> But when they controlled for household income, they found most of the correlation disappeared. This does not really prove much, since attitudes to long-term

> But when they controlled for household income, they found most of the correlation disappeared. This does not really prove much, since attitudes to long-term gratification are probably shared within households due to the effect of idiosyncratic cultural factors, which might affect both income and academic achievement. One would need to look for a "natural experiment" where divergence in income was totally exogenous and not due to any shared factor in order to conclusively resolve the issue.