The argument:
In large part because natural immunity from prior infection is far more common than can be measured by testing. Testing has been capturing only from 10% to 25% of infections, depending on when during the pandemic someone got the virus. Applying a time-weighted case capture average of 1 in 6.5 to the cumulative 28 million confirmed cases would mean about 55% of Americans have natural immunity.That's far higher than what the CDC measures. They have actual data on whether people show antibodies.[1] Their main data source is from random blood samples processed by commercial labs where the samples were taken for some other purpose. The trouble is, the blood lab data runs way behind; the latest data is about 60 days old. But that's what to watch. It's annoying that so little current data is available on this. Anyway, as of late December, those numbers are from 3% (Hawaii) to 23% (North Dakota).
Blood samples are also taken from blood donations. If you donate blood, you get an COVID-19 antibody test done on it, and the Red Cross will tell you the results. (Yes, there's an app for that.) Those provide a crosscheck. People who have blood drawn for medical purposes tend to be sicker than average, and blood donors tend to be less sick than average. So one would expect the value for a whole population to be somewhere above the blood donor rate and below the medical sample rate.
Kansas is seeing 22% of blood donors showing antibodies, as of mid-January. Note that this includes people who have been vaccinated, which, for Kansas, is about 8%. That compares to 12% showing antibodies in lab samples as of mid-December, which was pre-vaccine. I haven't been able to find data from other states. Anyone finding that?
There are a number of approved blood-based antibody tests, and they have sensitivity levels from 89% to 99%. So test error isn't a big contributor to error.
So, while the measured values are moving up, nothing as high as 55% is showing up in the actual data yet. More current data would be a big help. Perhaps the data collection and publication will be improved under the new administration. Anyway, looking at the actuals is more useful than trying to project this from a measurement of something else.
[1] https://covid.cdc.gov/covid-data-tracker/?CDC_AA_refVal=http...
[2] https://www.ctnewsonline.com/news/article_a97ef828-6122-11eb...