Dr. Kentaro Fujita

And this matters.

And so when the researchers, they had like 30 or 40 other covariate variables that they were controlling for.

When they control for all these other variables,

children's delay of gratification was no longer predicting these outcomes it was supposed to.

And so this paper got a lot of attention for basically saying, look, the marshmallow tests are bunk.

Now, this has been controversial because the question is, was that statistical adjustment appropriate?

And are we interpreting that statistical adjustment correctly?

There have been other experimenters, other researchers who have come along.

One of them is named Yuko Munakata and her team.

They took the same data set.

And they reanalyzed it with a different set of assumptions, a lot more conservative.

So rather than throwing in 30 covariates, they put in theory-driven covariates, ones that made sense from what we know already about research as opposed to like throwing in the kitchen sink.

And when they did that, they still found that delay of gratification predicted reports of problematic behavior, which suggests a very clean replication of the original marshmallow test.

Some people have suggested that failure to replicate the original marshmallow test.

It got a lot of attention, but it may not have been the final answer because these experimenters, again, came along, looked at exactly the same data set, and came to the opposite conclusion.

So there's still a bit of a debate out there.

But I think the main point to take away here, again, is that the way that you set up the marshmallow test is really important.

You have to have trust.

And the argument about socioeconomic status is that

Kids who grow up in high SES environments, they're very stable, they're very predictable.