PPM System is Terrible

[uncamark]

I notice that WUMB and WERS are still listed, and I know WUMB doesn't buy the book. Does WERS?

WERS and WUMB are Radio Research Consortium members and therefore are considered clients. WERS is the only student station member in Boston and the other two RRC members in Boston, of course, are WBUR and WGBH/WCRB.

 

[reelyreal]

I think some people are missing the point that ratings AREN'T necessary for a radio station. You don't need Arbitron to be successful financially and to know you're pleasing your listeners. There are hundreds of stations in non-rated markets that make money. I know, I've worked/work for them. Ed Christian at Saga Communications is another champion of not buying the book, and they're doing just fine.

The book is a nice luxury and can be a good tool if you have the money, but it's not a life-and-death matter.

If you're a commercial station, encode the PPM, don't buy the book. Selling on numbers is a slippery slope. Then, if you do well enough to show in the books, the media buyers who buy the book see your numbers and you'll get a buy.

 

[Seltzer]

I am going to advise my non-subscribing clients to turn off PPM encoding, since they will no longer be listed in the publicly released numbers.

It's a just & appropriate protest of Arbtron's new policy. If enough stations turn off their PPM encoding signal, then maybe Arbitron may lose its accrediting in that market from the Media Ratings Council or whatever its called. But be careful, once a station signs an agreement with Arbitron, it may have to keep the PPM encoding signal up & running for awhile. Stations should check their contract before turning off, so they know WHEN to legally turn it off.

What a nutty idea. The agencies..the one's who make the buying decisions..see ALL the numbers of ALL stations. Turn off the encoding and you won't show up and will get no buys.

 

[4CX1000A]

What a nutty idea. The agencies..the one's who make the buying decisions..see ALL the numbers of ALL stations. Turn off the encoding and you won't show up and will get no buys.

If any of these stations had any prospect of attracting an agency buy, I'd agree with you. But they do not; these are small stations that have never done business with a major agency in their history.

 

[davideduardo]

How is the PPM encoder connected?

The PPM encoder is inserted in the audio chain prior to the transmitter.

http://www.arbitron.com/portable_people_meters/thesystem_ppm.htm

It has quite a few failsafe and monitoring safeguards.

 

[Seltzer]

But I must say.... the PPM system is far better than the old diary system.

I like the diary system a lot better ( my opinion ).

The meter is better, It actually picks up real radio listening. not what someone thinks they were listening to.

 

[Neanderpaul]

But I must say.... the PPM system is far better than the old diary system.

I like the diary system a lot better ( my opinion ).

The meter is better, It actually picks up real radio listening. not what someone thinks they were listening to.

How many billboards do you pass on the way to & from work each day?

According to PPM, you saw them, digested the message, and purchased the product & service of *every* one you passed.

What's your favorite beer?

WRONG!

According to PPM, it's Random Beer Lite. Because a guy in Malden passed 26 stores on the way to work today with Random Beer Lite signs in the window. PPM "determined" he stopped at every store and purchased some.

When the compression of in-demo AQH Persons is in the hundreds, and 1200 AQH Persons can be the difference between top 5 and mid teens, and 900 meters represent the entire 6+ listening for a city of 3 million, and each meter equals over 3 thousand persons.

Yeah...that's happening.

The total number of usable 6+ panelist data, in some markets is less than one station's AQH person total.

PS...of those 900 meters, 1/3 might be distributed within the same home.

AND...

When you break down 5-year demo data, there are entire demos unrepresented. For (completely theoretical) example, perhaps your station targets Women 18-49. There *may* be absolutely no women 18-24, or 45-49 represented in 3 out of the 4 weeklies that made up your month. That's right. If the wrong meter holder in a sub-divided 5 year demo within the demo you target, isn't in town, *no* woman between the ages of 18-24, & 45-49 listened to radio in Boston that week.

Look at those compression numbers again.

*That's* the system we as an industry have adopted to dictate our success.


Have fun!

 

[N.E. Streetz]

But I must say.... the PPM system is far better than the old diary system.

I like the diary system a lot better ( my opinion ).

The meter is better, It actually picks up real radio listening. not what someone thinks they were listening to.

How many billboards do you pass on the way to & from work each day?

According to PPM, you saw them, digested the message, and purchased the product & service of *every* one you passed.

What's your favorite beer?

WRONG!

According to PPM, it's Random Beer Lite. Because a guy in Malden passed 26 stores on the way to work today with Random Beer Lite signs in the window. PPM "determined" he stopped at every store and purchased some.

When the compression of in-demo AQH Persons is in the hundreds, and 1200 AQH Persons can be the difference between top 5 and mid teens, and 900 meters represent the entire 6+ listening for a city of 3 million, and each meter equals over 3 thousand persons.

Yeah...that's happening.

The total number of usable 6+ panelist data, in some markets is less than one station's AQH person total.

PS...of those 900 meters, 1/3 might be distributed within the same home.

AND...

When you break down 5-year demo data, there are entire demos unrepresented. For (completely theoretical) example, perhaps your station targets Women 18-49. There *may* be absolutely no women 18-24, or 45-49 represented in 3 out of the 4 weeklies that made up your month. That's right. If the wrong meter holder in a sub-divided 5 year demo within the demo you target, isn't in town, *no* woman between the ages of 18-24, & 45-49 listened to radio in Boston that week.

Look at those compression numbers again.

*That's* the system we as an industry have adopted to dictate our success.


Have fun!

I Agree.

 

[WNTIRadio]

Diaries weren't any better. Lots of "voting" went on in that system. Do you remember every 15 minutes what you listened to? People would write down their favorite station or what they actually listened to for 15 minutes, but report it for a lot longer.

I agree that PPM isn't perfect. But as for your billboard, did you see it? Did you remember something about it? (Let's say it was one for a Ford dealer). You don't need to go out and purchase right away for it to be effective. It's about impressions and top of mind, the same reason Coke, Pepsi and McDonald's continue to advertise.

If anyone can come up with a better method, let me (and Arbitron) know. Personally, I think a split system, with diaries and PPM, with the results averaged, would probably give a better picture of what's going on.

 

[Neanderpaul]

as for your billboard, did you see it? Did you remember something about it? (Let's say it was one for a Ford dealer). You don't need to go out and purchase right away for it to be effective. It's about impressions and top of mind,

Unfortunately, assuming the results *are* more accurate than the diary, the data is not real-time. It's 3 weeks old. And knowing how some are prone to knee-jerk, overreacting to a bad monthly, which may be no fault of the station, causes multimillion dollar format flips, and missed annuals.

I think if people are voting with their hearts, it's an indication of the kind of brand loyalty advertisers would desire.

Indifference is death in the ad world. You gotta hook them to the point of motivating consumers to part with their income for your product. The best imaged brand, wins. You make a deliberate effort to purchase your favorite product. I believe those who express a deliberate choice for their favorite station, are doing the same. So, the trick is...to be their deliberate choice.

Problem is...we are often in environments throughout the day, where the choice of what we hear, is out of our control. Consequently, whomever controls the environment, skews the results.

Exposure is *not* listening.

The person wearing a meter, walks in to the SS Plaza. In the Plaza is a national retail outlet, who've cut a deal with a major radio company to *only* play their listed stations in the audio section of their stores. They're shopping for a TV. They walk through the store for 45 minutes. They're *not* listening to said radio company's station. They're just in the store. They didn't hear anything. The meter just picks up a signal that the panelist may not have even known, much less chosen. That station gets "credit" for 3 quarter hours, and an artificial kiss in ATE.

And the panelist wasn't listening at all.

I understand that it's about exposure to advertising. But, when jobs are lost via incidental data, there's a huge problem.

We work in one of the *only* industries where proficiency at your job means nothing. No matter how much effort your staff puts forth, nothing they do can circumvent the scenario above.

I don't believe we can assess our station's performance based upon data that even Arbitron won't stand behind as reliable. Read the disclaimer at the bottom of each page.

What's the number of meters Boston saw in the last 3 months? What percentage of the total population of the market is that? Can you show me *any* other industry that uses the estimates of fractions of single digit percentages to effectively assess their product?

There's *got* to be a better way. I don't know that there's one we as an industry are willing to fund.

Accuracy is expensive.

 

[PirateJohnny]

Since having all stations encoded makes radio, as an industry, look better, there is no reason to not encode. Since it costs nothing to encode for stations inside a PPM MSA, I can't see why anyone wouldn't.

How is the PPM encoder connected?

The TV equivalent is inserted in the program line to the transmitter -- which means a failure will take you off the air. The devices are very reliable, but they aren't 100% reliable.

LOL. Been there, done that.

 

[davideduardo]

How many billboards do you pass on the way to & from work each day?

According to PPM, you saw them, digested the message, and purchased the product & service of *every* one you passed.

Bad example. The PPM measures hearing... an actual volume level adequate for a person to understand. When you drive by a billboard, you likely don't even look at it.

A PPM detection is a real impression. A billboard registers an impression if it is read, not if it is just passed by.

When you break down 5-year demo data, there are entire demos unrepresented. For (completely theoretical) example, perhaps your station targets Women 18-49. There *may* be absolutely no women 18-24, or 45-49 represented in 3 out of the 4 weeklies that made up your month.

But the PPM uses a panel... it is nearly all the same people every week. So if you got nothing in a cell one week and something another week, then it's because the same folks did not listen in that other week...

Arbitron would be glad to increase sample. Stations would not be glad to pay for it, and that is why the sample is the size we have today.
[/quote]

 

[reelyreal]

Since having all stations encoded makes radio, as an industry, look better, there is no reason to not encode. Since it costs nothing to encode for stations inside a PPM MSA, I can't see why anyone wouldn't.

How is the PPM encoder connected?

The TV equivalent is inserted in the program line to the transmitter -- which means a failure will take you off the air. The devices are very reliable, but they aren't 100% reliable.

LOL. Been there, done that.

The PPM encoder is installed in your audio chain, yes. That's why Arbitron sends you two encoders, a primary and a back-up with huge, clearly-marked bypass buttons on the front. Sure, I suppose both of them could fail simultaneously. I also suppose your tower could fall down tonight.

Getting knocked off air by a PPM encoder is pretty much a non-issue.

 

[w9wi]

Who hits that bypass button?

I'm certainly not saying I think these things are a serious reliability issue -- they aren't. What I'm saying is that while the downside of installing an encoder at a non-rated station may not exceed the upside, there *is* a downside.

 

[davideduardo]

Unfortunately, assuming the results *are* more accurate than the diary, the data is not real-time. It's 3 weeks old. And knowing how some are prone to knee-jerk, overreacting to a bad monthly, which may be no fault of the station, causes multimillion dollar format flips, and missed annuals.

We get weeklies just over 2 weeks after the fact; except for rare special news days and sporting events, there is a point of diminishing returns on the timeliness issue.

We get books, also, about 2 weeks to 2 weeks (depending if you are market #1 or market #40) after the fact. The book (there are no "monthlies" in the PPM) contains data that is at most 6 to 7 weeks old, and which averages 4 weeks old; in the diary, we got the data 5 to 7 weeks after the end, and the data averaged 10 to 12 weeks old.

And since the panel only turns about 8% a month, the data from even a few weeklies is more reliable for format and program decisions than several diary books combined.

I think if people are voting with their hearts, it's an indication of the kind of brand loyalty advertisers would desire.

Having reviewed millions of diaries over a period of 4 decades, I saw scant evidence of "voting" but lots of evidence of bad memory and good ad campaigns. For example, in Chicago I used to see entries of "Howard Miller - WIND" more than a decade after WIND had gone to a Spanish format .

That's not loyalty... it is the frailty of human memory.

Exposure is *not* listening.

Advertisers who buy with ratings want a measure of exposure, and that is one of the prime reasons the PPM was developed. Or do you think that radio stations really wanted to pay 60% more for ratings than before?

The person wearing a meter, walks in to the SS Plaza. In the Plaza is a national retail outlet, who've cut a deal with a major radio company to *only* play their listed stations in the audio section of their stores. They're shopping for a TV. They walk through the store for 45 minutes. They're *not* listening to said radio company's station. They're just in the store. They didn't hear anything. The meter just picks up a signal that the panelist may not have even known, much less chosen. That station gets "credit" for 3 quarter hours, and an artificial kiss in ATE.

That is just wrong. First, the locations where audio from a radio station is detectable by a PPM for five minutes out of every fifteen and covering a total of 45 minutes are few. Second, for a station or group to get any advantage out of in-store exposure, they would have to be in hundreds of stores in each city... as the chances of a metered person spending much time in such a store is minimal and totally unable to affect ratings.

But the biggest error of logic here is thinking that such short, one-time exposure would improve TSL (or ATE). Actually, short exposures will lower the average listening time of the station. If a station averages 5 hours TSL a week, a person who gives 3 quarter hours one time will reduce the average slightly.

In fact, half the cume of a station listens an hour a week or less, and is useless; half of that short-exposure cume does not even remember the station, so the exposure is of no benefit to the station in the long run. When a reach and frequency run is done, those short-span listeners can't be effectively reached, anyway.

I don't believe we can assess our station's performance based upon data that even Arbitron won't stand behind as reliable. Read the disclaimer at the bottom of each page.

I have not seen a "Page" of Arbitron data for a decade, i think. But in any case, the printed books going back to 1965 said "audience estimates" and used to contain a nomograph showing how much error there might be. Statistics is the only science where "error" is not a dirty word. We select an acceptable error based on cost and benefit analysis for the sale of advertising. It's not a rocket launch.

Can you show me *any* other industry that uses the estimates of fractions of single digit percentages to effectively assess their product?

Yeah, start with P&G going back to when they essentially invented consumer research 80 years ago. Small samples are typical in consumer research... particularly since processes like replication can show that larger samples do not improve the results significantly.

 

[jefflaurence]

Can here be abuses exercised with the PPM?

Why not?...here's a rhetorical example. I do voice imaging for many large market stations.

I have a client that ended up with something like a whopping 60% share of the audience during a recent summer's ratings period. The most listeners to one radio station in that market from what I understood. I thought it MUST be the great imaging 8) ..Technically they have a decent signal in the metro...although not great..just okay NOBODY could figure out how they had such an awesome summer book! Waaay out of whack.

Here's what they found out happened. Evidently the station was encoding their PPM data like all the other stations in the market..however..they were also the source of the "soundtrack" for the 3 hour Fourth of July Fireworks Extravaganza simulcasted over a local TV station that was also providing regional affiliates with fireworks programming on digital sub channels all over The state..and also on a local cable news channel. Guess what? Their PPM data was also simulcasted as well, so that anyone watching or in a room with a TV on that had that program on..automatically gave my client radio station credit for many many solid quarter hours of listening. That translated into a huge artificial bump.. And put the flashlight on ways many stations would be able to slightly "tweak" the numbers of their underperforming stations in a cluster with a couple "reversed" patch cords in a rack for a few hours a week...casting all kinds of suspicions on the true accuracy of these PPM devices...interesting crap.

 

[PirateJohnny]

...I do voice imaging for many large market stations....

You're so modest, Jeff. I work at a TV station that used you before new owners used their own team. I'd like a tour of our studio some time.

Seems like I heard stories about the new PPMs attributing new demos to stations because, for example, middle-aged men were shopping in a major retail outlet that was piping in a radio station aimed at younger females. The station was "surprised" to find out that so many 49-54 year old men listened to a station that aired Delilah. Or was that story just an urban legend.

 

[davideduardo]

Here's what they found out happened. Evidently the station was encoding their PPM data like all the other stations in the market..

I find it unusual that a station would feed any kind of outside distribution at the just-before-the-transmitter side of an audio chain, which is where the PPM encoder goes.

Usual feeds are from a studio output or some kind of distribution amplifier... generally before the separate processing that a web, HD and air feed would have. Since stations separately encode internet streams and on-air audio, the general location for feeding outside carriers would be right before that split.

Of course, the feed may not have been a feed at all but just a receiver picking up the station... and that would explain the encoding going across to other distributors of the event audio.

In the early days of PPM, Arbitron used to show the math about how putting the PPM encoded audio of a station on the stadium sound prior to an NFL game would have essentially no effect on the numbers. Of course, their intent was to discourage things that might bias the ratings... so who really knows. ???

 

[Neanderpaul]

Bad example. The PPM measures hearing... an actual volume level adequate for a person to understand. When you drive by a billboard, you likely don't even look at it.

A PPM detection is a real impression. A billboard registers an impression if it is read, not if it is just passed by.

Absolutely untrue. If you're in any environment for over 5 minutes, and within range of the signal, PPM records it as having been "heard."

Rush hour was the element missing from my equation. I stand by the analogy.

But the PPM uses a panel... it is nearly all the same people every week. So if you got nothing in a cell one week and something another week, then it's because the same folks did not listen in that other week...

And this is a good thing? You're in a city of millions, and relying on one, or two F meter holders between 25-29 to represent *all* listening in that demo for the market? Did you even re-read that when you typed it?

That's not exactly a sound defense.

(there are no "monthlies" in the PPM)

That's funny. I have 13 monthly "books" delivered every year. Somehow there's an unaccounted for period known as "Holiday" which allows for certain stations to artificially inflate their ratings based upon an ever-increasing overflow period into January, and ramping up period in December, based upon Xmas music airplay and exposure in retail outlets to create a holiday atmosphere.

And since the panel only turns about 8% a month, the data from even a few weeklies is more reliable for format and program decisions than several diary books combined.

No. It is not. Because the panel is too small to be representative of the listening habit of millions. Plain & simple. We should not be relying on the *same* group of people to represent that size audience. One meter skews the weekly sample enough to knock someone out of the top ten. We're talking about (in some markets) AQH compression of less than 1000 people. When a meter represents over 3000 people, and it's normally a listener, and out the market for several weeks per year, it can ruin a station's performance.

Having reviewed millions of diaries over a period of 4 decades, I saw scant evidence of "voting" but lots of evidence of bad memory and good ad campaigns.

Having reviewed PPM, I see total demographics that represent less than 30% of the population of a market, represent 60% of the top listening P 25-54. I see inaccuracies blown off with excuses when addressed. I see people losing jobs for a system that can be every bit as manipulated and skewed as it's predecessor. I see us paying more, for the same. I see certain companies given concierge service by Arbitron, based upon their having executives sitting on the advisory board.

Two wrongs don't make it right.

And the one I currently see, which is supposed to be better representative of "actual" listening, isn't. It's *exposure.* And I see lots of stations raising concerns, with very little response. I see Arbitron posting nice profits off of radio, while delivering data that it won't stand behind as true.

I'm not even going to get in to the mother who has her daughter carry two meters. Or, the meter on the ceiling fan/dog collar. Or, the disparity in financial remuneration given to certain demographics that cause the skewing of listening.

Advertisers who buy with ratings want a measure of exposure, and that is one of the prime reasons the PPM was developed. Or do you think that radio stations really wanted to pay 60% more for ratings than before?

Were they given a choice? They were told this was more accurate. It is not. It is nothing more than a consistent control group. If you don't get one of your station's listeners in that group, you're screwed for 18 month, or until that listener makes up part of the 8% turnover.

What is 8% of less than 2/3 of 1% of total population of the market?

Good luck with that.

the locations where audio from a radio station is detectable by a PPM for five minutes out of every fifteen and covering a total of 45 minutes are few. Second, for a station or group to get any advantage out of in-store exposure, they would have to be in hundreds of stores in each city... as the chances of a metered person spending much time in such a store is minimal and totally unable to affect ratings.

The chances that 900 people represent the actual listening patterns of a city of 3 million are equally as ludicrous. I can at least prove that my scenario will do what I claim it will do.

But the biggest error of logic here is thinking that such short, one-time exposure would improve TSL (or ATE). Actually, short exposures will lower the average listening time of the station. If a station averages 5 hours TSL a week, a person who gives 3 quarter hours one time will reduce the average slightly.

In fact, half the cume of a station listens an hour a week or less, and is useless; half of that short-exposure cume does not even remember the station, so the exposure is of no benefit to the station in the long run. When a reach and frequency run is done, those short-span listeners can't be effectively reached, anyway.

Wait one minute! Arbitron actually preaches that the *average* listener listens between 8-12 minutes per exposure. That means the *average* listener listens *more* that the short-span listener. We're told there's "nothing we can do to increase that ATE beyond 12 minutes. It is what it is." We're being conditioned to strive for *more exposures.*

More total exposures = better ratings is the formula.

Whether they hear it, or not is irrelevant. As long as the station is *on* where they are.

I have not seen a "Page" of Arbitron data for a decade, i think. But in any case, the printed books going back to 1965 said "audience estimates" and used to contain a nomograph showing how much error there might be. Statistics is the only science where "error" is not a dirty word. We select an acceptable error based on cost and benefit analysis for the sale of advertising. It's not a rocket launch.

"PPM ratings are based on audience estimates and are the opinion of Arbitron and should not be relied on for precise accuracy or precise representativeness of a demographic or radio market." - Arbitron

There's the current disclaimer.

What it means is "We're guessing." And people's careers are being gambled away upon their guesses.

If they don't actually represent the audience we're attempting to reach, why are we using the service?

 

[davideduardo]

Absolutely untrue. If you're in any environment for over 5 minutes, and within range of the signal, PPM records it as having been "heard."

If you are driving a car and focusing on the road, you do not see all the billboards... you may not see any of them in peak transit times. There are no impressions if you do not actually see the board, whether you read the board or not.

The PPM registers true impressions. Whether we are listening is immaterial since what advertisers want are impressions and the PPM gives that.

And this is a good thing? You're in a city of millions, and relying on one, or two F meter holders between 25-29 to represent *all* listening in that demo for the market? Did you even re-read that when you typed it?

In a diary or a PPM survey, looking at a very tight demo is dangerous. If you look at a tight demo (25-29 is a custom demo, not a standard demo, anyway) in a single daypart you are getting a tiny sample of the universe, anyway... diary or PPM.

That's funny. I have 13 monthly "books" delivered every year.

Yeah, but they are books, not "monthlies."

The PPM has 13 28-day, 4 week survey periods, named after the months and the "holiday" period (which was unmeasured in the diary as it was the "hiatus" between Fall and Winter).

"Monthlies" are what we got with the diary method via a breakout of one unweighted 28-day period of the three month sample.

Somehow there's an unaccounted for period known as "Holiday" which allows for certain stations to artificially inflate their ratings based upon an ever-increasing overflow period into January, and ramping up period in December, based upon Xmas music airplay and exposure in retail outlets to create a holiday atmosphere.

Actually, in most markets it's just one station or at most two that get the Christmas music "kiss" and the listening is real. Why wouldn't advertisers, such as many retailers who derive more than half their profits from Holiday sales, want to know where the best ad environment for that season is to be found?

And since the panel only turns about 8% a month, the data from even a few weeklies is more reliable for format and program decisions than several diary books combined.
[
No. It is not. Because the panel is too small to be representative of the listening habit of millions. Plain & simple. We should not be relying on the *same* group of people to represent that size audience.

A panel is uniquely able to mirror, hour by hour, day by day, week by week, the precise composition of the universe under study. If 12% of a market is Hispanic, then 12% of the panel is Hispanic.

In the diary, the weekly sample is less than 1/5 of the weekly sample in the PPM, and totally unweighted and totally wobbly. So the diary lets you look at, somewhat reliably, 12 weeks. The PPM is equally reliable down to the one-day quarter hour level, allowing even morning show bits or sporting events or celebrity interviews to be viewed with the same accuracy as an entire book.

One meter skews the weekly sample enough to knock someone out of the top ten.

Maybe in a tight discreet demo... but in the broad ones like 18-49, 18-34 and 25-54, we are talking about large percentages of all meters and one or two will not make the difference you speak of for major "top 10" stations. And, as often or not... or maybe more often than not, the changes you see are real...

We're talking about (in some markets) AQH compression of less than 1000 people.

I have never heard the term "compression" used for "sample size" but I guess that is what you mean. The sample, as I have said, is what radio is willing to pay for. Arbitron would gladly add sample, but radio would not gladly pay for it. So what we have is what we, as an industry, are willing to pay for.

When a meter represents over 3000 people, and it's normally a listener, and out the market for several weeks per year, it can ruin a station's performance.

Ratings are supposed to mirror the market. If a percentage of the market is out of market for vacation or work each year, then the survey should reflect this.

I see certain companies given concierge service by Arbitron, based upon their having executives sitting on the advisory board.

I've never seen this, and the Arbitron Advisory Council has existed for decades. The council members are representatives, each, of their class... by format and market and measurement type.

I see Arbitron posting nice profits off of radio, while delivering data that it won't stand behind as true.

All surveys have disclaimers. No survey based on a statistical sample, can guarantee 100% accuracy. Even the U.S. Census, which is a census and not a survey, has a margin of error. So a disclaimer, in today's "I'll sue you" society, is essential.

I'm not even going to get in to the mother who has her daughter carry two meters. Or, the meter on the ceiling fan/dog collar.

Or granny who fills in everyone's diary and pockets the dollar bills. Every survey of any kind has some extraneous data in it. Arbitron, by the way, looks at "matching" PPM performance including motion time and matching listening habits and they call the users to determine if there is something odd... the PPM is so costly precisely due to the amount of personal attention each metered family or household requires week after week.

Or, the disparity in financial remuneration given to certain demographics that cause the skewing of listening.

Incentives are used to insure parity representation in the survey of groups that do not cooperate at rates as high as others. The incentives are for participation, not for using radio more. So if Spanish dominant Hispanics are, let's say, 14% of a market but normal techniques and incentives only gets 7% of the panel, but paying more gets the full 14%, then DST achieves the purpose of making the survey proportional, which is the goal of a survey, of Arbitron and of MRC. I don't get why you would object to techniques that improve proportionality... I don't get it at all.

The chances that 900 people represent the actual listening patterns of a city of 3 million are equally as ludicrous.

900 meters in-tab is about what you get in Austin. And a replication study would likely show that going to 1800 or 2700 or whatever would not change the overall performance of any but the very bottom tier of stations.

Wait one minute! Arbitron actually preaches that the *average* listener listens between 8-12 minutes per exposure. That means the *average* listener listens *more* that the short-span listener. We're told there's "nothing we can do to increase that ATE beyond 12 minutes. It is what it is." We're being conditioned to strive for *more exposures.*

Each exposure or incident is finite because people do other stuff. They may listen to a certain station all day at work; in the diary they would write "9 AM to 5 PM" or whatever. But in the PPM, every coffee break, phone call, trip to the shop floor, lunch break, visit to the next cubicle, potty break and whatever else occurs in the office, shop, loading dock or delivery truck cuts the listening into very thin slices.

What you are being told by the studies that have been done is to make sure people want to come back to you after lunch, after going on a sales call, after going for coffee. That is what an "incident" means.

If they don't actually represent the audience we're attempting to reach, why are we using the service?

If you want to get rid of the disclaimer, do a daily census of radio listening. Measure every single person, every moment. Get back to me on the cost.