The Monthly Secret Behind PPM Ratings

When you look at a PPM monthly, you probably assume that you’re looking at one cohesive sample. You’re right…up to a point. In this column, I’ll attempt to take some of the mystery out of how PPM samples are used to create the Nielsen estimates.

First, there’s average quarter hour, the bedrock estimate, and the one that drives your bonus plan, if you have one. In the diary service, the samples differ each week. As we’ve all known since the beginning of our careers (because the diary dates back to the mid-’60s, which is likely before any of us started in this crazy business), the diary keeper tracks their listening for seven days, meaning there is a unique sample for each of the twelve weeks of the survey period (or 24 weeks if you’re in a condensed or “two book average” market). That’s fairly straightforward.

PPM is different. Panelists can be “intab”, in other words, supply good data and are in the tabulation (hence “intab”) for one day and not the next. Remember that the panelist’s meter has to have eight hours of “carry time” or motion in a PPM day to qualify (five hours for 6-17s). Being intab or not for any given day has nothing to do with media usage. 

If the panelist’s meter registers the requisite eight hours of motion, all of their listening counts.  If not, well, none of it counts even if your morning show was part of their listening for that day. In case you’re wondering, the typical intab rate is in the 80-85 percent range. You can find your market’s monthly average in the e-book.

The number of panelists in the sample varies each day and you can see this in the PPM Analysis Tool. Weekdays have more intab than weekends. Holidays are a problem and Nielsen has typically offered bonuses to panelists if they carry their meters during major holidays. I once did a “Welcome to PPM” panel in a new market and remarked about parents making their kids put on their meters before they could open their Christmas gifts. 

Also, remember that new households are being added and dropped from the panel on a continuing basis. Some new panelists may have been in the sample for just a few days when the month ended and others were there for a week or two at the beginning and then left.

The best way to think about PPM is that a monthly is an average of 28 one-day surveys. The individual days are averaged together to create the monthly, but the number of panelists differs for each day of the measurement month. 

Then there is the cume question. Let’s see a show of hands: how many of you knew that when you run a weekly cume, the sample base is smaller than for the AQH? 

For example, if you run Monday-Friday 6 AM-7 PM, the sample base for the weekly cume is smaller.

In order to be in the weekly cume estimates, a panelist must have been intab for at least six days of the week (Thursday-Wednesday). However, even if the panelist was intab for just one day that week, their listening was used for that one day and is included in the AQH average daily cume, but not the weekly cume. Here’s a little cheat sheet for you:

I knew you’d ask why not put all the intab into the weekly cume and here’s the main reason: The more days you participate in the survey during a week, the more likely we’ll see a full picture of your listening, especially the different stations you may hear for less time than you spend with your P1 station. 

This likely increases the cume for all stations and it’s something worth knowing and understanding when you look at the data. The Analysis Tool will tell you the sample sizes in the upper right-hand corner of the screen when you do a run.

Next obvious question: Does it matter? Well, maybe. First of all, the policy makes sense. A panelist who may have been in tab for only a couple of days in the week will not give as good a representation of their weekly cume listening as someone who was in more days. 

Further, that policy syncs up with the diary where six “good” days are generally required (there is a way for a five “good” day diary to be intab as well, but it’s a more unique situation). And if there is something that ratings services like, it’s consistency.

The one place where it may have an effect is when you really “slice and dice” the data. It’s possible to have a run where you can see AQH estimates, but no cume data, because you have at least 30 panelists for AQH, but not 30 for cume. Won’t happen often but it can. Otherwise, the policy has always had my seal of approval.

Let’s meet again next week.

Dr. Ed Cohen

One of the radio industry’s most respected researchers, Dr. Ed Cohen writes a weekly business column, heavily focused on ratings research for Barrett Media. His career experiences include serving as VP of Ratings and Research at Cumulus Media, occupying the role of VP of Measurement Innovation at Nielsen Audio, and its predecessor Arbitron. While with Arbitron, Cohen spent five years as the company’s President of Research Policy and Communication, and eight years as VP of Domestic Radio Research.

Dr. Ed has also held the title of Vice President of Research for iHeartMedia/Clear Channel, and held research positions for the National Association of Broadcasters and Birch/Scarborough Research. He is adjunct faculty in the Department of Broadcast Communications at Western Kentucky University and welcomes your thoughts. Reach him by email at doctoredresearch@gmail.com.