The Weekly Reading List - June 30

A little brief this week, yet worth the read.

Welcome back to The Weekly Reading List. Before we get into this week’s links, let me ask you a question: are you a fan of this series? This was always intended to be a trial, and since it has been going on for over a month (the first installment was May 19), now is a good time to assess whether it is worth keeping. If I decide against keeping the series, I can spend that time writing other things.

Let me know down below.

A Great Way to Handle Failure

Over at Retraction Watch, a story on a simple mistake that cascaded — and some virtuous researchers who owned up to their mistake.

The paper, “In Sickness and in Health? Physical Illness as a Risk Factor for Marital Dissolution in Later Life,” garnered coverage in many news outlets, including The Washington Post, New York magazine’s The Science of Us blog, The Huffington Post, and the UK’s Daily Mail . 

But an error in a single line of the coding that analyzed the data means the conclusions in the paper — and all the news stories about those conclusions — are “more nuanced,” according to first author Amelia Karraker, an assistant professor at Iowa State University.

Stories of coding errors, poor research methods, and a lack of transparency with other researchers abound on blogs like Retraction Watch. But what was so refreshing about this story was how Karraker and her research team handled it.

After other researchers asked for clarification about some data, as they were unable to replicate the results, it was discovered that a very basic coding error massively skewed the results. Participants who left the study were treated as getting divorced, which made the apparent divorce rate for women with serious illnesses skyrocket.

Karraker and her team immediately shared their code with the other researchers, and when the error was found they contacted the journal to retract the paper and publish a correction (with a much more modest conclusion). It all happened quickly, and it prevented the further spread of bad research.

This is unusual, as many bad science papers don’t get retracted, and too often researchers obfuscate when challenged. But some of the damage is done: I doubt The Washington Post is going to run a story about this or the more modest result actually supported by the evidence.

Some Favorite Substacks

I’m going to take an opportunity to point you toward two of my favorite smaller Substacks. Some of these are totally free, some write almost exclusively behind a paywall, and some mix their content like I do.

First up, Stevereads, by Steve Donoghue. Here’s his latest.