Thread by @lilacCourt

our @PALS_SCALES paper showing language, but not general non-verbal cognitive ability, longitudinally predicts emotion recognition from face & voice cues has finally been published: peerj.com/articles/9118/ 1/

This is fabulous work led by @SarahGriff90 who has shown amazing persistence and resilience in what has been the most dispiriting journey to publication in my working life. 2/

we started over two years ago by asking to submit a registered report to Developmental Science - they said no b/c it was a longitudinal study and we'd seen earlier data. 3/

undaunted, we did our best to make this the most robust study we could:
pre-register hypotheses/analysis plan on OSF
test 350+ children in a population study
make all experimental materials open
make all data and code open
very measured in conclusions 4/

what I love about @PALS_SCALES is it is a pretty inclusive population sample - ~350 kids age 10-11 did the emotion task and there is a huge range of language and cognitive ability in our sample - few studies include such a range 5/

despite all of this, the paper was desk rejected by Dev Sci for being 'too clinical' - a frankly shocking outcome given they regularly publish much smaller studies of clinical groups such as autism, dyslexia... 6/

we have since submitted to three more journals - nine people have reviewed this paper, seven reviewers were extremely helpful and positive, two going so far as to explicitly state it should be published. 7/

we submitted to a prestige journal that prides itself on open science and badges. But we had one reviewer that wrote about three lines completely trashing our paper. This person did not like our tasks 8/

"non-verbal measures and/or measures of moment-to-moment dynamics of emotion recognition could have increased the extent to which this manuscript made a unique contribution to the literature" 9/

however, no reference or link to such tasks was given, no justification for why such tasks would be better than the ones that we used, no consideration of whether or not children with clinical conditions could engage with such tasks 10/

but one such review is enough to sink the paper. I did appeal but was told our work 'was just not good enough' (it never works - does appeal ever work for women)? 11/

when I talked to other academics about this they either said:

1. don't worry - good they reviewed it but getting in that journal comes down to luck of reviewer draw
2. known biases in publishing - more likely to publish men/from famous labs 12/

but then they also said would be much better for career/grants to publish in these prestige journals! WHY? Why do we reward luck and bias? 13/

Luckily one of the last pre-lock down things I did was attend @BrianNosek lecture @UCL where I was reminded that UCL is a signatory of DORA - where we publish does not matter as much as what we publish. 14/

and he reminded us that senior academics can make a difference in making sure that we change our reward structures and incentives. I'm a senior academic, and here is what I'm going to do... 15/

I'm not going to prioritise prestige journals - I will look for journals that value sensible peer review and evaluate what we've done, not what might have been possible in some fantasy alternative universe 16/

I will emphasize excellence science & contributions to the field when introducing speakers, not how much or where they've published 17/

I will remind hiring and promotions committees of our commitments to open science & DORA when evaluating candidates. I've just been on @UCLBrainScience promotions committee & am pleased that where papers were published has not featured in any discussions 18/

I'm also on REF... 19/

a side note - it would have been great to have comments on our tasks before we did the study - which is why we tried to go down registered report route. BUT I don't think what was suggested would have answered our question or had been practical. 20/

in which case we would have dropped emotion recognition from our study & that would have been a shame. We cannot answer a causal question, but establishing that early lang is important for later emotion allows us to think about mechanisms that we could test in later work 21/

anyway, I hope that our data will be useful. the take home: many children with language deficits find it hard to recognise emotional states from faces/voices. other work also shows they struggle to predict emotion responses from situational cues. 22/

clearly this has implications for social interaction and lots of thoughts now in the @PALS_SCALES team for what we can do about it. 23/

It has been a brutal process but I am proud of this work and so privileged to work with @SarahGriff90, the whole @PALS_SCALES team & the young people in the SCALES cohort who are teaching us all so much! END

PS: can I just say our experience of PeerJ was brilliant - quick, open review (my first experience of this) that was focused and extremely helpful. Will do it again and highly recommend.