Friday, January 6, 2017

Logical speculation vs. the empirical data

Hmm. From Warning: You Still Need to Stress About Stress by Peter R. Orszag.
Differences in life expectancy by education and income are increasing. And while differential trends in smoking and other habits partly explain the lifespan gap, it's mostly a puzzle. I suspect that a major factor is stress -- that prolonged exposure to stress has risen more among families of moderate income than it has among those of high income, and that the stress is literally killing people.

A new study raises doubts about this hypothesis. The researchers found that British women aged 50 to 69 who said they were unhappy or felt as if they had no control over their lives were more likely to die over the next decade than those who did not. But when the researchers adjusted for self-reported health at the beginning of the study, the correlation effectively disappeared.

The researchers concluded that the women in worse health had shorter life expectancy, and that their health problems were also what caused them to feel stress and unhappiness. As the lead author, Bette Liu of the University of New South Wales, commented, “Illness makes you unhappy, but unhappiness doesn't make you ill.”
Translation: More educated and richer people live longer than poorer, less educated people, creating a Longevity Gap. There are many potential causes of the Longevity Gap. One of those causes might be that poorer people are more stressed than richer more educated people. It is logical that that might be a cause. A new study indicates that stress is not a cause. There are three possible design flaws in the study which might challenge their conclusion but those flaws are speculative.

With such weak evidence, what is the basis for the headline, "Warning: You Still Need to Stress About Stress?" The evidence suggests otherwise. The most likely explanation for the difference between the body of the text and the headline is 1) Headline writers are different from authors, 2) Headline writers are more focused on clicks than consistency, and/or 3) the author is subject to motivated reasoning (he wants the logical speculation to be supported by the empirical data.)

All three explanations might be in play simultaneously. A striking example of letting logic outstrip the evidence.

No comments:

Post a Comment