Wednesday, May 10, 2006

Correlation and Causality

Scoop today fawns over the report Decades of Disparity III by Pete Hodgson which attempts to blame inequality in social circumstances for the inequalities in mortality rates and health outcomes in our society.  This is despite it explicitly realizing: "Socio-economic inequalities explain much of ethnic inequalities in health but do not account for all the disparity."
When you are only able to explain a portion of a statistical correlation but are left with a large remainder co-efficient this is frequently a good sign that you are looking at exactly that - a correlation NOT a cause.  Please note I am not disputing that poorer socioeconomic groups have poorer health - what I dispute is that the statistics PROVE that the lower socioeconomic conditions RESULT in poorer health outcomes.
Being poor does not CAUSE someone to smoke, participate in less sporting activities, or be obese - and yet these are all factors that are ALSO highly correlated with lower socioeconomic conditions - and ethnicity.  I have no issue with a health policy that targets these risk factors, and in attempting to target them in an effective way means different policies may be required for different ethnic and social groups - because they frequently think and act in different ways (hence the correlation in the first place...)  However for this to be effective people need to stop BLAMING the social strata as being the CAUSE of those conditions.  Otherwise these people become powerless to affect the necessary changes in their own lives.  It needs to be recognized that having less money does not CAUSE someone to smoke and/or drink, what needs to be found is why poorer people tend to CHOOSE to smoke / drink and give them the actual responsibility for that choice.
Tags: , , , ,


burt said...

It always stuns me how many people have no idea of the concept of Correlation and Causality. A lot of Govt spin relies on this ignorance.

iiq374 said...

Unfortunatly it doesn't surprise me - although it does dismay me.

Given the number of 2nd year stats students that still got B/C passes without knowing the difference; the general populations understanding is abysmal!

Which is why I've decided to try and point it out a little more when I can.

Anonymous said...

I say briefly: Best! Useful information. Good job guys.

Anonymous said...

Really amazing! Useful information. All the best.