The authors of a 2017 paper on emotional and behavioral gaps between boys and girls have retracted the article after discovering a coding error that completely undermined their conclusions.
The revelation prompted the researchers to republish their findings in the same journal, this time with a title that flips the narrative.
The PsychJournal study, first published in March, looked at self-regulation — loosely defined as the ability to get stuff done and keep a lid on it — in boys and girls in German elementary schools. Although previous studies had found girls might do better on this front, the authors, from the University of Leipzig and New York University’s Abu Dhabi campus, initially found the opposite:
Sex differences in executive functioning and behavioral self-regulation were found, suggesting an advantage for boys. …Findings suggest that the male disadvantage in self-regulation documented in previous studies might be specific to characteristics of the sample and the context in which the data were collected.
Strike that: According to the (arguably understated) notice, a major error forced the authors to change the title of their paper from “Boys have caught up” to “Boys have not caught up:”
The retraction has been agreed upon due to an error that may influence the result and accuracy of the content, which would have implications for how readers use the article.
The first author, Catherine Gunzenhauser, told us:
We discovered the error when we worked with the same dataset to look at a different research question not related to the focus of the PsychJournal paper.
In these analyses, we found a gender difference (in a variable for which gender differences had not been investigated for the PsychJournal study) that seemed very strange, because the direction of the gender effect we found was clearly at odds with a large body of literature. This raised my suspicions, and as I could not find any error in the electronic data file, I went back to the paper questionnaires. After some research I found there had been a systematic coding error during data entry (the 0/1 coding for “boy” and “girl” that we had on the paper questionnaires was opposite to the one in the SPSS labels). This error was unfortunately overlooked when the coding was double-checked.
We then immediately contacted the editorial office of the PsychJournal in order to ask how to proceed to make the error public.
Gunzenhauser, who guest edited the issue in which the article appeared, said she and her colleagues “do not expect” any other retractions or errata related to the error.
The initial title of the paper plays off a 2015 article by Gunzenhauser and von Suchodoletz in Early Education and Development, “Boys Might Catch Up, Family Influences Continue: Influences on Behavioral Self-Regulation in Children From an Affluent Region in Germany Before School Entry.”
Walter Edwards, managing editor of the journal, told us:
Since the corresponding author of the article was also a Guest Editor of a Special Collection in which it appeared, and had worked admirably to put together an outstanding collection of contributions, the journal had no reason to believe that the error was anything but unfortunate oversight, purely unintended. Accordingly, no additional scrutiny was felt necessary, and the revised version (pchj.180) was accepted after the same level of review process as the initial submission.
Hat tip: Rolf Degen
Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up on our homepage for an email every time there’s a new post, or subscribe to our daily digest. Click here to review our Comments Policy. For a sneak peek at what we’re working on, click here.
This kind of error occurs when coding is used, and when the coding is changed to non-standard coding. When non-standard coding choices are made, it is easy to slip from the correct interpretation (which is not typical) to a typical interpretation which is not correct. This is the reason why the CDISC effort was developed. Using a standard coding approach for standard variables (Yes = 1, No = 0), a more interpretable and more correct data file will be set up. To avoid these kinds of problems, use standard codes for standard variables.