Part of the problem here when discussing specifics and trends like this is that we use a term like "middle class" to mean two things; either the simple middle q

Part of the problem here when discussing specifics and trends like this is that we use a term like "middle class" to mean two things; either the simple middle quantile (1/3 or 1/5 or 3/5 depending on who you talk to) of income or the more historical definition of less-than-upper-class (those people who do not need to work; who have sufficient capital that they can live off of passive income from it) and more-than-working-class (those who have to work; who do not have sufficient income to accrue savings). And even the more-than-working-class definition clashes with other trends; some pundits seem to define "working class" as nothing more than "does not have a college degree", which means that some trends, rather than being interesting demographically, instead just become a restatement of an underlying trend (like you said, the increased prevalence of college degrees). And the final problem of all attempts to analyze this is that we don't really have comprehensive data that supports a single way of measuring these aspects historically, because the definitions and criteria keep changing. Even when there is insightful research, it really is hard for it to be applicable outside its direct area because the definitions and proxies for historical values are not guaranteed to be uncorrelated with new phenomena that are attempting to be measured.