Causation versus correlation
EDS Skin Care Forums Forum Index » Skin Care and Makeup Forum

free skin care · Thu Feb 19, 2009 9:42 am

There are lots and lots of studies out there which show that people with a particular characteristic or behavior have higher rates of certain diseases or health issues. Seeing this kind of information--particularly if you have that characteristic or behavior--can be alarming and raises questions about whether we should stop using or doing whatever it is that is linked with the bad consequences.

We can be more informed consumers if we look critically at the study to determine whether it seems reasonable to change our behavior to avoid the bad consequence--cancer, skin aging, or whatever. That involves determining whether the study likely shows causation or just correlation. If the study shows that x *causes* y, then it may make sense to avoid x, at least if the risk of y is significant and the benefit of x is comparatively minimal.

On the other hand, correlation just means that, when you look at a group that does x, we see greater amounts of y than you would see in a random population. Sometimes that indeed means that x causes y, but sometimes is unrelated, so that stopping to do or use x won't help prevent y at all. Here's a real world example: A recent study showed that people who had only cellphone service and no landline phones had more than twice the binge drinking rates of people who had cellphones and landlines. Does this mean that we could reduce binge drinking by encouraging people to get landlines? Of course not. It turns out that those with cellphone service only tend to be a lot younger than those with both cells and landlines, and younger people have higher rates of binge drinking. Here there is a correlation (phone type and drinking pattern) with no causation.

One way to think about this in the context of skincare studies is to ask, "Did the scientists consider other factors that could have caused the result, and did they regress out those competing explanatory factors? Is there a theoretical mechanism that would explain why x would actually cause y?"

Hope this helps evaluate information we get about health issues.

Thu Feb 19, 2009 10:46 am

Interesting topic, Flitcraft....

free skin care · Thu Feb 19, 2009 11:41 am

This is very true, flitcraft. We should always have what you say in mind and question whether there really is a causation or not when reading about findings from studies. I think media far too often reports on results from studies without questioning whether there exists a causal connection or not and without asking critical questions. The reason may be that many journalists have too much respect for scientists.

I remember a very stupid study from my home country a while ago. A researcher was interviewed on the radio, warning people against searching on the Internet for information about symptoms, diseases and diagnoses if they felt ill. Why? Because a study had shown that people who were actively searching for information about diseases and symptoms on the Internet, were less likely to recover from their disease than those who did not.

I was very surprised to hear researchers jumping to conclusions like that. I think it's obvious that people who suffer from a disease that is easily cured go to the doctor and get cured, whereas patients with long-term symptoms or rare diseases that are not easily cured, turn to the Internet for information. I don't think googling symptoms makes you worse or kills you. If that is the case, it must be the radiation from the PC that makes you ill.. Shock

Automatic Message

If this is your first visit to the EDS Forums please take the time to register. Registration is required for you to post on the forums. Registration will also give you the ability to track messages of interest, send private messages to other users, participate in Gift Certificates draws and enjoy automatic discounts for shopping at our online store. Registration is free and takes just a few seconds to complete.

Click Here to join our community.

If you are already a registered member on the forums, please login to gain full access to the site.