You can’t split out sub populations when you are looking at herd immunity type calculations. If the populations can mix, then the virus is flowing between group A and group B. Even if group B has a lower transmission rate, the data are so mixed you may not see it in case rates. ( Group B has case rates that include infections from group. A, and vice versa.)Do we have a "real-world" comparison of mask mandates vs. vaccinations? I'd say hospitalizations and deaths are the two results that will be most reliable as cases will be undercounted in the non-vaccinated population. Have they taken all the states that had both mask mandates and participated in the trial vaccine program and compared the % of hospitalizations and deaths (by age) in the non-vaccinated population and the population that was vaccinated (vaccinations after the first 2 weeks)? That is a reasonable comparison between the vaccine and mask policy.
You’d have to look at different countries with different mask/distance/vaccine rates. For example, you could look at England (Feb) versus New York (Mar). Cold weather and roughly 25% vaccinated, but one had serious NPI and the other did not. So far, that one indicates that strong NPI work better than opening indoor dining.