I hope this exchange of emails won't act as a trigger for Allstat to sprout a series of such cases of limited thought - somewhat analogous to the RSS 'Forsooth!' column which lampoons obvious failures by the media etc. in numerical and graphical presentation of statistical data. More constructively, one of the identified roles of public health practitioners in the UK is advocacy, and much the same applies to statisticians - we should see it as part of our role to help others to understand how with observational data, the possibilities of reverse causation and substantial confounding are the norm, not the exception. Personally, I'm not particularly surprised that neither the media nor a charity's research arm had considered such alternative explanations. In my experience, few other than those with a statistical or epidemiological mindset and experience tend to reason this way - unless there's an ulterior motive to fire the neurons. The tobacco industry has often claimed that it's people who are more susceptible to cancer who tend to smoke. While this is theoretically possible, it's very difficult to take such an argument seriously - the degree of plausibility of a particular effect being causal must always be borne in mind. Moreover, Mendelian randomisation studies can now unpick and finally debunk such arguments, and will be increasingly used to do so - in relation to disease aetiology and physical/chemical/pharmaceutical causative agents - though its applicability to sociological issues such as the relationship between teenage drug use and family breakdown seems much less feasible.
Robert G. Newcombe PhD CStat FFPH
Professor of Medical Statistics
Department of Primary Care and Public Health
Centre for Health Sciences Research
Cardiff University
4th floor, Neuadd Meirionnydd
Heath Park, Cardiff CF14 4YS
Tel: 029 2068 7247
Fax: 029 2068 7236
Home page http://www.cardiff.ac.uk/medicine/epidemiology_statistics/research/statistics/newcombe
For location see http://www.cardiff.ac.uk/locations/maps/heathpark/index.html
>>> "David B. Klein" <[log in to unmask]> 10/09/07 14:13:54 >>>
Precisely. I was surprised at the BBC and the study's investigators
for accepting that one-sided causal interpretation so uncritically.
At 04:15 AM 9/10/2007, Robert Newcombe wrote:
OK - but this disregards the real possibility of reverse causality,
that failing/nonfunctional partnerships may well predispose to
teenage drug abuse. I would guess that both effects are operating
here - a traditional vicious circle.
>>> "David B. Klein" <[log in to unmask]> 28/08/07 03:31:27 >>>
"Drug abuse [by teenagers] causes family splits "
http://news.bbc.co.uk/1/hi/uk/6966086.stm
|