The entire reason they did the marshmallow study is because most studies on impulse control cannot avoid confounding factors.

Time value money matters if I am offering you money now vs later. E.g. if you are in debt money later effectively involves the interest earned on that money at likely 25% or worst case 900%. If you aren't in debt the alternative is investing at 7% with risk or 2-5% without risk.

Trust is incredibly important. Money now is money now, money later might be money later if they actually fulfill the promise. And this isn't income agnostic as the risk of this varies wildly based on the impact of the money. A "get back on your feet" amount of money today or a slightly larger amount in a year implies a lot more risk than some spending money on either case.

Additionally while genetic markers have sometimes been effective at predicting even those have trouble with the random nature of gene transference.