Over the last few years I have seen a number of tussles over income distribution measurement in the economics journal literature between people drawing data from different sources and maintaining that data from one source measured over one unit of analysis is more reliable or more closely related to welfare outcomes than measurement over another, e.g. tax records (tax filers), or SSA records (individuals), CPS-ASEC/ACS/SCF/etc. (households), family income, also calculated for the CPS, etc.,and sometimes used as a bridge between household and taxpayer data.
I'm trying to understand how one should answer these questions in principle, what difference it makes empirically, whether some measures are better for some purposes (e.g. for measuring distribution of consumption, income, wealth, etc.), how one might try to convert, compare, or harmonize distributional data from different sources, or even from the same source, when defined over different units of analysis.
The best I have seen on this are papers by Burkhauser, Larrimore, and various co-authors, e.g. "Deconstructing Income and Income Inequality Measuress: A Crosswalk from Market Income to Comprehensive Income" with Phil Armour, and then the whole industry of people arguing with Saez -- not that Saez isn't a worthy opponent for the lot. But I seem to recall a set of careful annual comparisons by something like interagency teams of people from the BEA, BLS, Census, Treasury or IRS SOI, and so forth doing a series of careful comparisons at least of measurement and definitional differences, (including, I think, NIPA/CEX comparisons on consumption). But I can't seem to find them at the moment and I don't know whether they are still being done.
And I have never seen a good welfare-economic comparison of measurements using alternative family/household/tax return units. The closest I've seen to that is the adult-equivalence literature, but that usually assumes a unit and just asks how to adjust for variations within that unit.