There are a lot of factors that go into determining the sample size for any study or research, cost, time, how hard it is to collect the data and how much data is needed to offer sufficient statistical accuracy. Obviously the smaller the sample size the higher the risk of statistical errors and the lower the confidence in the results. Most researchers will tell you a good sample size is typically 10% of the population up to 1,000 participants. 1,000 participants is generally considered a sufficient sample size even for very large populations because it provides a reasonably accurate representation of the population with a small margin of error. So, what are the sample sizes of some of the research and studies radio bases many of their programming decisions on? Let’s walk through a few.

Traditional radio music testing uses some fairly small sample sizes to make the most important decisions music radio stations make, which songs to add, move or drop. Call out research typically gets 300 people or so for each region of the country unless groups pay to have it done specifically in their market. Online music testing does a little better if there’s an incentive for doing so, garnering more like 500 to 1,000 participants. So, online testing can hit that magic 1k number. Focus groups are typically too small to base music decisions on but they are great for getting specific branding, content and other overall feedback on our stations. However, they’re also expensive to conduct. In my opinion, market-level streaming and sales data is the best music research any programmer can look at because the sample base is literally everyone within our markets. That’s because we’re all listening to music on at least one of our devices over the course of any given week, certainly so when we look back longer than that. Our goal in music programming has always been to identify what the audience likes so we can give them more of that and market-level streaming and sales data has now removed the guesswork. Part of the reason I steer programmers away from putting too much stock in watching what other stations play is because we don’t know what research they’re looking at, how accurate it is or whether they’re properly weighing all of it. For instance, most radio programmers put too much weight on listening to songs and getting a gut feel for whether they fit and too little weight on the data for that song. That’s why I’ve yet to look at any station that’s not missing on a couple of their heavy currents, even in the larger markets.

Audience measurement is of course the big stat we all look at in radio and, although I’d be the first to admit the process isn’t perfect, the sample sizes aren’t that bad really. Nielsen has almost 60,000 active panelists carrying around PPMs spread across the top 49 radio markets where they’re in use. Those markets have a total population of 159,726,000 people. While 60,000 people determining the listening habits of 159 million plus may sound like a small sample base. That is roughly 1 PPM for every 2,662 people and most researchers would say that’s enough to get a fairly accurate measurement. While it’s well under the 10% of the population number (That would require 15.9 million PPMs), it’s still over that magic 1k participants number most researchers look for across those 49 markets. Nielsen diary markets, which by the way are supposed to switch to mobile diaries instead of paper in 2025, have a slightly higher sample base but of course are less accurate because they’re based on recall. But, overall Nielsen does have large enough sample sizes to pretty accurately measure radio audiences. However, with audio fingerprint technology that’s available now and the fact that we’re all carrying around a small but powerful computer in our pockets throughout the day in the form of a smartphone, I can’t imagine we’re far out from a way to measure audiences even more accurately by using a much larger sample base.

What do you think? What sample size are you looking for when you do research in-house? Comment below or email me at Andy@RadioStationConsultant.com.

Pic designed by FotoArtist_1 for Envato Elements.