Slashdot Mirror
Cat Ownership Correlated With Heart Health
Ant tips us to a story making the rounds lately, based on reporting a couple of weeks old, that owning a cat could cut your heart attack risk by one third. No such effect was seen from dog ownership, but the researchers say that could be because there weren't enough dog owners in the study population to provide meaningful statistics. The study: "...analyzed data on 4,435 Americans, aged 30 to 75, who took part in the federal government's second National Health and Nutrition Examination Study, which ran from 1976-1980. According to the data in the survey, 2,435 of the participants either owned a cat or had owned a cat in the past, while the remaining 2,000 had never done so. [The] team then tracked rates of death from all causes, including heart and stroke. Cat owners 'appeared to have a lower rate of dying from heart attacks' over 10 years of follow-up compared to feline-free folk..."
Clearly this proves that people with heart problems choose not to buy cats.
http://alternatives.rzero.com/
"The story is about a common kind of mistake that shows how important it is to understand the scientific method. Someone does a study and finds that there is a statistical correlation between one phenomenon and another. Then there is a claim that one of the phenomena is the cause of the other. Actually, however, they can both be related to something else that is not understood."
It's not a mistake at all, and your example is terrible.
First, what you're talking about is called a Confounding Variable http://en.wikipedia.org/wiki/Confounding_variable
Second, you're making the typical mistake of assuming that because confounding variables are sometimes present that they are ALWAYS present, or not controlled for. Do you know what confounding variables were controlled for in this study before you make the assumptions you have? No you do not.
Third, that ridiculous "correlation does not equal causation" mantra that is so often tossed about is designed like so many other easily remembered but relatively useless memes. It's not a scientific principle, it's a caution, nothing more.
The fact is, most of the time, correlation has some effect on causation. If nothing else, it indicates a relationship worth examining.
"Apparently those who have no strong ties to another person..."
This makes me ask, why denounce his study then do exactly what you denounced it for?
I can't tell you how tired I am of people getting modded insightful for misunderstanding then regurgitating something that most people who discuss this subject should understand at a base level.
There's nothing remotely insightful about restating "correlation does not equal causation".