Alright, big RCT time! Denmark did a big RCT on some preschool intervention in 2015. n=5436. We set a reminder to wait for their follow-up. That's now.
So, their test scores on verbal skills went up immediately after intervention ended.
And then, as you saw, there was NO MAIN EFFECT in follow-up! 😨
Not to worry though! Researchers then found a subgroup with p < .05. 😌 Kids with parents "without college" (defined via <14 years of education) had d = 0.26 improvement.
Did they do a formal interaction test or the subgroup fallacy? It seems it's the latter. Their p values in this subgroup is not even p < .01 despite n=597 kids.
Things got even worse when they tried adding some covariates. Look at that error bar. 90% confidence interval. Lols
So all in all, they found nothing that would convince a skeptical reader. Actually, they found that this program didn't work by their own planned main analysis.
As a fun side-note: they talk about the importance of G. Although the misidentify the G, I guess it's a start!
So the n=597 sample size is from their intervention arm which was 597+963=1560. They began with a lot more than this in the original report, but it gets more confusing if we read their sample description!
I don't know either.
Here's the original sample sizes. I don't know what happened to SPELL, not the one they are talking about here, which is Fart på sproget (speed up the language, yes, haha, fart jokes).
nb-ecec.org/en/articles/po…
Oh, here's the report, it's in English.
rockwoolfonden.dk/app/uploads/20…
And a prior report about these projects, in Danish:
childresearch.au.dk/publikationer/…
Share this Scrolly Tale with your friends.
A Scrolly Tale is a new way to read Twitter threads with a more visually immersive experience.
Discover more beautiful Scrolly Tales like this.