Monday, January 21, 2013

The Suncorp data-unfriendly report?

The big news last week was release of the Suncorp Bank Family Friendly City Report. There was outrage when this report placed Cairns 27th among the most populated thirty cities in Australia. King Parrot, the Lance Armstrong of the Cairns parrot cage, demanded an apology from Suncorp.

The report and response managed to make front page of the Cairns Post two days running. Chris Harrison followed up in the Weekend Post with a critical opinion: Family index simply sticks to the figures. However, Chris didn't question the numbers themselves, quite the opposite:
Numbers are numbers. I'm not questioning the bank's ability to count. What I am questioning is the report's self-confessed disregard for the subjective.
Perhaps he should have questioned the numbers because it is beyond me how this could have been so widely reported nationally without at least some professional journalist observing such obvious flaws. The accuracy of this report is so bad it is surprise it was ever released by Suncorp.

How bad? Well, in the income category they have somehow managed to excise Hervey Bay from Queensland altogether and move it over to Western Australia. However it was the health category which initially had Loose Change head scratching.

Suncorp declare that "Burnie is Australia's healthiest city" with citizens attending the GP barely 1.3 time each year on average. This somewhat clashes with a subsequent health indicator in the report where the citizens of Burnie assess their own health among the worst of the cities. Perhaps this says something about the GP's in Burnie but more likely it indicates that the data is flawed.

What stands out most about the health data is the huge dispersion. The bottom of the scale is Sunshine Coast with 8 GP visits per person. Townsville tops Queensland with just 2.9 visits. Even allowing for demographic differences it should be obvious that this dispersion on a basic health measure is unlikely. The report provides a link to the source of this data at A Social Health Atlas of Australia, 2011.

Data here is in excel file by state for various statistical areas. GP visits are based on a rate per 100,000.  GP visits for Burnie LGA are listed as 526,371 per 100,000 which would be equivalent to 5.3 visits per person. That is a long way from 1.3? Too far away for any difference to be explained by a different statistical area which would all still encompass the main body of Burnie.

The statistical division of Burnie-Devonport is 5.0 visits per person. Scrolling through all the Tasmanian data provideded at the source link whether it is broad statistical divisions, LGA's or smaller local areas there is not a single number within cooee of the 1.3 for Burnie! What have they done?

Meanwhile, up in Queensland, it is claimed by Suncorp that the average citizen of Townsville visits the GP 2.9 times a year while the Sunshine Coast is the centre of Australian ill health with 8 visits per year. Again, the data at the source link contradict this.

The Townsville Council region is 5.3 visits. Statistical division data still splits Townsville & Thuringowa but both are above 5 visits. There is no other data in the source link which supports the Suncorp numbers. Similarly with the Sunshine Coast there is nothing anywhere near a number of 8 visits for the Coast or any of its components.

Cairns is actually above Townsville on this health measure rather than below as in the Suncorp ratings. However, as suggested most urban regions are within a relatively close range on this rather than the extreme diversion implied by Suncorp. Smaller remote regions actually have lower rates of GP visits. Hence, those in the Far North seeking a family friendly destination on this measure should make a beeline for Torres Strait, followed by Kowanyama, Lockhart River and Yarrabah which all have lower rates than Cairns or Townsville!

Lets move on to education where Suncorp would have us believe that Townsville schools are a crowded Dickensian hell with 1,127 kiddies on average packed into each school? There is census data on those at school in the relevant age groups. Also found was a list of schools in Townsville at Wikipedia. This list may not be accurate but doesn't have to be to prove the Suncorp numbers wildly wrong.

A few hundred at primary schools and larger numbers close to a thousand at high schools would probably be expected and this is what the numbers actually show. Suncorps own description of their methodology appears to imply that they included everybody in the 15-19 age group whether they were at school or not. How any sensible and knowledgable person could have looked at these numbers and swallowed them defies belief!

Suncorp regarded disposable income as the most appropriate measure in that category. This would be nice except the broad data they have used from ABS doesn't provide detailed data for regions outside the capitals. Disposable income for each state ex the capitals is applied to all regional centres within each state. So all regional cities in each state get the same score.

This is the biggest clanger in the report because somehow they have managed to move Hervey Bay to Western Austrlalia and rank it at #6 with the highest ranking regional income centres. While not disposable income data, it would have made more sense to use the census numbers on income. This would place Hervey Bay stone motherless last and move it down from #6 to #30. Hervey Bay would then drop out of the paloton as the least family friendly city in Australia!

Do we need to go on? The King Parrot has demanded an apology from Suncorp on the crime and unemployment numbers for Cairns. There are similar issues with the employment numbers but the problem is that the Cairns numbers are actually among the more accurate in the report and wouldn't much change the ranking of Cairns in these categories. Most slighted in the employment category was Ballarat which was given an 8% unemployment rate when the census data is about 5.9%.

Problems with the employment numbers again should have been obvious with reference to Tasmania. Tassie is a relative economic and demographic basket case yet Suncorp managed to rate their two largest population centres in the top ten on employment. The numbers used for Hobart and Launceston, supposedly based on the 2011 census, are simply wrong by a substantial margin.

Cairns employment would move up a few places on appropriate census data but mostly because the numbers for other cities are more wrong. Suncorp used 7.2% census unemployment for Cairns while the SUA (Significany Urban Area) census data is 6.8%. It should be noted that the state electoral division of Cairns is significantly higher than the broad city average at 8.3%!

One could go on but shouldn't. Suncorp should be embarrassed by this report which looks like just another PR spin-job swallowed by the MSM media. However there is no basis for any apology specific for Cairns.



1 comment:

  1. Great work pulling this report apart. Very embarassing for Suncorp.