Test positivity is a weird metric because of things like this. It's interesting week-over-week, but loses meaning over the long term as testing protocols, other behavior changes, and changes to the disease. This makes it a lousy metric for policy making.
There was a county in Louisiana that went back into a shutdown Thanksgiving week, when nothing at all had changed, except that a local university was on break and had stopped flooding the daily statistics with thousands of negative tests.
Not to mention the perverse incentive: if you feel sick, you want to avoid getting tested and counted as a positive if you don't want your locality or employer to shut down. There is a theory that this accounts for much of Japan's perceived success - antibody studies show just as much prevalence as everywhere else, what happened was that all the mild cases never got tested.
There are also non-perverse incentives. If I feel sick and live alone, a test test result doesn't give me much actionable information, takes time, and puts the people giving the tests at slightly more risk.
