What does this mean, in practical terms? Well, they used a modified M-CHAT scale, which gives you a 0-20 score. Given that the average score for this scale is pretty low (<10), a 4.2% increase would be a fraction of a point Pretty meaningless stuff, that
-
Show this thread
-
It's also worth noting that, while the study did control for some demographic factors, they didn't look at many I can think of ~dozens~ of other things that may have contributed to potential ASD symptoms in kidspic.twitter.com/cvmipnP7os
1 reply 0 retweets 6 likesShow this thread -
On top of this - and not reported in the press at all - is the fact that the authors actually conducted TWO analyses One was linear (we'll get back to this) and positive. The other looked at a yes/no autism-consistent score for diagnosis
1 reply 0 retweets 2 likesShow this thread -
Everyone reported the positive linear analysis, but no one noted that actually, in terms of A CLINICALLY USEFUL DIAGNOSIS OF AUTISM, the study was entirely negative No association between screens and autism at all!pic.twitter.com/aPb02ebRaI
2 replies 1 retweet 6 likesShow this thread -
Somehow, it gets worse The measure of "screen time" used in the study? Turns out it...wasn't great
1 reply 0 retweets 0 likesShow this thread -
Specifically, they asked people "Does your child watch TV/DVDs?" as a yes/no answer, and then asked "“Over the past 30 days, on average,how many hours per day did child watch TV and/or DVDs?” In 2010-2012
1 reply 0 retweets 3 likesShow this thread -
Now, I'm no childhood autism researcher, but I'm pretty sure that asking only about TV and DVDs in 2010-12 would've missed a huge swathe of potential screen time For example, the first iPad came out April 2010
1 reply 0 retweets 4 likesShow this thread -
But this isn't the last huge potential issue here Remember when I mentioned the LINEAR regression that spat out all the positive results? Yeh, about that
1 reply 0 retweets 0 likesShow this thread -
The variable that they were using this linear model to predict was a 0-20 numerical scale made up of integers (+1 for every 'yes' answer), likely clustered around low values This is not an ideal target for linear regression
1 reply 0 retweets 2 likesShow this thread -
It's essentially an ordinal outcome - an ordered scale from 0-20 indicating autism risk - and without any transformations (which were not reported) it's unlikely that the linear model used was appropriate for this analysis
2 replies 0 retweets 0 likesShow this thread
It's also worth remembering that this study DIDN'T LOOK AT FORMAL AUTISM DIAGNOSIS Even the more clinically useful variable was yes/no on a relatively arbitrary scale of autism risk
-
-
The only real take-home from this study is that the children of parents who report that they watch more TV/DVD may have a tiny increase in their risk of autism-like symptoms but who knows because the stats were a bit dodgy anyway
1 reply 0 retweets 5 likesShow this thread -
TL:DR - study wasn't great - didn't show that screens cause autism - barely even showed an association - don't worry about screens and your kids too much, there's a pandemic on
2 replies 1 retweet 12 likesShow this thread
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.