Correlation is not causation (#1124)

Topics/tags: Rants

Correlation is not causation.

It’s the Statisticians’ mantra [1]. You think people who do research would have mastered it by now. But we keep seeing studies and newspaper articles that connect the two. When it’s news articles, I realize that it may be the traditional issue of reporters not understanding the core issue [2], rather than the researchers’ fault, but it’s still frustrating.

When I see an article in which correlation is used to imply causation, I usually let it pass. But there are times that it just sticks in my craw [3]. Yesterday [4], Inside Higher Education had an article about a study of scientists’ behavior during the pandemic.

Here are three paragraphs of interest.

Some 28 percent of scientists said they wore pajamas at least once a week – a cohort who were twice as likely to report worsened levels of mental health than those who dressed normally each day, according the study, by David Chapman and Cindy Thamrin, from the Woolcock Institute of Medical Research, which is affiliated with the University of Sydney.

Thamrin told Times Higher Education that the finding was consistent with evidence supporting blue pajama syndrome, in which hospital patients who remained in bedwear during longer stays were assessed as being more depressed than those who changed into day clothes. However, those who wore pajamas did not report lower levels of productivity, she added.

To those who choose to continue working from home at least some part of the week, we would hope to reassure them that the occasional day in pajamas won’t affect their productivity, but perhaps they should consider changing out of them as a matter of routine for the sake of their mental health, said Thamrin.

That’s a huge leap, isn’t it? Wouldn’t your first assumption be Those who are depressed are less likely to make the effort to change their clothes? At least in that case, there’s an obvious connection. Why would wearing pajamas [5] make you depressed? Wearing soft clothing makes many people happy.

I suppose there could be some evidence that taking positive steps can improve your mood, but it doesn’t appear that this study shows that.

So if you’re a scientist and working from home, don’t worry about what you are wearing. But if you’re a scientist working from home, you probably knew that correlation is not causation. If you’re in a non-statistical discipline and read the article, don’t rely on this evidence. Let me go a step further. Whoever you are, if you find yourself more depressed and you have the resources to do so, consider one of the more common approaches to treating depression: seek therapy, talk to a counselor (religious or otherwise) or a friend, visit with your physician to have them prescribe appropriate medicine, etc.

It’s times like this that I miss the Inside Higher Ed comments section [6].

[1] I suppose statisticians have other mantras, too.

[2] Although one would hope that those charged with reporting on science would be expected to have basic scientific and statistical literacy.

[3] This evening, I’m too lazy to do any deep exploration of the origin of the meaning and origins of the phrase, other than sometimes you try to take something in, and you can’t swallow it.

[4] Or at least it was yesterday at the time I wrote this musing.

[5] Or pyjamas, if you’d prefer.

[6] Or maybe not. I’m sure there would be a bunch of people writing something like, If it weren’t for the Democrats promoting the coronavirus myth, these people would be at work where they belong and not home in their pajamas.

Version 1.0 of 2021-01-23 .