KVI "GREATEST HITS" FORMAT NOT A HIT WITH LISTENERS

TVradioguru said:

Maybe you can hear KVI at a long distance on your super-special radio with a giant antenna strung across your backyard, but this is hardly atypical with the 99.99% of actual radio listeners.

Click to expand...

Before I left Seattle last May, I drove down to Portland in my car and listened to KVI on the way.

"...hardly atypical..." I believe you mean "...hardly typical..."

Bongwater said:

1,317 people in a market of 3.5 million isn't a "sample", it's a JOKE.

Click to expand...

Don't know much about statistical sampling, do you?

beachguy3b said:

Bongwater said:

1,317 people in a market of 3.5 million isn't a "sample", it's a JOKE.

Click to expand...

Don't know much about statistical sampling, do you?

Click to expand...

Beach, quite honestly neither do I and I've been around this biz for several decades. Just never really got into it. Would you take a moment and give us all a Reader's Digest overview about statistical sampling b/c I'd like to learn something. I really would. As an example I know that double the sample certainly doesn't double the accuracy. But let's say a sample of 750 for 3.5 mil as opposed to 10,000 for 3.5 mil. Something like that. Thanks.

RR

beachguy3b said:

Bongwater said:

1,317 people in a market of 3.5 million isn't a "sample", it's a JOKE.

Click to expand...

Don't know much about statistical sampling, do you?

Click to expand...

I'd like to hear from someone with experience. There is a vast difference between a sample sufficient to track voter preferences in a two-way presidential race and listening patterns involving numerous day parts, demographics, and dozens of signals.

beachguy3b said:

Bongwater said:

1,317 people in a market of 3.5 million isn't a "sample", it's a JOKE.

Click to expand...

Don't know much about statistical sampling, do you?

Click to expand...

No I don't. Nor do 99.99999999998ths of the entire human race.

Please enlighten.

I think my biggest ??? when it comes to the PPM technology is how does that little beeper sized meter determine if you are an active or passive listener?

I mean I am pretty sure I have been in earshot of tons of easy listening music in various offices and buildings I am in but I can assure you I don't know what station they are tuned too because I tune it out, so how does the little meter account for people like me?

Maybe if that dang beeper had a button on it the wearer could push if they were listening to something they really actually heard and enjoyed...

uh huh, and another big question is how them small scant percentage was picked to ppm. and what is the demographic makeup of them? imagine if they picked a good % of nursing home folks. then KIXI would have a good book. what if they just picked urban young youths? then the rap channels would be #1. what if they picked rural pickup drivin 46 year old males? then the real classic country station would, rise and shine. oh thats right, we aint got any to rise and shine. obviously, they try and give a good mix of demographic makeup, dont they? or is this like trying to weigh a jury in a trial to be more favorable to whatever side. obviously, the side that likes republican talk aint too well represented here....

"...hardly atypical..." I believe you mean "...hardly typical..."
[/quote]

Ya, that double-negative sort of threw me. Since atypical is NOT-typical, hardly atypical would mean typical?? HUH??

RR

beachguy3b, still waiting for statisical sampling 101.

RR

beachguy3b said:

Bongwater said:

1,317 people in a market of 3.5 million isn't a "sample", it's a JOKE.

Click to expand...

Don't know much about statistical sampling, do you?

Click to expand...

Yet another UW class I used to catch some "Z"'s ... but I do remember there was a magic "minimum" that would work in any crowd size being surveyed. The point being that if you increased the sample the results were typically the same as with the "magic" minimal-number, so basically suggesting wasted effort.

We can probably google the "magic minimum" -- so question is whether that vs. the 3.5M market reaches the accuracy threshhold or not. I suspect this is bigger issue in smaller markets where overall population is smaller and a much smaller sample set is used (the point BillW has been making).

Randy Roadz said:

beachguy3b, still waiting for statisical sampling 101.

RR

Click to expand...

Don't hold your breath.

Okay, I'll stick my neck out here. My MBA was marketing with a minor in statistics - lots and lots and lots of statistics! It's been a few years - so go easy on me. If the question is how small a sample size can be to accurately represent a population, the answer is 'pretty damn small' - surprising small for that matter.

More important than size (in this case size doesn't matter much) is WHO the sample size is. There is random sampling (whoever is walking by), quote sampling (try and get a representative population) and variations. Who and where the meters are is much more significant than how many there are, in fact, having too many out there can actually DECREASE the accuracy of the reporting.

We'd get into things like standard deviation, chi square values, means, modes and medians that would start to make my head explode and you all go to sleep but we can go there is you want.

Suffice to say the 'problem' if there is one, is likely not HOW MANY, it WHO and WHERE.

@littleboyblue. Oh and the 'magic minimum' you likely remember from your classes is is n =30. Granted that is too small for what we are talking about here but you get the idea.

You know who is very well versed in PPM measurement is David Eduardo, a frequent poster on several of the PPM and ratings discussions. Unlike David, I've attended a few of the Arbitron seminars, but not as many as David. Plus David is very good at explaining the methodology basics than I would be trying to post on my IPhone at an airport somewhere. That and I doubt few here would believe me anyway...

I'll send David a message if he doesn't catch this post before while he's on line. His ears might be burning as we speak.

It is Groundhog Day. I haven't commented here in about a year I think. I've stopped by a few times to read, including today. I've got a few minutes so I think I'll make a few comments.


I won't venture a direct estimate of the statistical significance of a 1300 person sample in a 3+ million market because it probably depends on the sophistication of the targeting and the analysis.

But to take a few steps into the topic, the listenership is comprised of 2 sexes, say 5 age bands and 4 major ethnicities. Between these criteria you have 40 buckets that people fall into. Put a 1300 person sample into 40 demographic buckets and you may have 50-100 people in the buckets with the greatest populations, a few dozen in others and less in the rarer bucket combinations of just these 3 criteria.

Add say 3 income bands, 3 listening frequencies and primary preferences for 6 radio formats and you'd have 2160 buckets. Most of those specific buckets would probably get 0-2 meters. Few buckets would get more than a dozen or a few dozen.

Arbitron surely goes to extensive lengths to get the sample as a whole to represent the the basic demographics of the market to the level of sex, age and and ethnicity but I don't know now how close they get to representing the exact population distribution of the 40 sex- age-ethnicity combinations.

And it would really take great statistical talent and stretching to proportionally represent buckets with 6 criteria including the first 3 and income bands, listening frequencies and preferences for formats. Hard to imagine even attempting to cover anything else.

Can you get a decent idea of what white women between 29-44 years old listen to? Yeah probably. Can you get a decent idea of what white women between 29-44 years old with high income, heavy listening and a preference for contemporary hits listen to? Or an asian male 18-25 with moderate income, moderate listening who likes rap? Or a hispanic male 44-54 with low income, low listening who likes country? I am less sure about getting good estimates of those level of groups from the outside. Of course this is other people's field of expertise but it does seem challenging.

You've got to remember that there's more to ratings sampling than doing the math. There are qualitative and lifestyle issues at work as well.
Well dressed businessman and women aren't going to want to walk around with a little "I'm a geek box" attached to their purses or belts. Ear-bud listeners with mini receivers aren't going to attach some monstrous dongle to intercept their audio and feed it to another box they much attach to their belt. The PPM doesn't perform well inside a purse and what professionally dressed women will clip it to her dress, blouse, suit, etc. The system has myriad cultural biases built-in to its design. Radio was sold a bill of goods by Arbitron which has a virtual monopoly on radio ratings.

some good thinking on your post towerlamp.....and there possibly was the key wound that spelled the ratings death of smooth jazz, KWJZ and the death of republican talk, KVI?

i think you nailed it. what normal business/proffesional person, wants to walk around with a geek box hooked up to them.

do these people recieve any compen$ation for this?

what can be a more accurate rating system in the future? have satellite hookup/ ppm box on every AM/FM radio sold that reports back to big brother and or arbitron what your tuned to?

note: when i first signed up for XM in 03, i asked XM associate if they had the ability to monitor what the recievers are tuned into to for channel rating purposes. they said "no, but good idea, sir"

Steenman said:

Okay, I'll stick my neck out here. My MBA was marketing with a minor in statistics - lots and lots and lots of statistics! It's been a few years - so go easy on me. If the question is how small a sample size can be to accurately represent a population, the answer is 'pretty damn small' - surprising small for that matter.

More important than size (in this case size doesn't matter much) is WHO the sample size is. There is random sampling (whoever is walking by), quote sampling (try and get a representative population) and variations. Who and where the meters are is much more significant than how many there are, in fact, having too many out there can actually DECREASE the accuracy of the reporting.

We'd get into things like standard deviation, chi square values, means, modes and medians that would start to make my head explode and you all go to sleep but we can go there is you want.

Suffice to say the 'problem' if there is one, is likely not HOW MANY, it WHO and WHERE.

Click to expand...

Another piece of the puzzle is what you're trying to measure. 1300 people is a pretty decent sample size for, say, a poll on who people on going to vote for in an election that has two choices. It was also a pretty good sample size back in the days when the top rated radio station in a market might have had a 10 share.

Without pulling out my old stats text, I'm thinking that the confidence on a sample that size is going to be on th eorder of +/-1.5 or 2 percent. When trying to measure a radio station with a 1 share, that's a pretty large uncertainty.

scott salvatori said:

i think you nailed it. what normal business/proffesional person, wants to walk around with a geek box hooked up to them.

Click to expand...

What normal person posts on radio message boards?

do these people recieve any compen$ation for this?

Click to expand...

Yes, they get money/premiums, and there are incentives like gift cards. Since a whole household must particpate, there is a lot of family group pressure to keep it up.

TowerLamp said:

Radio was sold a bill of goods by Arbitron which has a virtual monopoly on radio ratings.

Click to expand...

The system was pretty much demanded by the big agency-type advertisers who felt it was time to get faster delivery of data and more frequent books than the diary could permit. Radio opted in because advertisers were demanding the same kind of technology in measurement as they got for other media.