There was some challenge to this experiment. Once family and parental education were accounted for, the difference largely disappeared after age 15.

I read elsewhere theorizing, that kids from more secure backgrounds will delay more likely as the delay has been rewarded in the past. Whereas, if every dinner is fighting over scraps, they will probably take the initial marshmallow

https://www.theguardian.com/education/2018/jun/01/famed-impu...