Science

I saw this headline:

Feeling young at heart wards off death, scientists find

The headline was followed by these sentences:

Researchers at University College London found that those who felt younger than their actual age were 41 per cent less likely to have died in the follow up period.

And

Scientists have proven that people with a youthful spring in their step and an unswerving optimism about the future seem able to cheat death.

I see this kind of thing (“scientists have proven …”) on the Internet all the time, and I am going to write about it because it touches on one of my pet peeves: journalists who don’t understand science but write about science.

I’m fairly sure that the researchers at University College London made no statement to the effect that feeling youthful helps a person live longer. If they did make such a statement, they should be sent back to school to study science.

The gist of the article was that researchers followed more than 6,000 people for eight years monitoring their happiness levels and health, and they found that happy people tend to live longer than unhappy people. In the field of epidemiology (the science that studies causes and effects in health and disease processes) this kind of study is called an observational study.

I’m sure many people have wondered if happier people live longer than less happy people. Framing this question scientifically, we’d like to know:

• Is there an association between happiness and life expectancy?
• If there is an association, does happiness help us live longer or is something else happening?
• How does happiness help us live longer?

An observational study usually begins by collecting data. In this study, the data is the happiness and health of 6,000 people. Collecting data allows us to observe associations that develop over time. After collecting and carefully analyzing this data, the researchers reported that they observed a correlation in the data which can be stated: happiness is associated with longer life. Let  me stress that at this point, nothing has been proven.

When we see an association between two variables (let’s call them A and B) we might assume that A causes B. But perhaps B causes A. Or, perhaps both A and B are caused by a third variable, which we’ll call C. In this case, C would be what is called a confounding variable.

After running an observational study and finding an association, the next step is to form a hypothesis – a possible explanation. The hypothesis must be testable. To test the hypothesis, we conduct experiments and collect data from the experiments; this process is called a clinical study. Then we analyze the data and reach a conclusion. Not until all this is done do we have anything remotely like “proof.” Even then, nothing is considered proven until other researchers repeat our experiments and reach the same conclusions.

It may seem reasonable that feeling happy could make us live longer, but perhaps having better health is the confounding variable that makes people happy and helps them live longer. Or perhaps having more wealth is the confounding variable that makes people happy and, by allowing them to eat healthier foods and have access to better health care, helps them live longer. The researchers acknowledged their study had not proven anything when they stated “The mechanisms underlying these associations merit further investigation.”

So folks, when you read a headline that says, for example, “Eating an egg a day is associated with diabetes,” just remember: words like associated with or correlated with mean that you are reading about an observational study, and observational studies are not proof of anything! To repeat: such studies only allow us to form a hypothesis, which must be tested by a clinical study, and the results of the clinical study must be reproducible. An observational study is the beginning of the road, not the end. Don’t let sensational health headlines worry you. Read carefully and apply at least a little skepticism.

Does every journalist get every story one hundred percent correct? Unlikely. Don’t be the person who thinks, “I read it on the Internet so it must be true!” It may be better to think, “I read it on the Internet so some of it may be true.”