Would you go ahead and break out the the impact you're claiming as it pertains to adult children living with their parents? I disagree that it makes it completely broken, the effects shown in the study are dramatic, rather than subtle. You'll need to show an equally dramatic impact from adult children living with their parents having a counter effect.

That study also makes adjustments for household size. Further, average household sizes are smaller today than they were in 1980, by about 10%. That trend has been persistent, with household sizes shrinking considerably since the 1950s/1960s.

The upper middle class expanding from 12.9% of households to 29.4% of households, cannot be explained by your premise.

The change in adult children living with their parents between 1994 and 2016, is a mere 4%. From 27.5% to about 31.5%. And most likely that figure is set to decline a bit from here given the very low unemployment rate.

That 4% change, for example, could explain some of the drop in the poor and near poor, from 24.3% of households to 19.8%; and the lower middle class falling from 23.9% to 17.1% of households. However it very obviously fails to explain the change on those two combined. And it entirely fails to explain the epic increase in the upper middle class.

> Would you go ahead and break out the the impact you're claiming as it pertains to adult children living with their parents?

No, barring a grant to do so, I'm not going to dig into the paper’s source data and rerun the analysis without the methodological errors, of which the treatment of the demographic noted is just the most glaring, should be obvious to even a layman at the time who had been basically attentive to the news, much less anyone seriously trying to do honest analysis. Among the other methodological errors (and this is not intended to be an exhaustive list):

(1) The paper commendably recognized that current money income is a different thing than class, which “knits together multiple factors such as income, wealth, education, prestige, and cultural sophistication”, and then proceeded to use money income alone to measure class mobility on the assertion that it “current income is closely related to all the factors associated with social class and because data on current income are readily available.” While prestige and cultural sophistication are clearly hard to get annual statistics for, education and wealth distribution are not; there is no excuse for not including these of the intent to is assess a feature which includes them.

(2) Related to #1, Rose does not justify his thresholds as meaningful class thresholds,

(3) Also related to the use of money income as a proxy for class, Rose justifies using fixed real income thresholds (which mean that changes in the median inflation-adjusted income with even with no distribution change within the non-excluded population will result in upward “class” mobility, which, even to the extent that money income correlates to class, doesn't seem to correlate with how it usually correlated with that concept) with the only justification being that that's how the federal poverty line works (which is true, but it's worth noting that this is both a controversial thing about the federal poverty line, and dubious in relation to class generally even if it is appropriate to the purpose of the FPL; there's probably a good reason to think that what is right for the FPL corresponds in some ratio to the boundary for the lower class, but that with successively higher classes income mapping to the class is not simply a constant multiple of the poverty line.)

(4) No justification is given for the method of calculating “family of three equivalents” (using the square root of the ratio of the actual household size to 3 as an income divisor to get the “family of 3” equivalent income.) Note that if one views the FPL as having a reasonably constant relationship to class boundaries as Rose suggests in his justification of the use of constant real-income thresholds, this method of size adjustment is tolerably close for families under 3 (though off in different directions for size 1 vs. size 2) but artificially pushes families over three upward in “class” because it understates the additional income needed to stay at the same level with more family members.