The article, “When Words Hurt: Affective Word Use in Daily News Coverage Impacts Mental Health,” was published in Frontiers in Psychology in August 2018. The study has been cited six times, according to Clarivate Analytics’ Web of Science. In March 2020, an article in The Conversation used the study’s findings to argue that kids should reduce their television intake during the coronavirus pandemic to ward off anxiety.
First author Jolie Wormwood, an assistant professor of psychology at the University of New Hampshire, said she decided to pull the study after revisiting the dataset. She found that some of the study participants—95 people in the Boston area—who completed a questionnaire three different times during a nine month period, gave inconsistent answers about their memory of an event. That normally might not be too worrying, since memories “shift over time”, according to Wormwood, but a bit more sleuthing revealed that the researchers had inadvertently mixed up the IDs that were assigned to study participants.
Wormwood explained the error in an email:
Tracking down the information needed to re-align the data correctly was no easy feat, particularly during the pandemic, since I’d changed Universities and identifiable data associated with the project was kept in hard-copy only in a locked lab out of state. But my co-authors and collaborators on the paper were all incredibly gracious and supportive of figuring out what happened and correcting the record as needed.
It was obviously very disappointing to discover that the observed relationships were no longer significant in the corrected data set. We worked to put in the retraction as quickly as possible, though it took much longer to process and be finalized than I anticipated. However, Frontiers was very understanding and pleasant to work with throughout the process. We are hoping to explore similar hypotheses in a larger data set in the future.
When Wormwood discovered the error, she said, “I don’t think I slept for a day and a half while I was frantically trying to (dis)confirm my suspicions that something was wrong with the logs we used to match participant IDs across time points.” The authors asked for the retraction in July 2020, Wormwood said; the journal accepted it in November and published this notice in December:
This paper reports a study that involved data collected at three time points. The authors recently discovered a clerical error in how the data were merged across time points. Specifically, data collected during the final longitudinal time point (wave 3) were incorrectly assigned to the wrong participants when merging data files. The authors have corrected this error and re-conducted the analyses and the pattern of findings reported in the paper are no longer supported by the analyses.
The authors concur with the retraction and sincerely regret any inconvenience this may have caused to the reviewers, editors, and readers of Frontiers in Psychology.
Gilad Hirschberger, a psychology professor at the Interdisciplinary Center Herzliya, in Israel, who was the editor on the original study, told us in an email that:
The authors initiated the retraction after they discovered a fatal error in their data. Once they asked to retract the paper, the journal notified me.
Hirschberger also commended Wormwood’s discovery of the error and decision to pull the study:
It’s reassuring to know that there are such honest people in our field who have the courage to admit that they were wrong and pay a price for it.