Monday, 10 September 2012

The worst abuse of statistics by Rev Peter Jensen

Tonight on Q and A, we witnessed one of the worst abuses of statistics - and it causes real harm.

A correlation is not causation

Peter Jensen - you sinned by misusing statistics
Anglican Archbishop Peter Jensen asked for a sensible debate about the health issues of being gay, on the basis that gays have shorter lifespans on average than non gay people.

This is the most heinous misuse of statistics.  A correlation (when two things coexist at the same time) does not mean that one thing causes the other.

Observe the following facts:
- People who live in Africa have a shorter lifespan than people who live in Australia
- People who are overweight have a shorter lifespan than average
- Men have a shorter lifespan than women
- People who are gay have a shorter lifespan than average
- Indigenous people have a shorter lifespan than average
- Smokers have a shorter lifespan than average

Does this mean...
- Africans are genetically weaker?
- If you are overweight, you should lose weight to live longer?
- If you are a man, you should become a woman to live longer?
- If you are gay, you should become straight to live longer?
- Indigenous Australians are at fault for their shorter life spans?
- To live longer you should stop smoking

Clearly, some of those statements are nonsensical, some are provocative, but they are all equally plausible (or implausible) from the data.  None of them should be stated without more facts.

What is really going on?
  • Africans and indigenous Australians have a poorer standard of nutrition, water and healthcare among other things.  It is these things that cause shorter life spans.  The reasons that these groups and others around the world have these conditions is complex and varies from location to location.  However, this does not means that Africans or indigenous Australians are at fault for their shorter life spans.
  • People who are overweight, and men have a higher risk of certain conditions which end your life.  Diabetes and heart failure are two of these.  Would losing weight reduce your risks? Yes, controlled studies have shown this.  Would becoming a woman reduce a man's risk? Who knows - there have been insufficient studies of transsexuals to evaluate?
  • Gay men live on average shorter lifespans than average.  Why? There are a myriad of possible factors including higher rates of mental illness, which does shorten life spans.  Why do gays have higher rates of mental illness? There are many contributing factors, but the suggested solution that gays should try to not be gay is more likely to increase mental illness than decrease it.

  • Smokers have shorter lifespans, because smoking increases the risk of lung cancer.  Lung cancer will end your life.  Therefore: smoking - causes - lung cancer - which causes death.  This is the only one of these statements that is a proven causation.

Misuse of statistics is not an accident

Peter Jensen, you are a learned man.  You studied the rigorous discipline of law, you have multiple degrees.  You know the difference between a correlation and causation.  You know that you are being deliberately mischievous when you were asked to comment on Jim Wallace's quote that included...
"The life of smokers is reduced by something like seven to 10 years and yet we tell all our kids at school they shouldn't smoke...But what I'm saying is we need to be aware that the homosexual lifestyle carries these problems..."

He is trying to argue that there is a causal relationship between both smoking and death and a gay lifestyle and death.  There is a relationship for smoking.  There is no relationship for a gay lifestyle.

You know the difference Peter Jensen.  You know.

Peter Jensen, your knew your comments would get traction. You misused statistics deliberately.  That is sinful. Christ would not have done that.

No comments:

Post a Comment