Debunking the Marshmallow Test Myths

Video link

Was the original marshmallow test study defuncted? My last video that mentioned it had a few people commenting, with some claiming that the latest research showed that all it measured was whether or not a kid came from a wealthy home. I can't help it that I'm popular. There are evidence? Articles like this one from the Atlantic, which you'd think would be trustworthy. But let's take a look see. They say that in this 2018 study, the researchers restaged the classic marshmallow test. They administered the test, and then tracked down the kids years later to see how they did in school. But as scientists, we like to read the original research. Here's the paper. They used data from the blah blah blah study of yada yada, i.e. no restaging, they used an existing data set, which, unlike the original marshmallow study, which measured 15 minutes, this time and only measured if the kid could wait seven minutes, at which point they call it quits. And more than half the kids hit the seven minute mark, which the researchers themselves said made a lot of their resulting calculations less reliable. And therefore, we consider our study to be a conceptual rather than traditional replication of the original study. And they state later that these issues, render it impossible to provide a definitive answer to whether the between early delayability and later achievement differs by socioeconomic status. Translation? It's not clearly just all about money. But what did the study show? Just like the original study, they found a pretty strong relationship between performance on the marshmallow test and later grades in school. But here the correlation was half as strong as that original study. Then they tried controlling for 30 different variables related to home and background, after which the relationship was still significant, but a lot smaller. Only when they also controlled for a bunch of variables of behavioral and cognitive performance, turned to be no longer statistically significant. Just barely. And overall, the strongest predictor of performance of the marshmallow test was not family background or money, but other measures of cognitive performance of that child. So, of course, family background matters. And helping to improve the children's home environment and economic situation is one of the best levers that we can pull to try and improve their outcomes later in life. And no, the marshmallow test cannot be taken in a vacuum. But saying that it predicts nothing after controlling for all these variables, it's sort of like saying how much you can bench press predicts nothing about how many pull-ups you do. Once you control for a socioeconomic background, grit, and how much you can curl. End result being, the marshmallow test is still an interesting measure of one aspect of a child's performance. And always read the original research. Or follow for me because I'll do it for you and explain.

Additional notes

Replying to @Avisha - 🧬Science Made Simple ⚠️ HOWEVER: Another 2020 study re-analyzed the same dataset as the 2018 one using better statistical methods and found that better performance on the marshmallow test DID seem to predict better academic performance, social skills, and fewer problem behaviors later on in life. Interestingly, they found that the most important contributing factor to this was NOT socioeconomic status OR willpower, but the level of social support that the child had. Which then played into whether or not they were likely to trust adults in general, for instance when they promise to “give an extra marshmallow later.” The great thing about this is that social support IS NOT GENETIC. Want your kid to do better later on in life? Be trustworthy and support them when they’re young. 📚 DOI: 10.1177/0956797619896270 DOI: 10.1177/0956797618761661 #science #Psychology #stem #marshmallowtest

References