What methodological individualism means

It is interesting that I continue to see people make the assertion that if one accepts individual explanations as valid, then one is a methodological individualist. That is plain wrong: Methodological individualism, at least as it was presented by Mises, is a doctrine much stronger than that. It says that the only good explanations in the social sciences are explanations at the level of individuals. Explanations that rely on higher-level phenomena, such as national interests, class struggles, collective manias, collective action problems, and so on, all of those are at best shorthands for the true, individual-based explanation. Mises is very, very explicit on this in Human Action.

The equivalent doctrine to methodological individualism in the social sciences is atomic reductionism in the physical sciences. The latter doctrine says that the higher-level sciences only exist because of human mental limitations: it's just too hard for us to calculate what each particle in a hurricane is doing, so we have a science called meteorology, and so on. Rejecting that view of the physical sciences is not equivalent to denying that atoms exist, or asserting that atomic level explanations are universally wrong. Someone who rejects atomic reductionism in the physical sciences may simply contend that chemistry, biology, meteorology and so on will continue to exist no matter how much we know about events at the atomic level, because there are emergent phenomena that happen at these higher levels that simply can't be described solely in terms of atoms. Atoms of iron are not rigid: only iron bars are rigid.

If methodological individualism and methodological holism were the only choices on the table, I suppose I would be a methodological individualist. But there is another option: methodological pluralism. We offer an explanation at whatever level helps us to best understand the phenomenon we are interested. If consumer choice, say, is best understood at the individual level, then we model it at that level. On the other hand, if languages are best understood at the level of speech communities, then we model them at that level.