One thing that annoys me about much political analysis is the use of the demographic characteristics of the
total population of a state or congressional district. Surely what is relevant for electoral analysis is the demography of
eligible voters. It therefore makes much more sense to use the CVAP (citizen voting age population). (Admittedly, even that is imperfect: for example it does not take into account the disfranchisement of felons, but it is still a much better guide to eligible voters than using total-population figures could ever be.)
If it were really that difficult to find the CVAP for congressional districts, there might be some excuse, but in fact such figures are readily available for the 115th Congress:
https://docs.google.com/spreadsheets/d/1RhiFHMsETYQuWjvHfJgEQGvYhUjumQVtejP8KAcqoCE/edit#gid=1241347791 See
https://www.dailykos.com/stories/2018/5/24/1766053/-A-district-s-total-population-and-eligible-voter-demographics-can-vary-wildly-Here-s-how-to-tell?_=2018-05-24T15:09:11.578-07:00 for an explanation.
An example of why this matters: Some people are trying to explain (or explain away?) Ocasio-Cortez's victory by saying "well, she's a Latina and the district is half Latino/Latina." But actually while the total population of NY-14 is 49.1% Hispanic, only 39.9 percent of the district's CVAP is. (And she won some non-Hispanic white areas of the district anyway...)