xxx Census results are out.

The 2016 Census results have been released today. 27 June 2017.  Not bad for timeliness, especially considering the chaos around collection day.

A note of caution:  I have subsequently discovered that the boundaries of State Suburbs have been substantially revised in this area since 2011.  The names haven't changed however so I only picked this up later when comparing population sizes and wondered why Hoskinstown and Primrose Valley had shrunk so much.  So be wary of material for State Suburb level.

Further, I found that in my earlier deliberations in making the comparison between Table Builder for 2011 and a Profile for 2016 I didn't pay enough attention to the fine print regarding what was included and what was excluded form the various tables.  Thus some of the analysis of dwellings was well wrong.  I have now corrected it

It also appears that the ABS has got its servers sorted out as when I tried to get data (about 90 minutes after the release) I was able to

• download a General Profile for Carwoola 2016; and
• get data for a table for Carwoola 2011 through Table Builder.

So my concerns - following the release of the TableBuilder shells - appear to have been allayed, at least as far as the capacity of ABS systems.

I haven't really got into the data yet but the age profile looks quite similar between the two censuses.

Considering the polynomial line it seems to have moved up an age class, which I'd expect with a pretty stable population having got 5 years older between the two counts.

Of some small concern is an apparent decline in population between the two censuses.  In 2011 there were 1433 people counted in Carwoola which had declined to 1418 by 2016.  There are two broad ways in which this could have occurred:

1. a net drop in the number of occupied dwellings; or
2. a decrease in the average number of people per dwelling.

My observation around the area is that there has been a reasonable increase in the number of letterboxes which I know in some cases has meant that new houses have been built.  I suspect there have also been some extra dwellings built in the development at the end of Wanna Wanna Rd.  Thus I am inclined to doubt reason 1.  Fortunately information is available from the Census on the number of dwellings.

I am not aware of any non-private dwellings in the area although it appears that there were a small number (it appears in the results as 3, but this may be an artifact of the confidentiality processes) found in both Censuses.  However I will ignore this and focus on the private dwellings

2016 2011 Occupied private dwellings 490 462 Unoccupied private dwellings 41 30 Total 531 492

So my expectations of an increased number of dwellings were met, once I was able to compare two datasets through Table Builder.  I have mentioned to ABS that it would be good if the apparently similar tables in Table Builder and the Profiles could be based on similar definitions rather than having to perform relatively complex adjustments to match them.  (As I am not sure that all the data required - eg on Household Type - was in the first data release this becomes very tricky.)

What I did observe after the Census was a high proportion (~20% from memory) of the letterboxes to which I deliver the Stoney Creek Gazette contained copies of the Census Form labelled "Final Reminder".  Since it seemed that the Census Collector for that area didn't visit houses but just dropped the forms and other correspondence in the letterbox I strongly suspect that:

• he (I met the person at one point so can attest to his sex) will have missed any dwellings that didn't have a letterbox (including ours since I had removed the box as we were travelling away for a period around the Census - we completed the form in a caravan park near Rockhampton); and
• If the form wasn't removed from the letterbox when the Collector came back the property was regarded as either not having a dwelling or at best an unoccupied dwelling.  

All in all I am not very happy about this situation!  However, down the track it appears that only 6% of identified dwellings didn't respond - but how many were not identified?

I also note

1. an article by Bill McLennan, former Australian Statistician, expressing some doubts about whether the quality of the data can actually be established; and
2. the report of the Census Independent Assurance Panel which appears to give the count an rating of OK.  They do however qualify their findings:

"Given the limited time and data availability, the Panel made this assessment at the national and state and/or territory levels for selected variables. Specifically, it was not possible to assess the accuracy of statistics for small geographic areas." (emphasis added).