However, that "research" is sometimes rather questionable. There is a problem with people not wanting to be surveyed these days.
No, there isn't a "problem" with people not wanting to be surveyed. It just costs money to get people to take a lengthy or ongoing survey.
In any case, the results are valid for radio stations and that is because the same characteristics that are found in a radio station or audio streaming research candidate are found in those who would participate in Nielsen radio and audio surveys.
Most research of any value today includes significant compensation in exchange for participation. Families that participate in the Nielsen PPM surveys earn incentive credits which can give them very nice items and compensation. Diarykeepers get some money and are made to feel "important" by contributing to the success of the stations they listen to. Station research pays nice money (anywhere from $50 to $150 for participation in cash or nice gift cards) for music tests and perceptual research.
In both station "private" research and ratings research, skilled and competent recruiters seek out respondents who meet the specifications of project design.
Going back to when we started doing call-out in the mid-70's and AMTs (Auditorium Music Tests) and in the early 80's and focus groups or one-on-ones in the same era, it takes a lot of work to design the sample specifications and to recruit conforming candidates for participation. In addition to a reward for being a respondent, internal station research may pay as much as $100 to $150 per respondent to the recruiter to find the "right" people for a test.
Between respondent incentives and recruiter costs, a lot of money goes into station research to guarantee that the right people participate. In ratings, while everything is done in-house by Nielsen, the costs are the same or greater to insure a representative sample (diary) or accurate panel (meter).