So I'm a little skeptical on that front because military aircraft are still not that often broadcasting ADSB even during routine flights, at least in my (also close to the border) region. In theory the Air Force was supposed to have completed ADSB installation on their fleet last year but they blew the deadline pretty bad on even installing transponders, and of course they still reserve the right to disable them during military operations.

Maybe with the data we can figure out what portion of military flights are included?

For the helicopter training flights that I notice most often, it's still rare to see one that broadcasts ADSB, probably <10%. C-130s usually don't either here but it's more often, maybe more like 25%. Perhaps for other categories of aircraft they've installed more transponders. But in the city where I live, even passive mode-C MLAT is probably around 50% success on tracking military flights for ADSB Exchange. FlightAware might have better coverage for mode-C. mode-C can't contribute to this GPS reliability data anyway but it illustrates that even C-130 pattern practice is sometimes "stealth" from a radio perspective due to the low installation rates for ADSB and difficulty of good mode-C coverage.

The paper linked elsewhere (https://web.stanford.edu/group/scpnt/gpslab/pubs/papers/Liu_...) mentioned issues with military training flights resulting in spurious low-NIC cases but unfortunately doesn't quantify it. With the way the AF rollout has gone it probably depends on the specific installation, command, and aircraft type.

In the border region specifically we would tend to expect the majority of non-military flights to be civilian CBP aircraft that aren't performing unusual maneuvers. CBP has a somewhat complicated and limited authority to disable ADS-B that I don't know the contours of, I'm not sure how often they do so on their larger (non-sUAS) aircraft. Involvement of the Air National Guard in the Texas area might complicate the analysis though.