strategy papers | articles about us
our clients | our team | contact us
buy stuff | cool audio | book 'em
our picture gallery | New Walter Sabo Video

Arbitron Consultants Fly-In

Source: Harry Valentine -- Dec 09, 2002

For the past year or two, Arbitron's Portable People Meter has been a hot topic. During that time it has seemed that the PPM has been the centerpiece of what we have heard from the company. Last Friday's Consultants Fly-In at Arbitron headquarters in Columbia, MD was a return to a more typical program. The PPM was covered, but only as part of a well-rounded agenda that also included a diary service research update, a simpler way to credit programming, streaming usage and Arbitron MeasureCast ratings, PD Advantage and language weighting. Here are some highlightsfrom the day.

Diary Service Update

While Arbitron still maintains that the PPM is the future, it was made abundantly clear that their commitment to the diary is stronger than ever. To that extent, they report spending more than ever on testing it.

Perhaps the biggest issue with respect to diary measurement is response rate. It's down, but Arbitron is not to blame. As with nearly every telephone research operation, response rates are near their lowest levels on 20 years. Arbitron offered some possible reasons for the decline in response rates and they all made sense:

• Too much telemarketing
• Improvements in privacy technology
• Respondent concern about being the target of a scam

It's important to make a distinction here. Response rates are low because of low consent rates. Diary return rates, once a respondent has agreed to participate, are very good, near their highest levels in 20 years.

Arbitron has been working on a number of things too increase response rates but one of the most encouraging ideas is being called promised incentives. It's a simple premise; Arbitron will pay participants to return their diaries.

Diary premiums, of course aren't new. But promised incentives are a little different. Arbitron slightly reduces the normal diary premium to a range of $1.00 to $3.00, but promises an additional $10.00 per person for a completed diary. Arbitron has tested this plan with black, Hispanic and young adult respondents and noted double digit increase in response rates. As of this fall's survey, promised incentives are being used in all top 10 markets except Detroit. In the spring, Arbitron will test promised incentives with a $1.00 diary premium and reduced follow-up incentives.

In phase 1 of the winter book, Arbitron is planning to test promised incentives off line for consent, paying $5.00 for each respondent who agrees to take the survey materials.

Some other initiatives to increase response rates being studied by Arbitron are:

• Trying boxes for pre-placement. Everyone has been receiving boxes with survey materials since 1999 and they have worked well.
• Pre-placement invitations in order to convert people who initially decline to participate. With Arbitron, you have to refuse twice.
• E-consent: Allowing people to consent to survey participation for a $5.00 premium via a secure Arbitron web site. E-consent was pilot tested this fall with more testing planned for Winter 2003.
• A cell phone summit planned for February. Arbitron wants to see if they can interview cell phone users. Previous experience has yielded a surprising low refusal rate. The problem is getting through to the cell phone users instead of their voice mail.

A Simpler Way To Credit Programming

In the past few years, some stations have played games with the personalities-programs information they submit to Arbitron. Imagine that! As a result, there have been station disputes over submitted data, five to ten such challenges a week, according to Arbitron. Also, 40 percent of stations don't have complete information on file at Arbitron for any given survey. This has led Arbitron look for a better way to credit personality/program mentions, and they found one.

Instead of using the personalities-programs information submitted as part of the station information packet, Arbitron will now use diaries themselves to determine how to credit a personality/program diary entry without call-letters or frequency. Here's how it works:

• If elsewhere in the same diary, the program/personality in question is mentioned with a station, that station will get credit.
• If other diaries in the survey mention the program or personality only with one particular station, that station will be credited.
• If more than one station in the survey is mentioned in diaries in conjunction with the program/personality, ascription will be used to assign station credit.
• Arbitron will call stations to confirm broadcasts. There are specific procedures in place toensure Arbitron speaks with an appropriate person in management.
• Arbitron will wait to the end of the phase to credit entries lacking station identification. In this way they can see what all the diary keepers in the phase do before assigning credit.
• There are some very specific procedures in place for referencing diary comments. Historically, this has been the one place in diaries where a station mention has not counted.

As with everything else the company does, Arbitron has extensively tested the new personality-program crediting procedure. What they have gleaned from the various studies should not be surprising:

• About 3% of diary entries contain a program or personality mention.
• Only 0.44% of diary entries are lone program/personality mentions.
• 2% of the quarter-hours come from combined station/personality entries.
• Just 0.5% of the quarter-hours come from lone program/personality entries.

Arbitron found that the vast majority of diary entries were credited the same way using diaries compared to the personalities-programs lists. If anything, diary keepers yielded slightly more appropriate information. Unidentified listening dropped to 0.04% from 0.06 percent using the new procedure. The fact that Arbitron put so much effort into an issue that affects such a small percentage of diaries, illustrates the distance they are willing to go to have a quality product.

Arbitron's new way of crediting programming goes into effect in the Winter 2003 survey period. The bottom line is stations should continue to submit regular station information packets. Arbitron just won't need the programs-personalities listings anymore.

Streaming Usage and ARB MeasureCast Ratings

Bill Rose, VP/GM of Arbitron Webcast Services presented some data from the latest Arbitron Internet study. Here are some highlights of consumer attitudes found in the study. It's worth checking out the full study, which is available free on the Arbitron web site at www.arbitron.com.

Media And Entertainment

• Despite cynicism about the business and consolidation, consumers think radio is doing well. 59% say radio is getting better while 43% say the Internet, 37% say newspapers and 33% say television is getting better.
• Television is perceived to be the most essential medium (39%), followed by radio (26%), the Internet (20%) and newspapers (11%).
• Respondents thought that television was the "coolest and most exciting medium" (35%), essentially tied with the Internet (34%). However, among 12 to 34 year olds, the Internet dominates the "most cool and exciting" image (46%), well ahead of TV (29%) and radio (21%).
• Among media, radio is the dominant source for new music (66%), followed by television and the Internet at 11% each.

Broadband

• Residential broadband adoption has doubled over the past 18 months. As of July 2002, 28% of Americans, about 38 million, have access to Internet at home (Jul 02), up from 13% in January 2001.
• 13% of respondents planed to get broadband within the next 12 months.
• 83 million Americans have tried streaming video or audio. That's 35%, up from 30% 18 months previous to this study.
• Three-quarters of "streamies" are 18-54 years old. 12-17- year- olds represent 19% of streamies.
• Streamies are more upscale and better educated than the general population. 57% of monthly streamies have a household income of $50,000+, compared with 38% in the general population. 45% are college graduates or higher, compared with 32% in the general population.
• People with residential broadband access use streaming media much more frequently than those with dial-up access.
• 36% of people online have listened to radio stations online. 48% of those listen to stations in their own area and 41% listen to stations from other parts of the U.S.

Audience Measurement

Arbitron acquired the streaming measurement license and related assets from MeasureCast on November 7, 2002. MeasureCast uses the same basic method as Arbitron's Webcast ratings. Arbitron is now in the process consolidating; transitioning stations or channels that were in Arbitron but not in MeasureCast.

The days of free Webcast ratings are going away. Arbitron will be evolving this into a subscription service. Free trials will be provided for a limited time, after which users will be offered the service commercially. A rate card and policies are in the works now.

Arbitron recognizes that an industry standard for webcast ratings is required to fuel ad sales and build credibility of the medium. They are talking about using traditional metrics to mimic Arbitron rules; average quarter-hour, time-spent and aggregate tuning hours, all of which involve the five-minute rule. There will also be Internet metrics, which remove the five-minute rule.

Nevertheless, there are some pretty significant variations between radio and streaming measurement because the media are used so differently. To start with, only about eight percent of Americans have streamed over the past week compared with 95% who have cumed radio. And that eight percent is sliced up many ways because of the thousands of choices available online.

Arbitron will use server-based measurement to collect data directly from the computer that sends information to the user. Server-based measurement gives a near census, not just a sample. It measures all tuning events with maximum reliability and almost no respondent burden. Server-based measurement yields representative data regardless of audience size. This is a crucial issue for the online audience since there are so many choices.

In order to get a near-census, there are some trade-offs. Only those who provide data can be measured so there is no share. And cumes equal PC's, not people.

Measure Cast provides monthly, weekly and daily reporting. Data is available within one week. Demos are available at the network level and come from an online survey, which is married to thequantitative data. Respondents are instructed to turn on a global unique identifier in their media players during the online survey. About 40 percent of total sample has it on. Arbitron then uses modeling to put the numbers together and have a good estimate.

PD Advantage

Version 4 of PD Advantage was just released. Arbitron says stations should have it now. We were treated a live demo of some of the software's features. There is a lot you can easily see about your station from this software including:

• How your station's vital signs are trending; and how your station's vital signs compare to other stations, the national average and top performers in the format.
• How much cume or time-spent listening you need to increase your share to a certain level.
• A five-book range of cume or TSL in which your station operates. Arbitron called this "the sandbox in which the station plays. It gives a reality check for the station based on history and looks like a good tool for planning.
• A leading indicators report. Arbitron says this is the report that everyone is gravitating to. The leading indicators report takes seven key elements and puts them into trend graphs plus a five-book average. The key elements in the report are cume, share, TSL, P1 cume, P1 AQH, and 100+ quarter-hours.

Language Weighting

Dr. Ed Cohen, Arbitron's VP of Domestic Research, discussed the issue of language weighting, specifically Spanish language weighting.

A little background: Arbitron has been tracking Hispanic diarykeeper language preference since 1997 and reporting audience estimates by language preference since 1998. In Winter 2002, Arbitron changed the way they find out language preference, asking the question in the recruiting call rather than in the diary. Participants are classified as "Spanish primary" if they report speaking only Spanish or mostly Spanish in the home.

In a recently completed study, Arbitron learned that their in-tab Spanish primary language estimates are lower than Nielsen's estimates, which for now Arbitron considers the standard. Since the U.S. Census does not provide information on language preference, Arbitron is reviewing alternatives, including the Nielsen estimates, to use as a benchmark for the universe of people whospeak Spanish primarily.

The result of this is Arbitron's commitment to language weighting. If the sample percentage of Spanish-dominant is lower than what is ultimately determined to be the population's percentage, the samples will be weighted higher. There is presently no set date as to when it will happen. Arbitron's current processing system can't handle it but they are working on a new system that can and will set a target date around the beginning of 2003.

Language weighting is not likely to be used in all 80 Hispanic DST markets. Most will continue to have Hispanic weighting only. Arbitron expects language weighting in potentially 25 markets. Arbitron is just starting to talk with Nielsen about this. And Arbitron is being careful because they need something defensible in terms of the primary Spanish-speaking population estimates. But they consider it a key priority. Watch for more on this issue early next year.

Portable People Meter

Arbitron saved the Portable People Meter carrot for the end of the day, saying that they focused on their diary service by design. Overall, Arbitron has had an excellent experience PPM in Philadelphia this year. A total of 86 media outlets, including 47 radio stations have been encoding. Items on the PPM agenda included the KYW encoder issue, PPM time line, latest audience estimates and PPM programming insights.

KYW Encoder Issue

You may have read in the trades about a problem discovered in the last few weeks involving KYW Radio's PPM encoder. The problem was not actually with the encoder; it was with the volume of broadcast material coming from the station being fed to the encoder. The encoder was working properly, but the input to the encoder was below specifications. There was no way to tell how long the problem was going on.

Arbitron discovered the issue when replacing the encoder at KYW with a newer version. Arbitron tested the encoder output and found it very low. There were nowhere near as many codes as there should have been. The encoder is designed to put out a code every 4 or 5 seconds. The PPM is looking for codes constantly but it captures only one code every 30 seconds to minimize its storage requirement. That gives the PPM redundancy, since it only needs to capture a couple of codes a minute. But KYW's level was so low that the redundancy was used up. The ratings data bumped up slightly when the input level is fixed.

In response to the problem with the input level to the KYW encoder, Arbitron checked airchecks from all other media outlets to make sure code levels were OK. Four other radio stations (WZZD, WPHT, WILM and WHYY-FM) had low levels, but the levels were high enough that enough codes were coming through. Arbitron quickly decided to extend the PPM test through the first quarter of 2003.

Arbitron is being very thorough in their response to KYW's PPM input level problem:

The present encoder monitoring system was set up to say yes or no; that the unit was working or not. Arbitron is working on a new monitoring system to catch problems like this, a system with gradations that shows how well the encoder is encoding.

Arbitron is changing its procedures to emphasize the importance of input levels to the encoder.

A feature that emphasizes input specifications is being added to the back of the encoder.

PPM Timeline

It was emphasized that Arbitron's commitment to the Portable People Meter has not changed. They are still committed the commercialization of the PPM. However, Arbitron's timeline for the PPM has been modified in response to:

• Radio's request to slow down,
• The need for further research on response rates,
• Television requests for research, and
• Arbitron's desire to pace its R&D expenditures.

The PPM time line is vague because it is a function of Arbitron's success with response rates, the outcome of lifestyle research and other TV tests, support of broadcasters and meaningful progress in discussions with Nielsen about the planned PPM joint venture. Arbitron obviously needs the Nielsen joint venture and customer support for the PPM to go to market in the U.S.

Arbitron will carry out additional research in 2003. The current Philadelphia panel will be downsized and converted to a research lab panel, which will remain in place after March of 2003. Arbitron says it is important for Philadelphia media to keep encoding through 2003. Due diligence work is active between Arbitron and Nielsen, but it's too early to make any predictions. Arbitron estimates that the PPM could be commercialized in 2004.

PPM Ratings

Arbitron compared overall PPM data from the spring and summer survey periods. In the spring, AQH ratings were up in all demos except 55+. In the summer book, diary data was higher than that of the PPM in every cell except teens. Here is a comparison of total 12+ PPM AQH ratings from spring to summer:

AQH Ratings

Arbitron showed similar results for weekends and male respondents. However, the male data was very close to the diary data in the summer. The disparity was more with in females.

As we have heard before, because of the small Hispanic population in Philadelphia, Hispanic panel size very small. Arbitron reported that with the Hispanic panel there were smaller reductions in AQH ratings from spring to summer and PPM numbers were still higher than the diary in the summer.

Arbitron admits not having answers now, but they are looking at many things to determine what caused the shift of PPM AQH ratings from spring to summer:

• Impact from panel turnover
• Proportionality of the panel in the summer
• Panel compliance
• Typicality of summer data
• Impact of vacations
• Impact of commuting in the summer vs. the spring
• Changes to station programming in the summer
• Whether it was a TSL or cume driven difference
• Panelists with and without kids

A slide was shown comparing PPM numbers over time from black respondents with and without children. Moving into the summer, there was a growing gap of discrepancy between diary and PPM ratings. While households without children stayed consistent, there was a reduction in listenership in black households. I asked if they had similar data for non-ethnic households with or without children. One Arbitron rep said there similar was a gap, but that it was not as dramatic.

While there were differences in TSL, the cume story didn't change much from spring to summer. With the diary service, only one station had a cume rating of 20 percent or more. With the PPM, 16 stations did.

As we have heard previously, while the average diary keeper records three stations in diary, the PPM captures six or seven radio stations per week, depending on demo. The numbers of the top three stations are down a little but aren't that different. What causes the average time-spent listening to drop with the PPM is the fourth through sixth stations, which have very low TSL.

PPM Programming Insights

One of the things we like most about the PPM is its sensitivity, or ability to accurately measure very short periods of listenership. To that extent, Arbitron treated us to graphic examples of what happened to listenership with various broadcast events on radio and TV.

• A Springsteen interview yielded a jump in listenership.
• A radio football game caused a dramatic increase in listenership.
• Dr. Phil boosted KYW TV's 3-4 PM numbers.
• A major guest on Howard Stern's show made the numbers rise. The numbers were soft during Howard's vacation.
• Numbers spiked for the final American idol audition.
• KYW Radio got a big increase after a sniper shooting.
• A Rolling Stones ticket giveaway made the numbers go up. In this case, it was TSL. The cume was not particularly high that day.
• KYW Radio spiked on a day of lotto winner announcements.
• KYW Radio had big increases on the day they gave away a computer every hour.
• WIP has higher Monday and Tuesday morning drive ratings, presumably due to football fans interested in sports the morning after a game.
• WPHT got a ratings spike on 9-11.

IMPORTANT: With the PPM, for the first time, radio will have access to the same type of information that TV has used to make programming decisions, accurate listenership information for short periods of time. Events like those mentioned above don't show up in diaries.

The movement of PPM and diary data are incredibly correlated. A three-week trend line for the diary matches up pretty well with the PPM. There's just less fluctuation with PPM because of its ability to much more accurately measure listenership in short periods of time.

Back to Index