return rate of the 2016 Census. Some of these numbers are a bit rubbery though, as we really don’t know how many unoccupied are already in the data, or what the total count “should” be. If I can go a little existential for a bit, that’s a problem with trying to estimate the success of the collection using a parameter it’s attempting to measure!
How does the current completion rate compare to last Census?
We are now almost 5 weeks after Census. If 95% return is correct, this is still a fairly good result, particularly with the added restrictions on field staff in NSW and Victoria due to lockdowns and safe conduct in a COVID environment. But there are still some households who haven’t returned the Census form, and we would reiterate the Australian Statistician’s comments about the importance of ensuring every household is counted. The value of the Census is in its completeness. Every person counts, no matter their situation – and the power of the Census data is to provide a picture of our population at a very local level. If you know anyone who hasn’t done the Census, there is still time – they can fill it in online from https://www.census.abs.gov.au/ or request a paper form by calling 1800 512 441.
Census results are due to be released from June 2022 and we will be getting them onto .id’s community profile and other sites sites as soon as possible after this.