Causation, Correlation and Good Health

By Michael Edmonds 25/08/2012

Whenever I see correlation confused with causation, it reminds me of a time in my early teens when the A&P (Agricultural and Pastoral) show was in my home town. People started getting sick, and the common factor seemed to be attendance at the A&P show. Suspicion started to fall on those providing food at the A&P show, particularly for some reason on the hot dog vendors.

A few days later, the cause of the illness was shown to be norovirus, but not before some rather nasty things had been said about the food vendors. As norovirus is highly contagious, it is more than likely that the close proximity of people at the A&P was more than enough to allow rapid spread of the norovirus. So while there was a correlation between those attending the A&P show, and those get sick, the hot dog vendors were unlikely to be the causation.

This event taught me two things – first, that when it comes to health, we are generally very conservative and cautious. Secondly, and less flatteringly, it taught me how susceptible we can be to scapegoating and behaving out of fear and ignorance.

Our history is littered with examples of scapegoating, for example – disease being blamed on the work of the witches, resulting in the deaths of innocent local women. And this carries through to the modern day where people still die in “exorcisms”.

During the emerging AIDS crisis of the 80’s and 90’s – most New Zealanders will probably remember the name of Eve van Grafhorst, the Australian pre-schooler who was hounded out of her local pre-school in Kincumber, New South Wales, with some in the community even suggesting that the family leave town (which they eventually did, moving to New Zealand where she was able to attend school with relatively few hassles).

Examples of confusion between correlation and causation are apparent in the anti-vaccine movement today. Most children receive vaccines at around the same time as the first  indicators of autism start to become obvious. Consequently while the correlation between vaccines and the development of autism is a temporal one – they tend to occur around the same time, it does not follow that the vaccines are causing autism.

We are extremely cautious when it comes to looking after our health, an approach that in terms of the survival of our species has served us quite well. However, as we grow more knowledgeable we need to learn to balance this precautionary instinct with greater rationality and empathy for those around us.


0 Responses to “Causation, Correlation and Good Health”

  • It does not help, when we have a public health system that is built upon correlative research being passed off as causative.

    • Interesting thought JC, do you have any specific research in mind? Correlative research has it’s uses so long as it is not confused with causative research.

  • My favourite correlation vs causation example, is the strong correlation between ice cream sales and drowning. Clearly, eating ice cream increases your chance of drowning ;-).

  • Michael, high cholesterol causing heart disease is an interesting example of correlative research being passed off as causative…
    Eggs, dairy products and nuts… even coconuts were vilified by the health system as “these things contain cholesterol (eggs & dairy) or saturated fats (dairy & coconut oil) which cause heart disease, therefore they increase heart disease.”

    Now, of course, science has demonstrated that that message was not only wrong, but unhealthful .

  • Ron, if you do a big study looking at peoples’ health and what they eat, and find that people who tend to eat an above avaerage amount of food X also tend to suffer more than average from disease Y, you run into an ethical problem. Should you warn people against food X without knowing if you have correlation or causation? Medical science has tended to be conservative and issue the warning, just in case. You may well be saving lives, and if the warning turns out to have been incorrect, you probably haven’t done any harm.

    Wth science, there is an honest attitude to basing opinions on facts, and changing opinions when more facts emerge change.

  • An interesting theory, possum. Apart from the billion$ spent on propping up the medical industry, and in particular, the pharmaceutical industry, how would you apply your logic to epidemiological studies showing selenium reduces the risk of prostate cancer, or folic acid reducing the risk of bowel cancer?

    An interesting paradox given you’d probably do harm by giving such advice…

  • I take it you are referring to the SELECT study, at this site:

    This was a study to see if increased Vitamin E and selenium intake helped reduce prostate cancer, but seems to have found an increase for men taking extra Vitamin E. However, by referring to Selenium I think you may has misunderstood the findings as at 2011 (paragraph 5), “Men taking selenium alone, or vitamin E and selenium, were also more likely to develop prostate cancer than men taking placebo, but those increases were smaller and are not statistically significant and may be due to chance.”

    I think this well thought out study tends to support my point.

  • Ron, yes the story of the evolving medical views on cholesterol is a fascinating one, particularly how it became politicized in the 60s?
    It is interesting how most nutritional research seems to often come back to a core of “eat a variety of foods and avoid excesses.
    Given the complexity of the human body studying it isn’t an easy task, and knowledge grows bit by bit. What we need to be wary of is any “quick fix” answers and any politicization.
    In science we have to accept that ideas evolve and some initial studies will be thrown out when new evidence comes along.