I wouldn't expect a response, as it would likely weaken the argument he is trying to make through manipulation of statistics. Move along.
This is not a manipulation of statistics. The data is there and easily found. You use your Nielsen data and do a one-month run on cume for AM stations in each market and see what percentage of the total market 6+ or 12+ population you get.
You can then get the cume persons and add them and find out what percentage of all radio users they are.
However, I have concerns as to how this was done.
First, cume is not exclusive. A listener may use several AM stations in a month, and unless the data run only counts that person once, you will have an artificially high number. For example, a news follower in New York City may cume both WINS and WCBS (AM) and, maybe, even WABC or WOR. They can only count once as a person, even if each station has that individual as a "cumer".
Second, was the national total weighted by market size to get the national percentage? New York City (Again) is worth about 36 Fargos.
Third, a some AM/FM simulcasts are listed in Nielsen's "Single Line Reporting under just the AM call letters. Their cume is NOT AM cume, as generally those simulcasts have about 70% to 80% going to the FM.
Fourth, many, many AMs have translators. Since a translator is dependent entirely on the station it rebroadcasts, Nielsen shows the audience of an AM/FM translator combo always under the AM call letter only. So many apparent AM listeners are actually listening to an FM that is the child station to an AM parent but which, because it is FM, gets most of the listening. (In fact, many AMs with translators have reduced power, particularly at night, and depend entirely on the translator for their revenue and listening).