Monday, May 06, 2019

Second Fundraising Post For 2019. Retractions and Corrections of Studies Are Mostly Invisible in Social Media.

This year my fundraising will be more than a week, but on the other hand you won't get these begging posts every day!  Win-win.

My heartfelt thanks for all of you who read and comment here.  If you can afford to chip in for the costs of maintaining this blog and like it to go on for a while longer, please check the instructions in the left column on how to give.  Thank you.

I'm reading Gina Rippon's book The Gendered Brain.  The New Neuroscience That Shatters The Myth Of The Female Brain.  There will be a proper review of the book later, but today I want to quote a longer bit from the book, because it ties to some of my own earlier findings about the way research is popularized and about the way corrections and retractions to such research are not popularized.

The particular study Rippon describes was about studying the brain structures of 143 one-month old infants (73 female, 70 male) with a high-resolution scanner.  Because one hypothesis argues that many sex differences in the brains are innate, getting information about very young babies can help us test it.  The results and what happened next:

The authors of this paper reported marked sex differences in total brain volume, grey matter and white matter.  Again, this was quickly disseminated into the public realm, this time by an online research summary source that pitched this report as an important breakthrough in the search for explanations for female-male differences in behaviour.  The source concluded that 'pretending these early sex differences in the brain don't exist will not help us make society fairer.'

The trouble was that the reported findings were actually wrong.  Although the researchers claimed to have corrected for brain size, an eagle-eyed neuroscientist noted that the data in the paper weren't consistent with this claim.  The authors were contacted, rapid checking and reanalysis followed and all the claims of significant differences disappeared.

A correction was quickly issued, published on both the journal's and the research digest's websites.  But there was a two-month gap between these events, and social media had already pounced.  Reference to the paper had already appeared on Facebook with one telling comment:'I actually had an argument about this with someone who claimed to have a degree in the field very recently.  Rubbing her face in this will make me so happy.'
In these days of ideological echo chambers it is the fake news, or in this case fake neuronews, that stick around, even if later disproved.
I include that long quote, because the story here is very similar to what happened to another study, also on purportedly innate sex differences.  I wrote about that in 2014.

Studies which appear to support various ideological beliefs are widely disseminated in social media, because they are click-bait.  Both those who hold those ideological beliefs and those who absolutely do not will read such research summaries.  But those who disseminate these studies tend not to check if the studies are later retracted or corrected.  When that happens, the popularization process leaves people misinformed.

This process is not a symmetrical one in the study of sex differences.  Studies which show sex similarities* almost never get the kind of online media boost that studies showing sex differences do, never get a thousand comments in the New York Times or the Guardian, and so on.
*  An additional reason making this bias worse is that sex similarity findings are often not published at all, and if they are, they are not stressed as one of the main findings of a study.