The idea that concepts depend on their reference class isn’t new. A short basketball player is tall and a poor American is rich. One might have thought, however, that a blue dot is a blue dot. Blue can be defined by wavelength so unlike
Luzius MeisserIf you read the actual paper (available at http://sci-hub.tw/http://journals.sagepub.com/doi/pdf/10.1177/0956797618761661), you will find that this AtlanticSee more article provides a very distorted view of what the researches actually found. First, this has nothing to do with the "replication crisis" as the article claims. The marshmallow test has been repeated successfully again and again, as the actual paper also says. What differs are not the test results, but the conclusions. In particular, they show that once taking the background of the children into account, the marshmallow test is much less informative. For example, if a child scored well in other cognitive tests earlier, then doing the marshmallow test does not provide that much additional information. That does not mean that the marshmallow test is nonsense. It just means that once we know that a child does well at other tests, it is less surprising anymore when it also does well at the marshmallow test. Furthermore, the article claims that this is mostly about income. But the researchers included many other control variables besides income, including race, many of which also turned out to be highly significant. So instead of focusing on income, the Atlantic could also have used the same findings to claim that it is not the marshmallow test that determines success, but being white. All the researches showed in the end is that the results of the marshmallow test correlate with many other things, but the data does not allow us to tell much about how they are connected, i.e. what causes what. This latter part is a question of interpretation.