How does a radio station know how many listeners it has?
Same way as TV stations know: Companies such as Arbitron conduct random polls to ask people what the watch or listen to.
Arbitron (which was the primary ratings service for radio stations) was bought by Nielsen in 2013, and rebranded as Nielsen Audio. As I understand it, they’re pretty much the only company doing radio ratings in the U.S.
Here’s an example of the overall ratings from them for the past few months for Chicago radio stations; they typically break it down further for specific dayparts (morning drive, middays, etc.). The “PPM” listed there is “Portable People Meter,” a device that members of Nielsen’s panel use and carry, which picks up and identifies the audio signals of the radio stations and TV stations that they are watching.
Ok, thanks.
So all this data is skewed towards listeners bored enough to carry around some clunky device with them all day? 
I did a Nielsen radio diary this spring. It suuuuucked because typically I would actually have many radio hours in a week due to baseball (canceled) and driving (nowhere to go) so I pretty much had like 30 minutes over 5 days, when I managed to listen to my shower radio.
There was a spot in the diary to write some thoughts so I did mention that I typically listened to baseball and NPR.
They sent me 3 crisp dollar bills for my trouble.
My grandfather used to tell me that radio stations would play a certain song and send watchers out to see which people were bobbing their heads and/or tapping their toes to that particular beat. 
Years ago, my parents received the radio diaries and I think they asked me to fill it out or to fill it out for them. (Their radio listening was really boring, by the way. They only ever had the radio on in the car, and it was set to WCBS-AM News Radio all the time.)
I got an Arbitron survey a while back. I was supposed to log all the radio stations I listened to (for a week I think). I made extra sure to skew their results by logging WWV, the local WX radio station, etc.
(all of my logging was truthful)
Brian
Are listener numbers for terrestrial radio down now that streaming and satellite radio are things?
I’ve never done the full Nielsen diary, but I’ve done a few online surveys that asked what stations I typically listen to. And since I only ever listen to NPR I always wonder if NPR even cares how many listeners they have, since they don’t sell traditional ads anyway.
I think they care. For one thing, the NPR stations want to know how many people listen versus how many people actually contribute during the pledge drives.
Disclosure: I am, in fact, a professional market researcher. I’ve spent my career doing, and interpreting, studies like this.
In part, though those people are paid to do so (though, as @ZipperJJ notes from filling out a paper diary, they don’t get paid a lot). And, paper diaries are still used, as well.
Regardless, the people who put together those studies work hard at making certain that the demographics of the respondents in the studies are demographically balanced to reflect the population in those markets (age, sex, income, ethnicity, etc.) So, while they are not perfect samples, they are also not purely skewed by being what market researchers are called a “convenience sample” (i.e., just some bored people we rounded up to participate).
The PPM is the most accurate because it doesn’t lie. If you’re an alt rock fan and you spend an hour in the dentist’s office forced to listen to a Country or Soft Rock station, the PPM records your listening anyway. In markets that still use paper diaries, people lie about their listening prolifically. Younger people usually filled out the diary on day 7 just by writing in their favorite station. Older listeners were always more honest and meticulous about it. The PPM is a small device you carry in your shirt pocket. You can’t turn it off and you can’t mess with it, as far as I know. And you don’t have to remember what stations you listened to and when. Then subscribing stations get a digital report with all of this information broken down for sales purposes. The PPM is as accurate a measure of radio listenership as is available at this time.
Not only does NPR care, many NPR stations in larger markets actually subscribe to the ratings so they can quote the numbers to businesses that underwrite programming.
The ratings for my local market show the local NPR station as #17 in the market. They even list four HD channels - even though the only peopl in town who listen to HD channels are people who work in radio.
Gads, that was word salad. That’s what I get for multi-tasking while I write. Here’s a cleaned-up version:
Yes, the sample is, in part, “bored people” (i.e., people willing to actually participate in a study), but the respondents people are, in fact, paid to participate – although, as @ZipperJJ notes from filling out a paper diary, they don’t get paid a lot. And, paper diaries are still used, as well.
Regardless, the people who put together those studies work hard at making certain that the demographics of the respondents in the studies are demographically balanced to reflect the population in those markets (age, sex, income, ethnicity, etc.) So, while they are not perfect samples, they are also not purely skewed by being what market researchers call a “convenience sample” (i.e., just a random group of bored people we rounded up to participate).
IIRC they called me to make sure I’d be willing to do it. Then they sent a letter with a dollar saying my diary would come soon. Then they sent the diary with another dollar. Then they sent one more letter with one more dollar reminding me to send it back.
Apparently they’ve upped their pay slightly. A few weeks ago they sent me a short survey and a dollar bill, and promised five dollars if I filled it out and sent it back. That wasn’t a full diary, just a very short survey asking what stations I usually listen to and what times of day I generally listen to the radio. I sent it back and just today I got a thank you letter and a five dollar bill.
So how much do they pay you to carry this device around? And for how long? What if you’re like me and only listen to the radio in the car and don’t drive much? Will it pick up Pandora on my TV or Alexa?
If the money is right how do I sign up?
Nielsen uses their “Nielsen Families” for both TV and radio ratings, so I’m guessing that it’s those panelists who are using the PPMs. Generally, it’s not possible to contact Nielsen and say, “I want to be a panelist”; they recruit you – and they carefully select panelists so that they can create a representative sample of the entire population.
They do get paid for participating, but not a lot. This article indicates that it’s in the range of $10-15 a month, plus some bonuses for staying in the panel for a while.
In theory, the PPM can pick up and register the audio from any TV or radio broadcast. However, what it’s doing is identifying and decoding a digital signal that stations embed in their audio streams, specifically for use by Nielsen’s devices (that signal is supposed to not be audible to humans, but identifiable to the PPM). I do not know if streaming services or satellite radio stations use those signals.