Chrome Prince
31st July 2012, 10:47 AM
Please indulge me, you will see the relevance to betting systems at the end.
Today a media release was made saying that a government survey had linked dodgy backs to psychological problems.
As a statistics man, I cannot refrain from looking deeper and finding the flaws.
The subject matter is also of interest to me.
It seems every time the government releases figures, they are skewed or flawed in some way, we punters would call it backfitting.
The conclusions drawn are as bad as researching Moonee Valley barrier stats and applying them to Flemington.
The first alarm went off when all they referred to were percentages and not numbers.
The second alarm went off when they wrote that Indigenious Australians were 25% more likely to report having back problems.
How do they know what people are thinking, how do they know who is more likely to report something?
Could it possibly be that Indigenious Australians are actually 25% more likely to have medical problems because of their lifestyle, location, lack of medical treatment, financial disadvantage etc etc.
No, perish the thought!
Next theory.... isn't it possible that a back problem, or any other chronic problem is going to increase anxiety or depression. Anyone with a chronic injury is going to have some degree of depression, whether it's back related or not.
But again, they have somehow linked back pain with depression.
I could list 20 medical conditions that lead to depression.
But enough of the ramblings, I took a peek at the data which is not available anywhere on the media release, nor anywhere on the AIHW website that is referred to.
So I had to delve into the government Bureau Of Statistics catalogues, and what I found was astounding.
This does relate to Betting Systems, the parallels are remarkable and I believe that punters are more likely to report more accurate figures than government agencies :D
What I found was the worst case of manipulation I've ever seen.
Firstly, they have taken data from 2007/8 and linked it to data from 2004/5 and linked 2004/5 data to 2009 data.
What possible hope of generating any kind of accurate picture.
They haven't just taken three snapshots, but linked the data and drawn conclusions.
In other words, as an example, it's like examining betting trends in 2009 based on turnover or races in 2004.
Oh it gets much much worse...
There is a tiny notation down the bottom that there is a 25% margin for error
in the data, yes folks a quarter of the data is unreliable!
O.K. not too bad if they have millions of records I guess, a bit of a worry if they have thousands of records, a major worry if they have hundreds of records.
Here's the clanger, in Victoria, guess how many records constitute non-Indigenious Australian born people?
1.2, that's right 1.2 people!!!!!
That's like saying all horses wearing green silks which won last start are winners because they surveyed one grey horse at odds of 100/1 and it was a winner.
83.3 people were born in other countries.
What a sample size.
I'm sure the pom pom system would be more accurate than these clowns.
North Africa, Middle East, South East Asia was 27.6 people, I'm getting a picture here, a great majority of underprivileged people have depression and medical problems, but even that is a wrong conclusion because you cannot possibly draw any conclusions from a sample size of 84.4 people.
The only possible conclusions one can draw is that:
a) The punters on this site would be more accurate employees of the Australian Bureau Of Statistics
b) Employees of the ABS should never devise betting systems.
Today a media release was made saying that a government survey had linked dodgy backs to psychological problems.
As a statistics man, I cannot refrain from looking deeper and finding the flaws.
The subject matter is also of interest to me.
It seems every time the government releases figures, they are skewed or flawed in some way, we punters would call it backfitting.
The conclusions drawn are as bad as researching Moonee Valley barrier stats and applying them to Flemington.
The first alarm went off when all they referred to were percentages and not numbers.
The second alarm went off when they wrote that Indigenious Australians were 25% more likely to report having back problems.
How do they know what people are thinking, how do they know who is more likely to report something?
Could it possibly be that Indigenious Australians are actually 25% more likely to have medical problems because of their lifestyle, location, lack of medical treatment, financial disadvantage etc etc.
No, perish the thought!
Next theory.... isn't it possible that a back problem, or any other chronic problem is going to increase anxiety or depression. Anyone with a chronic injury is going to have some degree of depression, whether it's back related or not.
But again, they have somehow linked back pain with depression.
I could list 20 medical conditions that lead to depression.
But enough of the ramblings, I took a peek at the data which is not available anywhere on the media release, nor anywhere on the AIHW website that is referred to.
So I had to delve into the government Bureau Of Statistics catalogues, and what I found was astounding.
This does relate to Betting Systems, the parallels are remarkable and I believe that punters are more likely to report more accurate figures than government agencies :D
What I found was the worst case of manipulation I've ever seen.
Firstly, they have taken data from 2007/8 and linked it to data from 2004/5 and linked 2004/5 data to 2009 data.
What possible hope of generating any kind of accurate picture.
They haven't just taken three snapshots, but linked the data and drawn conclusions.
In other words, as an example, it's like examining betting trends in 2009 based on turnover or races in 2004.
Oh it gets much much worse...
There is a tiny notation down the bottom that there is a 25% margin for error
in the data, yes folks a quarter of the data is unreliable!
O.K. not too bad if they have millions of records I guess, a bit of a worry if they have thousands of records, a major worry if they have hundreds of records.
Here's the clanger, in Victoria, guess how many records constitute non-Indigenious Australian born people?
1.2, that's right 1.2 people!!!!!
That's like saying all horses wearing green silks which won last start are winners because they surveyed one grey horse at odds of 100/1 and it was a winner.
83.3 people were born in other countries.
What a sample size.
I'm sure the pom pom system would be more accurate than these clowns.
North Africa, Middle East, South East Asia was 27.6 people, I'm getting a picture here, a great majority of underprivileged people have depression and medical problems, but even that is a wrong conclusion because you cannot possibly draw any conclusions from a sample size of 84.4 people.
The only possible conclusions one can draw is that:
a) The punters on this site would be more accurate employees of the Australian Bureau Of Statistics
b) Employees of the ABS should never devise betting systems.