Thursday, August 15, 2013

Open season for statistical nonsense?

The election has apparently brought an open season on statistical abuse. Conus at Mission Beach hooked a big fish this week from Rupert's flagship and then landed it with assistance from Grogs Gamut:

More deliberate misrepresentation from "The Australian"

Election 2013: Day 10 (or, I’m 70% confident there’s a preference for bias)

Recent reports in the Cairns Post citing various ABS data which could not be easily verified had been noted. This week the Post followed up on the serious topic of affordable housing and homelessness:

Affordable housing shortage now crisis

In no way do I wish to be critical of the work done by Anglicare for the homeless in Cairns or the seriousness of the issue. Indeed I would fully support their comments on the inadequacy of Newstart Allowance which actually entrenches homelessness and long term unemployment.

However some of the methodology and comments are significantly misleading. Anglicare did a survey of online rental vacancies over a weekend. They then compared this with the census data from 2011 and then decided:
"The most startling figure is that on census night 2011 the median rental weekly rent was $240 yet in the survey of nearly 800 rental properties just two years later it is $336," Mr Roberts said.
Comparing two different data sets like this is NOT wise. What Anglicare did was look at online advertised rentals. I did this too and what you come up with is a current median not far distant from Rick Carr's CairnsWatch reports. This quarterly data is derived from the Residential Tenancies Authority.

The RTA data provides a breakdown based on rental bonds for rental properties by number of bedroom and type of property and also number of new rental bonds each period. CairnsWatch then derive a weighted median from this. Note: they are new rental bonds for each period so will lead the trend as will advertised rates. One would have thought this pretty good data and worthwhile further exploration actually!

However what it also shows is that the reported weekly RTA rent for the 2011 census period as indicated by the CairnsWatch weighted median was about $40 or $50 above the census number. There are several good reason why this could/should be so and a quick look shows that the RTA number was typically above the census for any region.

Here are the RTA numbers for the various components in the Cairns region (ex Douglas) over the last two years with nothing like the reported $96 crisis increase in any component:

Jun Qtr 11 Jun Qtr 12 Jun Qtr 13
Rent New Bonds Rent New Bonds Rent New Bonds
1br unit $195 401 $185 436 $200 394
2br unit $250 928 $255 855 $270 819
3br unit $330 181 $330 153 $340 158
2br townhouse $230 128 $235 118 $240 110
3br townhouse $300 28 $290 30 $340 19
2br house $280 39 $290 57 $285 54
3br house $310 443 $320 444 $330 451
4br house $370 399 $390 419 $395 364

 Never mind, the Cairns Post then followed up with an editorial that the median rent had shockingly risen from $240 to $336 in two years based on a comparison of the census with real estate advertising rather than the consistent series in the RTA data.

My respect for Anglicare and its worthy mission with the homeless prohibits more robust language on the veracity of this study.

Note: This was published in the fish'n'chip wrapper edition as a "Nick Dalton exclusive". Perhaps that's why Anglicare failed to respond to a polite request for a copy of the study? There are also some other numbers here not verified in the quoted source? Previous post: Homeless numbers daltoned

No comments:

Post a Comment