Monday, 31 July 2017

2016 Census Carwoola: Income in the Gazette area

My original aim for this post was to do a general post about what could loosely be termed indicators of affluence in the area.  However when I started to investigate the most obvious indicator of this - income - I found I had enough to say about that, without pushing the strength of the data beyond breaking point, to complete a post on its own so here we go.

Personal Income

The question on the form asks for people's income in ranges, defining the response categories in terms of weekly income and the annual equivalent.  That reflects the different ways people in different situations think about their income.  Personally, I always think of my income in terms of annual amounts (because that is how Commonwealth Public Service salaries were expressed and that is what ATO is interested in) so I will use those values in what follows.

I am unsure what happens now, but back in the day most of the ranges quoted on the form related to income levels of interest to people concerned with social welfare so tended to split up the lower levels.  Some other users were more interested in getting more detail on the big end of town to that they could identify just which postcodes were a good target for marketing the bigger BMWs and dodgy tax avoidance schemes but they were unlucky.  I'll return to this below when comparing 2011 and 2016.

In all of what follows I have expressed values as the % of cases in which an income was reported.  This excludes the categories "Not applicable" (mainly persons aged less than 15 years) and "Not stated".  The Not Stated rate is traditionally higher for this variable than many others but is very similar between the censuses for the Gazette area at 8.7% in 2011 and 9.1% in 2016.

The first chart shows the income profiles for the Gazette area and Australia for 2016.  My use of a trendline is simply to try to reduce the "noise" in the graph and reveal a general picture.
Clearly incomes in the Gazette area in 2016 are higher than for Australia in total.

Comparing 2011 and 2016 was made complicated by changes to the ranges appearing on the form between the 2 Censuses.  In some cases the change was simply a split of 1 class in 2011 into 2 classes for 2016 while for 1 case the boundary between 2 classes was changed and in another 2 classes in 2011 became 3 in 2016.  As clear as mud?  Perhaps a graphic will help!
So I plotted the profiles for 2011 and 2016 using the combined ranges.
 The shapes of the distributions are remarkably similar between the two years with a broad implication that incomes are higher in 2016 than in 2011.  Which fits with expectations.  Whether this is because of rises in remuneration rates for the same jobs or because residents have changed jobs to higher brackets is well beyond the scope of these data (let alone my ability to sort that out).

Within the Gazette area I examined the profiles by sex.
 I will simply note that I am surprised by the shape of the polynomial curve for females.  If readers can explain (or even hypothesise) what that means please put a comment on this post!

The other 'basic' demographic classification is age but trying to present profiles for 10 age classes and 15 income groups would be an incomprehensible  mess.  Rather, I decided to present information about the median income class for each 10 year age group.  That is, the income class where half people were in lower classes and half above it.  If anyone wants detail about that it can be provided but generally, "trust me I'm a statistician"!
This is essentially consistent with the chart above comparing overall profiles for income in the Gazette area and Australia.  Note that we have few residents aged 80+ and the Income Not stated  rate for that small group (27%) was much higher than for any other age class.

Household Income

As well as including data on the personal incomes reported on the forms the ABS adds up the incomes within a household (=dwelling) to give a household income.  There are a few nuances in this explained in the census dictionary but that is the principle.  Here is a chart showing the percentage of in scope dwellings according to income levels (I have plotted the upper bound of the income class for ease of presentation).
I have chosen line styles to emphasise the polynomials which cut out the "noise" in the series.  They are a pretty good fit to the underlying data as shown by the correlation coefficients.  My major takes from the chart are that:

  • the data points are clustered to the left of the chart, reflecting the user-interest in detail of the lower incomes; 
  • overall the shapes of the two series are similar, with a long 'tail' on the right of the chart; but
  • yet again the Gazette area has a higher proportion of people at higher income levels than Australia as a whole
Looking at the original data I am intrigued by the presence in both series of a marked peak for the income group $104k to $130k (plotted at $130k).  I can't at this stage work out, with certainty, why that is so but it is interesting that an annual income equal to twice Average Weekly Earnings for November 2016 is $120k right in the middle of that class.   I need to think about that a little more.

Saturday, 29 July 2017

Some snaps of things natural

It was pretty cool (about -1oC ) and foggy when we took off on our dog walk at 8am this morning.  It was still that way as we dropped down the hill in Widgiewa.  I thought the consequent muted light made these trees look attractive.  Rather like the River Red Gums in Sydney Nolan's Big Bend paining.
 Later in the day a friend sent me an SMS to say there was a Yellow-billed Spoonbill at the Bungendore Cultural Centre (wet Annex).  ie the Sewage Works.  Indeed there was!

 There were several other duck species present but nothing of great note.

Going back home I passed down Trucking Yard Lane and noticed quite an amount of bird life at the small dam just off the Kings Highway (Trucking Yard Dam #2).  There were a good number of ducks on the dam and a very large flock (at least 100 birds) of House Sparrows in a leafless Hawthorn.  I have counted 75 birds in the image and there were others to the side and in the back of the tree.
 There were 24 Plumed Whistling Ducks on the nominate Trucking Yard Lane Dam.  Large numbers of Pacific Black Ducks (50) Grey Teal (40) and Chestnut Teal (at least 20 with at least half being males in full spiff) were on the dam or nearby paddock and at least 25 Australian Shelduck around the feed racks in the background.

The most numerous birds were Cockatoos with at least 50 Little Corellas, 100 Sulphur-crested Cockatoos and (I guessed) 600 Galahs.  They were in several very mobile flocks and I found it impossible to count them.  Here is a small part of the mob in quiet feeding mode.

Thursday, 27 July 2017

2016 Census Carwoola etc: Marriage, relationship and families

Before getting into the topics in the subject line, I will mention how useful I am finding the 2016 Census Dictionary in explaining the meaning and/or content of items.  Anyone planning to use the data should consider looking at the dictionary.

This post started off thinking about working my way through the traditional demographic variables with Marital Status the next cab off the rank.

I will confess that early in my more serious work with the Census (I think around 1993) I suggested - publicly - that there was no longer a need to have this topic in the Census.  Did I ever find out quickly that this was entirely a Satanic proposal.  I can't remember quite how many letters of protest we got (all of which were knee-jerk reactions by very conservative people) but vividly recall getting a 2 page spread in the Sydney Daily Telegraph with many Sydney identities (including Bobby Limb and Dawn Lake; and Warwick Capper and his wife) saying how important marriage was to them.  I was quite chuffed that this amounted to them saying how important the Census was  - very good publicity!

So the topic stayed in the Census.

In the intervening Census cycles things have changed a little such that social marital status (ie de-facto relationships) are now also covered.  (In the 1980s this wasn't possible in Australia although we noted that the French Census did include it as the fourth question).  So I will cover that topic as well.

Marital Status

Registered Marital Status

Social Marital Status

Comparison of Registered and Social Marital Status

This gets very complicated, due mainly to different definitions of "Not Applicable".  For Registered Marital Status the Not Applicable category is simply persons under 15 years of age.  For Social Marital Status it also includes Visitors from within Australia as a major component (and a few other small categories).  There is possibly also a minor issue with Registered Marital Status being imputed while Social Marital Status is not.

While it is conceptually possible to cross tabulate these and sort things out it would be very difficult to do.  Instead I will simply note that the two variables are generally consistent: 
  • 86% of people who indicate " in a Registered Marriage" in one series do so in the other; 
  • 85% of people who have never married (Registered)  are either in a de facto relationship or not married;
  • For those with a Registered status of Divorced or Separated close to equal numbers are either in a de facto situation or not married.
  • Only 10% of  Widowed people are in de facto situations
  • All those "not Applicable" for Registered (ie those under 15 years of age) are correctly also "Not Applicable" for Social Marital Status.

Same Sex Marriage

Until the 2001 Census if a person indicated that they were the spouse of the Head of the Household  but was of the same sex their relationship was adjusted to "unrelated".  Since that time the Social Marital Status relationship is shown as "Married in a de facto marriage" and a same sex couple indicator is set according to the sex of the persons concerned.  The counting unit for this variable is "Families": the Not Applicable value is set for Non couple-Family Households.

Under 0.5% of couple families in the QPRC area chose to indicate that they were same sex couples.  As a result I believe that attempting to produce data for the far smaller (in terms of population size) Gazette Catchment would run foul of the ABS confidentiality protection mechanism.  However I found the following chart interesting. (Note that "Sydney" refers to the LGA "City of Sydney" and not the entire Metropolitan area.)

Relationship within Household

During my time in a number of Census Programmes (about 15 years in various roles) I found sorting out relationships and family structures to be a source of great mystery.  It was in fact a very rigorous operation but always seemed to me to rather like this (with apologies to Gosciny and Uderzo).
A core variable is the information gathered on Relationship to Person 1.  (That term in and of itself has scope for much entertainment: on an observation study I noted that whichever member of a couple filled in the form they put their spouse as Person 1.)  Even at the condensed version of this classification shown in Table Builder there are 31 types of relationship listed.  I have cut this down far more tightly as shown in this chart with % for the Gazette area and Australia.
This shows a similar pattern to that I observed in 2011: a marked majority of couples and kids.  Overall these two large groups amount to 78% of the people in the Gazette area contrasting with 69% for Australia as a whole.

Family Composition and Household Type

Before getting on to the results from 2016 I will just pass a comment about the definition of a Household in Censuses.  It used be defined in terms of a group of people within a dwelling who shared living expenses, often expressed as "eating from a common pot".  This was based on history, and presented many issues in modern society, especially for group households and multi-family dwellings.  (At work a bunch of us who had lived in group households at various times exchanged stories about domestic arrangements and they varied widely between groups and over time within a group.  Communal eating rarely featured - unlike communal drinking.)

It is thus pleasing to see that the current ABS definition is now "A household is defined as one or more persons, at least one of whom is at least 15 years of age, usually resident in the same private dwelling." which matches far better to the operational procedures for enumeration.  I'd imagine this change has caused much alarm and despondency amongst academic sociologists who used to pontificate - often wrongly - about the interpretation data based on of the old definition.

Family Composition

There are 575 Family Households in the Gazette catchment of which 540 are single family households  and 35 are 2 family households. There are also 214 non-family households comprising mainly Lone Person households, Group Households, dwellings (= Households) at which the Collector didn't make contact; and vacant dwellings.

This chart gives further details about the composition of the households containing families.
Clearly single families are the commonest form of composition in the Gazette area.

Household type

In effect this can be used as an expansion of the "Not Applicable"category above.  It clearly shows that the Gazette area is dominated by family households.  In this classification "Not Applicable" means vacant dwelling and, in the Gazette area, "Other non-classifiable" means dwellings where the enumerator didn't make contact.  (There are some other types of dwelling which are not classifiable but they are not found in this area.)

Tuesday, 25 July 2017

Rebuilding begins

After a lot of visits by builders and contractors our fences and sheds claim was signed off by NRMA Insurance  earlier this month.  It took a bit longer than we'd thought, but they did have a bunch of claims in the area so I'm not complaining.

There are approximately 5 main stages to the work:

  1. Repairs to the woodwork on the deck and consequent painting;
  2. Fencing;
  3. Demolition of burnt sheds;
  4. Replacement of sheds;
  5. Replacement of fire-damaged main water tank.
Today saw the start of part 1.  This will be really good to get done as we have not been able to use one route off the deck since the fire (and, as always once you are blocked from doing something you realise just how often you did it).

The first two photos are before work started.  The unpainted wood is a make-safe done by some tradesmen very soon after the fires to replace a purling which had burnt out at the base.

Here are the guys from ACT Trade Services going at it. Giovanni on the ground and Roberto on the deck!
They certainly got stuck right into the work (and were very pleasant guys).  Here is what they achieved on the first day at the West end of the house ...
 ... and the East End (or sunroom end).  It wasn't interesting beforehand so I didn't photograph it at that stage.
After somewhat over a day and a half they'd done the business and were off to do good deeds elsewhere!  Here is the excellent result of their efforts.

 Here is half of the remains now added to a central pile of crud for the demolition dudes to clear up.

Sunday, 23 July 2017

2016 Census Carwoola: Age and Ancestry for Gazette area

The only place to begin this is with a discussion of what is meant by Gazette area.  In using Table Builder for the 2011 Census Results I combined 3 State Suburbs (Carwoola (CA), Hoskinstown (HO) and Primrose Valley (PV)) to approximate the catchment area as shown in the graphic below.  I noted that the area for Primrose Valley was not right, as the boundary crossed the Queanbeyan River to include Urila.
 In starting "work" on this post I was surprised to find that the populations of Hoskinstown and Primrose Valley had dropped by about 50% between 2011 and 2016.  This led me to look at the ABS Geography pages which showed that those two suburbs had both been split (which was not at all apparent as the same name had been retained).

Carwoola appears to be the same in 2016 as in 2011.  Primrose Valley has been split with parts now going to Urila (U) and Yarrow (Y) neither of which existed (nor indeed exist) in the State Suburb list for 2011. Hoskinstown has had 3 parts split off going to Forbes Creek (FC); Rossi (R) and Palerang (P).  It is possibly not the fault of ABS as their "Maps" page includes these boundaries (along with Electoral and Postal boundaries) as Non-ABS structures.    However, IMHO they should have a good hard look at how the boundaries are presented in Census output.

Another note for caution: "our" area is Palerang and the State Suburb of Palarang is much further South - close to Dalgety.

Here is the map of suburbs in the Gazette Catchment (or at least included therein in 2011).
The 'recognition' of Yarrow has little effect on analysis as there are no more than 15 people (and perhaps as few as 8) in that area .  With so few people the Census data are intentionally unstable to preserve confidentiality.  Palerang has zero population.  All the other 'new' suburbs have non-trivial populations.  

Population size and Age

Having combined the new suburbs to match the old definitions gives this comparison:
Carwoola and Hoskinstown show a slight - but puzzling - decrease since 2011 while Primrose Valley show some growth.
Using a polynomial to remove the random noise from age profiles shows very similar patterns in 2011 and 2016 for the Gazette area in total.
Getting back into detail here are the State Suburb populations for the Gazette area (plus Urila).


I decided that my second foray into the first tranch of data would be to look at ancestry: if there was a marked change in response rates between 2011 and 2016 I would have expected that to show up here as a major increase in "Not Stated".  In fact the % of records with no ancestry stated has declined slightly since 2011.  Possibly this is an effect of many forms being completed on line.

Looking at the incidence of Not Stated Ancestry and Imputation of age, in most suburbs it appears that in most cases a response of "Not stated" for ancestry means that the age was imputed (ie no form was received for a household).  In a couple of smaller suburbs an unusual situation appears in which the number of records for which age was imputed is greater than the number with Ancestry not stated.  I can only conclude that this is either:
  • an artefact of the confidentiality provisions; and/or
  • because a few people completed a hard copy form and the answer for age was not legible; and/or
  • because a few people are amazingly sensitive about their age!
At the most detailed (4 digit)  level of the ancestry classification, 35 ancestries were reported in 2011 and 30 in 2016.  Overall:
  • the number of people reporting each ancestry correlates very well (coefficient of 99.4%) between the years;  
  • The 5 most common "ancestries" English (EN), Australia (AU), Not Stated (N/S), Irish (IRE) and Scottish (SCO) appear in the same order in both years. Italian (now 6th) and German (now 7th) swapped places;  
  • 8 ancestries maintained the same rank, 14 moved up the list and 16 dropped down; 
  • 8 ancestries  disappeared between the Censuses while 3 others were added in 2016.
In view of these fairly marginal changes the ancestry profiles are very similar when shown as pie charts.

Looking at the individual suburbs tends to give a similar picture to the above although, with the much smaller population sizes for suburbs other than Carwoola, the domination of Australian and English is more evident.  It is also possible that a relatively high percentage (eg one ancestry in Rossi is 4% against the value of 1% for the total area) could be attributed to a single family.

My broad conclusion from this exercise is that the information for 2016 is consistent with that from 2011.  My reservation about the quality of the data continues to be a concern that some dwellings were not identified by the Collectors

Saturday, 22 July 2017

A further look at Imputation Rates

Imputation in the Census is a statistical process through which
  • estimates are made of the number of people in dwellings which do not complete a form; and
  • the core demographic attributes (age, sex and marital status) are imputed for each "estimated " person.
In a previous detailed post I concluded that it appeared that

  • problems with non-private dwellings (NPDs) - known by the UN Statistics Division as Communal Dwellings, which I actually think is a better term but not the traditional one used in Australia -  were more an issue in the capital cities; and 
  • incorrectly rating non-responding private dwellings (OPDs) as occupied was the core issue in rural areas.  

This post attempts to investigate that issue a little more rigorously.

I think I do that but will note that what follows runs a risk of illustrating two issues with statistical investigations:
  1. It is possible to find that one gets sucked in to following "interesting trails" well beyond the original focus of the research; and
  2. Analysis can go well beyond the capacity of the data to support the analysis with the result that one ends up where the sun don't shine (and I am not referring to the UK in that).
A first issue has been to find information about the number of people in NPDs.  At this stage I suspect that the data that actually counts people by type of dwelling (eg private x non private) is not available in Table Builder Basic.  However it can be closely approximated by the number of people coded to "not applicable" in the variable "Relationship within household" which can be cross classified according to whether the Age Imputation flag has been set.  That is what I have used below.

A second issue is that some of the most useful geographic codes such as Section of State will only be released as part of the final release in October.  That will be poked and prodded then but for the time being I looked at Capital City vs the rest of the State as an approximate indicator.
It is immediately obvious that the imputation rate for NPDs is much higher than for private dwellings regardless of the State, or part of State in which the dwelling is located.  For the 7 State-level areas for which a Capital and Rest split is made, the rate for NPDs is higher in the City in 4 cases (blue stars), and lower in 3 (red stars) .  For OPDs the rate is higher in the Rest of State in every case.

It is probably important to note the comment by the Census Independent Assurance Panel that a major issue was over-imputation in NPDs.  It appears (see detailed previous post) that more than 1/3rd of imputed records were incorrect.  I don't have any information about whether this effect was even across States or between Metropolitan and ex-Metro areas.  It appears however that it is still unlikely to reduce the real need for imputation in NPDs to the level required in OPDs.

At the risk of falling into risk type 2 above, I looked at the ratio of imputation rates NPD:OPD for states and the metro/ex metro split.  (Note that all the ACT is regarded as part of the metropolitan area - possibly, and reasonably - reflecting the very small population in rural ACT.)
The low ratios in the NT reflect the relatively high levels of imputation for OPDs in that area.

Making an heroic assumption that the behaviour of people in NPDs is similar between city and bush, I think that what this shows is that there is more of a response issue in OPDs in the "Rest of State" areas which makes the difference between the two dwelling types less pronounced (ie a smaller ratio).  I can't work out how to demonstrate whether the higher rate for ex-metro OPDs is:
  • A real response deficiency in such dwellings; or 
  • Due to Collectors in rural areas incorrectly identifying vacant dwellings as occupied on Census Night.
but I have a gut feeling, supported by a comment in the CIAP Report that the latter is (at least) important.  I have emphasised the reference to Census night since in the case where a dwelling is a weekender or such-like the Collector may have evidence that the dwelling was occupied at some stage during the Collection Period and conclude, wrongly, that this was the case on Census Night.  In theory the property owners should be able to indicate that the dwelling was unoccupied on Census Night through the on-line system but I don't think that was, in practice, the case.

I have given some consideration to investigating more detail about NPDs but have decided to leave this to another post.

Thursday, 20 July 2017

Three days in the life.

May you not live in interesting times! I make that statement having had an interesting 48 hours with the small dog.

We did an 8km walk in Queanbeyan with her on Tuesday 18 July and she was in fine form. On returning home she was put on the lead to put tooth to bone which concluded with her yelping very loudly. This is most unusual for her and when I went to investigate she was caught up in some vegetation. On untangling her she gagged and vomited up a lot of slimy foam.

This was repeated several times over the next couple of hours. She then seemed to come good until evening when she became very unsettled and appeared to be looking for a dark corner to crawl into.  A definite worry.  The unsettled behavior continued through the night and we didn't get much sleep.   The process of settling her down was to cuddle her and stay very still oneself.   By morning she seemed to be getting very unresponsive. 

As soon as our vet (John Montgomery of Sonza) opened on Wednesday morning we rushed her over there.  (Actually in comparison to when we knew she had been snake-bitten and time was of the essence this wasn't much of a rush - my guess is that it took me about 40% longer this time.)   

We were expecting John's advice to involve a green needle as a result of some cardiac situation.  Instead he diagnosed a spider bite (I suspect red-back) and gave her a shot of cortisone and some antihistamine, noting that she may be sleepy but overall the prognosis was OK. Indeed she was sleepy for most of the day. However by evening on Wednesday she had another drink of water and scoffed some beef and rice.  Then she sat on the bed in a pose which clearly said "Where's my Smacko?" 

A more restful night for all concerned.  She still seemed a bit quiet when she got up on Thursday morning but did explore the kitchen floor looking for any food scraps.  However when Frances went to the wardrobe to get her parka Tammy's ears were pricked and she trotted over to the back door looking very interested in going for a walk.  Off she went and despite having had next to no food for 2 days still managed to park a coil.

We kept the walk short, and on return offered her a couple of chop bones.  As she was still rated as a poorly dog she allowed them indoors on her mat.
Demolition dog rools!  The bones vanished quickly.

We then went into Canberra and while Frances took care of some business I went to North Weston Ponds with Tammy in the search for a decent bird of the day. The first interesting sighting was 2 Darters.
 A third Darter joined them but it was a fuss getting the camera out so didn't take that snap.  The hoped for species (White-fronted Chat) appeared but got chased off by a bolshie Magpie before the Lumix could be wielded.

Instead I took an image of the snow-free Brindabellas.
The weather wasn't that pleasant.  Some folk would say that 8oC and 35kph winds was somewhere between ordinary and average. So we didn't hang around.  But having a bit of time before due to pick Frances up I thought I'd swing in to the RSPCA shop to see what they had in the way of doggie jackets, in an effort to overcome Tammy's spitting the dummy when the road is wet.  They had just what was needed: Dog Hoodies at $5 a go..
I didn't run a client satisfaction survey but have put it on a couple of times without losing any digits, which is far from the worst outcome possible.

Overall a pleasant end to a rather fraught period.

Sunday, 16 July 2017


It was a tad cool out this way this morning.  We recorded -5.0oC while a nearby resident scored -8.1oC just about the time we passed her house on the dog-walk!

Of course this coincided with my test of the lawn sprinkler system firing up.  As a result some of the trees became a trifle ice encrusted.
 In this image I noticed how I had caught some gravid drops ...
 .. and decided to try and capture one at the point of separation from the twig.  Unfortunately this turned out to be a sort of "Whack-a-Mole" exercise but with scores of places for the target to appear.  So I haven't got a plop-drop but I found these make a nice sequence of capturing more mirror images of our house in the drop.

 This one makes it clear what I am on about!

Saturday, 15 July 2017

Some Winter images

This time last year we were in Atherton, experiencing maximum temperatures around 28oC.  This year it feels, some days, as though the number is similar but the letter has shifted 3 places down the alphabet.  Today was not flash.

The morning started off quite warm (0oC) but quite damp after some miserable drizzle.  This did give some interesting cloud FX on the dog walk.  Looking down on the Plain ...
 .. and back towards Taliesin.
 Later in the morning I took myself off to the Plain to see if I could discover some interesting birds.  Not really, was going to be the answer until coming back along Briars-Sharrow Rd I came across a flock of 318 (yes, I counted them) Sulphur-crested Cockatoos.  They have been feeding in that paddock for weeks.

 After the count I heard the distinctive calls of Double-barred Finches but couldn't get them into sight.  A heard record is good enough for Bird-of-the-Day!  As I climbed back into my car a male Nankeen Kestrel was busy sorting out its feathers.   Initially I thought it was a chestnut-headed female ...
 .. but once it put its wing down the grey head of a male was revealed.
 After a short flight its plumage was back in order.