I think the point is not that one should use larger samples, but rather than the conclusion is wrong since the sample is too small. Iow, there's no proof that the situation really changed in the US on the discussed topic.