PROGRESS TOWARDS
MEDIA MIX ACCOUNTABILITY
Portable People Meters' (PPMTM) preview of commercial audience results Roberta M. McConochie Leslie Wood Beth Uyenco Chris Heider
PPM commercial audience estimates offer insight about consumers' avoidance of tv commercials. Total commercial avoidance, an average 7%, is composed of nearly six- tenths channel switching and four-tenths due to other "interruptions." Program content appears to be the strongest predictor of avoidance. Gender and age exacerbate commercial avoidance with men, teens, and younger adults showing above-average churn. There is also variation in the relationship of exact, commercial-minute audience levels to average-minute audience. High-churn formats produce lower indices with even lower levels for men. In other words, there is potential for bias against particular media formats and particular targets in today's currency-based "proxy" measures of commercial audience -- average minute and AQH. PPM and Apollo estimates would help identify sources of bias and alternatives going forward. These results represent progress toward quantifying the mechanics of commercial avoidance for buyers and sellers. They also demonstrate the value of PPM's near-passive, direct, and precise capture of persons'-level media exposure. � Copyright by ESOMAR� / The ARF 2 Roberta M. McConochie, Leslie Wood, Beth Uyenco, Chris Heider INTRODUCTION
Background
In 2004 and 2005, worldwide advertisers have become increasingly vocal about the deficiencies of today's advertising models and repertoire of measurement tools (e.g. Glock, 2004, Stengel 2004). "Today's marketing model is broken. We're applying antiquated thinking and work systems to a new world of possibilities," according to Stengel. In Glock's words: "Marketing metrics are not keeping pace with marketing needs. We need to establish new metrics." Marketers' concerns, perhaps fueled by challenges of demonstrating the contribution of advertising to ROI (Campbell, 2005), have inspired one recent innovative information service: Project Apollo, the working title for the Arbitron-VNU planned service, with the collaboration of P&G, would link multi-media message exposure and consumer attributes with purchase behavior to quantify ROI (Dupree and Bosarge, 2004; McConochie, 2005). Beyond the measurement and ROI issues, a portion of the challenges facing marketers emanates from the abundance of competing, conflicting marketing messages. Commercial television serves up roughly one minute of nonprogramming content for every three minutes of substantive information and entertainment in the United States. Across all TV networks, total non-program clutter, including commercials, public service announcements and promotions, comprises over one-quarter of total programming (28.5%, according to Papazian, 2004). Primetime television commercial clutter has increased by 60% over the past two decades. Increasing clutter threatens the advertiser with the specter of commercial avoidance, particularly for "intrusive media like TV and Radio" (Ephron, 2005). Commercial avoidance may describe the behavior of over half of total prime-time viewers (Knowledge Networks, 2004). Research Questions
This investigation focuses on understanding consumers' audience choices visa-vis television commercials � whether they remain in the audience or not when a commercial or a series of commercials occurs. These results build towards estimating advertising ROI. More specifically, the authors address the following key questions: � Copyright by ESOMAR� / The ARF Progress towards media mix accountability 1. What is the extent of commercial-churn, specifically avoidance of specific advertisements? How does this relate to program context? 2. How does commercial-minute audience compare with context program audience at the average-minute and quarter-hour levels? 3. What is the impact of commercial load (commercials per pod and per hour) on commercial-minute audience? 4. Do the first ads in a pod produce less churn than subsequent ads? How does churn cumulate over units in a pod? The authors report results for the first two questions in this paper. Questions three and four will be addressed in the presentation. Though not an intrinsic part of Project Apollo, this investigation moves another step towards understanding whether and under what conditions commercials would actually reach consumers. Project Apollo will collect more precise information on ad-specific audience via encoding individual 15, 30 and 60-second commercials which can be "read" directly via the Portable People Meter (PPM)SM 3 Methods
Media data for the present investigation were obtained via Arbitron's Portable People Meters in Philadelphia in 2003. The in-tab sample included approximately 1,000 persons age six and older. This investigation focuses on one week of information, October 20-26, 2003. During this time, the PPM system measured six encoded broadcast stations and 26 encoded cable outlets, the largest cable channels in terms of audience. Radio stations were also measured, though these results are not reported in this investigation. The PPM streams of individual persons' media-exposure episodes, start and stop times, were integrated with independent third-party commercial verification data provided by Nielsen's Monitor-PlusSM. The Monitor-Plus commercial start and stop times were merged with PPM start and stop times of individual media exposures at the individual person's level. Over 100,000 individual commercials comprise the one-week database of this investigation. These include broadcast network and spot ads and cable network commercials. Local cable commercials are not included in the Monitor-Plus information and therefore are not part of this research. The present PPM data arguably offer more precise capture of individual persons' media exposures than today's worldwide currency alternatives. According to one of the authors of this paper representing the buying-side of media, Beth Uyenco, � Copyright by ESOMAR� / The ARF 4 Roberta M. McConochie, Leslie Wood, Beth Uyenco, Chris Heider "While the people meter is acknowledged as the best TV currency system we have, many have questioned its ability to capture accurate viewing behavior because it requires consistent button pushing for respondents to indicate whether they are in the viewing room or not. In other words, the people meter does not directly track whether people are viewing or not. A viewer may leave the room during a program break and neglect to push his or her button before leaving in which case the people meter identifies the viewing as continuous. The persons-level precision of PPM reflects the fact that individual panelists wear their own individual meters and that the meters passively capture exposure to media. In contrast, traditional Television currency measures television sets rather than persons. Traditional TV measurement also requires active participation � either writing in TV use in a paper diary or pushing a button each time each person starts and stops watching or listening." The strengths of PPM measurement at the persons' level also have been reported to the industry in previous research, e.g. Pellegrini and Purdye, 2004. This investigation's approximation of commercial audience, the result of merging two independent databases, may underestimate "true" commercial impact on consumer behavior. Also the present results, while arguably the most precise capture of persons-level audience, may be somewhat limited in precision at the granular level of individual commercials. For example, the present minute-level data do not perfectly estimate the audience to individual 15 and 30-second commercials. The authors believe that this limitation would tend to understate the "true" difference between commercial audiences and program audiences. The present one-week investigation thus sets the stage for a fuller, richer look at data from the PPM evaluation in progress in Houston Texas as well as a hint at what Project Apollo will find when specific commercials are encoded and the data are precise to the level of 15s as well as 30s and 60s. RESULTS
Extent of Commercial Avoidance: Seven Percent
This investigation defines commercial avoidance as the portion of persons exiting the audience on the base of total audience just prior to the commercial minute. The total commercial-minute audience includes new arrivals, continuing audience, as well as those leaving the audience. Previous limited investigations of PPM data, confined to a subset of brands and categories, have shown commercial avoidance levels of roughly 5% (McConochie et al, 2004). Including total categories and brands produced an overall level of avoidance of 7.3%, in the same range as the previous research. � Copyright by ESOMAR� / The ARF Progress towards media mix accountability Since the PPM data track each individual person's exposure to Television, minute-by-minute, and include each person's change in media status, in terms of discrete media episodes, PPM also can identify the audience flow from a previous episode. In this way, the PPM database indicates whether the exit from the commercial occurs when consumers switch to another TV channel vs. whether they are no longer exposed to TV. We use the term "channel switching" to describe the first, and "other interruptions" to describe the latter which would occur for example when the consumer turns the set off, exits the room, or is assaulted by other competing stimuli such as babies crying, dogs barking, etc. 5 Majority of Commercial Avoidance Due to Channel Switching
Given that most persons in the United States use remote controls to change channels, and that media and other attention spans appear to be dwindling in this age of quick-cuts, the authors suspected that the majority of avoidance would be due to channel switching. It was. Of the 7.3% total avoidance, shown below, 4.1 points came from channel switching. The remaining 3.2 points of avoidance were due to other interruptions of audience to television use. Total Commercial Avoidance Channel Switching Other Interruption of Audience 7.3% 4.1% 3.2% On the base of total commercial avoidance, in other words, setting the 7.3% to 100%, nearly six of every 10 commercial-avoidance episodes are followed by channel switching. Channel switching levels are similar for broadcast network affiliate and Cable channels (55% for Broadcast and 59% for Cable channels) as shown in figure 1. � Copyright by ESOMAR� / The ARF 6 Roberta M. McConochie, Leslie Wood, Beth Uyenco, Chris Heider Figure 1 CHANNEL SWITCHING COMPRISES NEARLY 60% OF TOTAL COMMERCIAL AVOIDANCE
80 59.2 60
% Avoidance Composition 54.7 45.3 40.8 40 20 0 TV Channel Switching Broadcast Cable Other Interruptions* *E.G. set turned off; leave room Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM, P6+ The remaining portion of avoidance resulting from other interruptions of viewing comprises 45% for Broadcast and 41% for Cable channels. In other words, the majority of audience exits from a commercial arguably reflect a decision to look for alternative Television content. To this end, the authors mine the data to look for the variables that account for switching and other avoidance, variables that provide clues for media sellers and buyers to retain their commercial audience. First we look at variations by broad dayparts. Daypart Variation Accounts for a Small Portion of Avoidance
Given that viewing and other consumer behavior varies by daypart, we expected variation in commercial avoidance as well. The authors suspected higher avoidance when people were able to sit in front of the TV, focus on the screen, and potentially react to content as is the case during Prime Time. At other times, we thought that less viewing with less concentration and more competing activities might result in less avoidance, though possibly more � Copyright by ESOMAR� / The ARF Progress towards media mix accountability "other interruptions" in viewing. Indeed the results showed some variation in avoidance in the expected direction. As expected, the highest levels of avoidance occur during evening programming. During this time, total commercial avoidance exceeded the respective averages for broadcast and cable channels. Total avoidance for 8 pm � 11 pm broadcast was 8% vs. the broadcast average across dayparts of 7% (figure 2). For 8 pm � 11 pm cable, total avoidance was 11% vs. the cable average of 10%. The slight differences imply that broad dayparts account for a small portion of variance in commercial avoidance Figure 2 FROM 8PM-11PM, TOTAL COMMERCIAL AVOIDANCE AT HIGHEST LEVELS
40 7 35 30 Perc ent Avoi d anc e 25 20 15 9 .4 5 .6 5 5.7 8 .6 5 .8 6 .4 9.3 6 .9 10 .3 7 .5 1 0 .8 8.1 6.7 10 8.4 0 7A M -9 AM 9A M -1 2N o o n 1 2 N o o n -4 P M 4P M -6P M 6 P M -8 P M C a b le Av g . = 9 .5% 8P M -11 P M 1 1 P M-1 AM B ro a d ca s t Av g . = 6.6 % Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM, P6+ Figure two also shows consistent difference in commercial avoidance levels between Cable and Broadcast channels. These differences and the likely causal factors are discussed further below. � Copyright by ESOMAR� / The ARF 8 Roberta M. McConochie, Leslie Wood, Beth Uyenco, Chris Heider Channel Switching also Highest During Evening, Prime Time
Also consistent with expectations, channel switching away from commercials reach maximum levels during Prime Time, and do so for both Broadcast and Cable channels (figure 3). For Broadcast, the Prime Time channel-switching level was 4.9% as opposed to less than half that in the morning. Similarly for Cable, Prime-Time switching was nearly twice that for mid-morning and early afternoon. Figure 3 EXTENT OF CHANNEL-SWITCHING ALSO HIGHEST BETWEEN 8 � 11 PM
20 15 10 5 .6 5 0 7 AM-9 AM 9 AM -12 N o o n 12 N o o n -4 P M 4P M -6P M 6P M -8P M 8 P M-1 1P M 1 1 P M-1 AM 2 .3 2 .2 5.7 2.6 5.8 3 .3 Broadcast
6 .4 3 .8 6 .9 4 .9 7.5 4.3 6.7 20 15 10 4 .5 5 0 7 AM -9 AM 9 AM -12 N o o n 12 N o o n -4P M 9 .4 4 .2 8.4 4.3 8.6 Cable
9 .3 4 .7 6 .2 1 0 .3 7 .6 5.9 10 .8 8.1 4P M -6P M 6P M -8P M 8 P M -1 1P M 1 1P M -1AM Channel Switching Avoidance Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM, P6+ Next, we take a deeper dive into the daypart data to look for further explanatory clues to explain variations in commercial avoidance. The hour-byhour avoidance data against the background of audience average-quarter-hour ratings build shed further light on avoidance variation. Broadcast Commercial Avoidance Builds to Evening High
The hour-by-hour avoidance results show a build over the day peaking from 7 pm to Midnight. Some of the lowest avoidance levels occur at 8 am (6%), vs. the 8% levels between seven pm and Midnight. � Copyright by ESOMAR� / The ARF Progress towards media mix accountability Figure 4 BROADCAST NETS' AVOIDANCE LEVELS HIGHEST DURING EVENING
40 35 30
Perc ent Avoi d anc e 9 25 20 15 10 5 .8 5 0 5 .5 5 .5 5 .7 5.9 5 .8 5.7 5 .5 6.2 6 .5 6 .4 5.9
A QH Ra ti n g 7 .7 7.5 7 .6 7.4 7 .0 3.6 11A M 10AM 12No on 11PM 7AM 8A M 9AM 1PM 2PM 4PM 6PM 8PM 10PM AQ H R atin g Av o id a n ce Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM, P6+ The cable hour-by-hour data also show a smooth build to the highest avoidance levels at 9 pm (11%) which exceed those occurring over most other times (figure 5). These avoidance curves for both Broadcast and Cable appear to confirm our hypothesis. During evening viewing, when people focus on their televisions, perhaps more so than at other times of daily routines, there is also more channel switching and commercial avoidance in general. � Copyright by ESOMAR� / The ARF 12M id 3PM 5PM 7PM 9PM 10 Roberta M. McConochie, Leslie Wood, Beth Uyenco, Chris Heider Figure 5 CABLE COMMERCIAL AVOIDANCE ALSO HIGHEST DURING EVENING
40 35 30
P ercen t Avo id ance 25 20 15 10 .4 10 5 0 8 .7 8.9 8.1 8 .3 8.9 7.4 8 .8 9.3 9.3 9 .3 10.1 10 .4 10.5 11 .2 1 0.6 10.1 6.3
AQH Ratin gs 11 A M 7AM 9A M 1 2N o on AQ H R a ting Avo ida nc e Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM, P6+ Despite the build in avoidance over the media day, daypart variation appears to account for a small portion of commercial avoidance. But this small variation is quite likely related to programming content as well as daily routines. We therefore progress to programming to help explain commercial avoidance. We look first at channel-by-channel variation then at program genres and specific programs where there are sufficient commercial data. Broadcast Channels, Genres and Programs Show Substantial Variation in Avoidance
Proprietary research conducted by DDB leads the authors to predict that genres and program content would account for substantial variation in commercial avoidance. Results confirm this prediction. The extent of commercial-audience variation across broadcast channels is roughly on par with that for dayparts. The highest overall avoidance level is 8% for Fox vs. a low of 6% for CBS (figure 6). Fox also shows the highest switching level (5%) vs. a low of 3% for low-avoidance ABC. The range of avoidance across broadcast channels is slightly higher than that for dayparts. � Copyright by ESOMAR� / The ARF 1 2M id 11 P M 10A M 6PM 9PM 8A M 10P M 3 PM 1P M 2P M 4P M 5P M 7P M 8P M Progress towards media mix accountability Figure 6 AVOIDANCE AND CHANNEL-SWITCHING VARY ACROSS BROADCAST NETWORKS
40 35 30 11 Pe rc en t A v oid an ce 25 20 15 10 4.7 5 0 Fo x WB U PN NB C ABC CBS 8 .1 4.6 8.0 4 .2 6 .8 3 .7 6 .4 3 .2 5 .7 2 .6 5 .6 C ha n n el S w itc h in g Avo id an c e Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM, P6+ Focusing on specific program genres across broadcast channels yields greater commercial-churn variation. Program Genre and Individual Program Content Explain Substantial Variation in Broadcast Commercial Avoidance
The impact of genre on avoidance gives more pronounced "effects" than for individual stations. For example, sports events show 9% overall avoidance levels vs. only 4% for daytime drama. The range of switching is also large: from 5% for sports events and adventure programs vs. only 1% for daytime drama (figure 7). In other words, sports and other action programming show more than twice the commercial churn of Soap Operas, confirming the hypothesized powerful impact of program content on commercial stickiness. � Copyright by ESOMAR� / The ARF 12 Roberta M. McConochie, Leslie Wood, Beth Uyenco, Chris Heider Figure 7 BROADCAST GENRES' TOTAL AVOIDANCE RANGES FROM 9% FOR SPORTS TO 4% FOR DAYTIME DRAMA
40 35 30 P ercen t A void an ce 25 20 15 10 5 .1 5 0 S p o rts E ve n t Ad v en tu re Fe a u re Film Ge n era l D ra ma S itu atio n C om e d y T a lk Show s K id s An ima tio n N ew s D ay time D ra ma 8.9 5 .4 7 .9 4 .4 7 .5 4 .5 7 .2 2 .7 6 .2 2 .9 5 .9 2 .6 5 .5 2 .6 5 .5 1 .3 4 .2 C h a n ne l S w itch in g Av oid a n ce Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM, P6+ Variation of commercial avoidance among specific programs is even greater than that for genres and channels, further confirming the impact of program context on commercial audience. For example, commercial avoidance ranges from 13% to 6% across a set of fall football and baseball events (figure 8). The Baseball World Series foregone-conclusion event earns the highest avoidance score at 11%. A pre-game event for football shows the second highest "avoidance" at 14%. Perhaps not surprising, the Philadelphia home team football game, a close scoring win for the Philadelphia Eagles, shows the highest holding power with only 6% commercial-audience loss. Variation in channel-switching shows similar large variation � from 8% switching for the CBS NFL pre-game to only 3% switching for the Eagles game. The large variation implies the importance of program context in planning effective communications campaigns and in marketing commercial inventory on the selling side. � Copyright by ESOMAR� / The ARF Progress towards media mix accountability Figure 8 ONE-SIDED WORLD SERIES GAME YIELDS TWICE THE AVOIDANCE OF HOME-TEAM EAGLES FOOTBALL
40 35 30 25 20 15 10 5 0 W o rld S e ries Ga m e 5 (F o x) S u n d ay N FL P re g am e (C B S ) W o rld S e ries P re g a me Avg (F o x) S u n d ay NFL (F ox ) S a turda y C o lleg e F oo tb a ll (AB C ) Mo n d ay N igh t Fo o tb a ll (AB C ) S u n da y NFL P re g am e (Fo x ) W o rld S e rie s Ga me 3 (Fo x ) E a g les vs . Je ts (C B S ) 7 .4 12 .6 7 .5 13 Percent A Voida nce 11 .4 7 .9 11 .2 5 .8 9 .7 4 .2 9 .3 6 .3 9 .2 6 .8 8 .9 4 .8 8 .8 5 .8 3 .1 C h an n e l S w itc h in g Av o id an c e Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM, P6+ In contrast to the large variation among specific sports programs, individual soap operas show more consistently low avoidance. All soap opera programs for this investigation show 5% or less total avoidance: The range is from 5% for The Bold and the Beautiful to 3% for The Young and the Restless and As the World Turns (figure 9). In departure from overall averages, channel switching was less than half that of total avoidance, ranging from a "high" of 2% to a low of under 1%. The implied higher levels of "other interruptions" to viewing appears consistent with daytime routines of house husbands and wives, dealing with a relatively constant flow of typical interruptions of children, telephones, dogs, visitors, etc. � Copyright by ESOMAR� / The ARF 14 Roberta M. McConochie, Leslie Wood, Beth Uyenco, Chris Heider Figure 9 DAYTIME DRAMA SHOWS CONSISTENTLY LOW TOTAL AVOIDANCE AND CHANNEL SWITCHING
40 35 30 Perce nt A voidan ce 25 20 15 10 5.3 5 0
T h e B o ld and the B e au tiful P assion s G u idin g L igh t O ne Life to L iv e All M y C h ild re n G e ner al H o sp ital D ays o f o ur L iv e s T he You ng an d th e R e stle ss As T he World T urns 4.9 1.4 2 .3 4 .8 1 .2 4 .7 1.5 4.6 1 .2 4 .3 2 .1 4 .3 0.9 1 .8 3.2 0.9 3 .1 C h a n n el S w itc hin g Av o id a n ce Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM, P6+ Channel Format and Genre Yield Even More Variation in Cable Commercial Avoidance
Turning to Cable channels, we see large variation across the channels, more so than among Broadcast stations. The format-driven Cable networks provide markedly different contexts for marketing messages with large differences in commercial avoidance. At the high end, VH1, Weather Channel, and MTV show commercial avoidance levels of over 14%, nearly three times that for the 5% levels of A&E and Lifetime (figure 10). As with the Daytime Drama genre, A&E and Lifetime show extremely low levels of channel switching, 3% or less. � Copyright by ESOMAR� / The ARF Progress towards media mix accountability Figure 10 AVOIDANCE FOR MUSIC AND WEATHER FORMATS THREE TIMES HIGHER THAN FOR A&E, LIFETIME
40 15 35 30 Percent A void ance 25 20 14.6 14.5 11.6 10.4 10.7 15 14.4 10.1 10.2 10.2 10 9.2 9.2 6.9 7.3 6.4 6.1 6.4 5.4 5.6 6.8 10 8.8 5.4 3.9 4.4 3.9 0 VH1 TW C MTV B ET CNN FX E S P N T B S C H IS T N IC K TLC TNT A& E LIF E C ha n n el S w itc h in g Avo id a nc e Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM, P6+ Genre of programming across cable channels also shows substantial variation in commercial avoidance. Talk and music programming show the highest levels, over 14% total avoidance with Talk showing the highest level of switching, 13%. Cable genres with the least avoidance are similar to those for Broadcast channels: General Drama and Sitcoms, both with about 7% total avoidance. � Copyright by ESOMAR� / The ARF 2.4 5 3.2 5.2 16 Roberta M. McConochie, Leslie Wood, Beth Uyenco, Chris Heider Figure 11 SIMILAR DIFFERENCES IN AVOIDANCE RANGE FOR CABLE PROGRAM GENRES Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM, P6+ Thus far, program content appears to explain a substantial portion of variation in commercial avoidance and churn due to channel switching, far more than that for daypart per se. Obviously many of the high vs. lower-avoidance channels target quite different demos. So we proceed to look at the impact of age and gender on avoidance. The expectation here is for higher avoidance for younger demos and for Men. That's generally what the results show. Teens, Adults 25-44, and Men Show Highest Commercial Avoidance
Consistent with expectations, Teens 12 - 17 and Adults under 45 show the highest avoidance levels of any broad age group, for both Broadcast and Cable channels (figure 12). The highest broadcast avoidance is for Teens with 9% total avoidance. For Cable, Adults 35-44 show the highest level, 12%. As expected, the lowest avoidance levels occur for the oldest panelists, 65+ who show only 5% avoidance for Broadcast and 8% for Cable. � Copyright by ESOMAR� / The ARF Progress towards media mix accountability Figure 12 COMMERCIAL AVOIDANCE HIGHEST FOR TEENS, YOUNGER ADULTS
1 1 .6 11 9 .8 9 .2 7.9 7 .4 10 .1 9 .6 8.9 7 .6 6.3
Averages, P 6+ Broadcast: 6.6% Cable: 9.5% 17 % Avoidance 7 .1 6.5 7 .2 5 .9 5 .1 K id s 6-1 1 Te e n s 1 217 18 -24 2 5 -34 3 5-4 4 45 -54 5 5 -64 65+ B ro ad c as t C ab le Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM Also as expected, Men in virtually every age group show higher avoidance (figure 13). From age 25 and up, Men show slightly or considerably higher total commercial avoidance than women. The greatest gender gap occurs for cable-channel avoidance for older men, 55+. For example, Men 65+ show total cable commercial avoidance levels of 11% vs. 6% for Women. Though some of the age/gender differences are substantial, the estimated impact of age/gender is not as extreme as that for program content with its three-fold multipliers of commercial-audience avoidance. � Copyright by ESOMAR� / The ARF 18 Roberta M. McConochie, Leslie Wood, Beth Uyenco, Chris Heider Figure 13 MEN SHOW HIGHER AVOIDANCE THAN WOMEN, ESPECIALLY FOR CABLE
Broadcast Avg: Men 18+ Women 18+ = 7.1% = 5.6% % Avoidance 7 .9 6.2 6 .7 6.5 7.8 6 .9 7 .4 5.7 6 .9 5 .3 6 .2 4.5 1 8-2 4 2 5 -34 3 5-4 4 45 -54 5 5-6 4
Cable Avg: Men 18+ Women 65 + = 10.9% = 8.7% 1 2 .6 % Avoidance 7 8 .7 10 .3 10 11 .1 1 0.7 7 .9 12 8 1 0.6 6 .4 1 8-2 4 25 -3 4 3 5 -4 4 M en 45 -5 4 Wo m en 5 5 -64 65+ Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM Currency Measurement Units vs. Commercial Ratings
We turn now from the topic of churn to that of the unit of measurement and consider the appropriateness of today's average-minute and average-quarterhour currency "proxy" measures for commercial minute audience. Two factors potentially affect the relationships between commercial-minute audience and currency measures: computation rules and viewer behavior. Computation rules for minute-level processing. Procedures for the PPM 2003 database of this investigation awarded one minute of credit for each 30 seconds of encoded-media capture. Arbitron's investigations of the Philadelphia data indicate that the vast majority of media episodes extend well beyond 30 seconds. Therefore PPM processing would round up 30 seconds to one minute in what we estimate to be well under half of individual minute-level computations. To compute average minute estimates for this investigation, we calculated the arithmetic average audience over the 15 minutes in which the commercial occurred. In contrast, the commercial minute estimates focus on the exact-minute audience in which the specific commercial occurred. � Copyright by ESOMAR� / The ARF Progress towards media mix accountability Average quarter-hour processing. For this investigation as for today's quarter-hour currency, a minimum of five minutes of audience per quarterhour, whether adjacent minutes or not, earned credit for the entire fifteen minutes. The authors' guesstimate that the potential triple crediting of observed raw media-use data probably occurs in fewer than half of possible instances. Viewer behavior. We have seen above that viewer characteristics affect churn. Specifically, the genre forms of cable channels, and the viewing behavior of younger viewers and of men produce above average commercial avoidance, particularly for sports programs, and formats of cable channels such as music and weather. On average, we would expect exact commercial minute estimates to closely match average-minute estimates. However, high churn could occur for program content as well as for the commercials. We suspect that this is the case for those contents comprising the context for the high-churn commercials: sporadic, episodic, briefer discrete media-use episodes. If and when this is the case, then it is possible that the commercial minute estimates could index above 100 against the base of quarter-hour estimates. If this effect occurs, it would focus on those formats and demos that have shown higher-than-average avoidance and channel switching above. To the extent that the indices for commercial minute closely match those for the index base, average minute and average quarter, the currency procedures would appear to be a good proxy for commercial audience � as they are intended to be. To the extent there are differences, especially complex ones, then today's proxy measures for buying and selling may bias the selling, planning, and buying of specific media and targets. 19 Commercial Minute Audience Index against Average Minute: 98 for Broadcast; 89 for Cable
Because of the observed evidence of commercial avoidance, we expect commercial-minute audience would be less than that for average-minute program audience. For the quarter-hour indices, because of the greater rounding-up of quarter-hour estimation, we expect commercial minute estimates to fall further under the AQH standard. We report these comparisons as indices � computed against the appropriate base of average-minute data or against the base of AQH. Audience was measured in terms of GRPs for these estimates. � Copyright by ESOMAR� / The ARF 20 Roberta M. McConochie, Leslie Wood, Beth Uyenco, Chris Heider As expected, commercial minute data index at slightly less than 100 vis-a-vis the average minute data, 98 for Broadcast and 89 for Cable channels, reflecting the higher levels of audience churn for Cable formats (figure 14). The results imply that average minute is a good stand-in for commercial audience for broadcast, but possibly not so good for cable channels. Figure 14 COMMERCIAL-MINUTE GRP INDEX < 100 AGAINST BASE OF AVERAGE MINUTE: LOWER FOR CABLE
120 108 100 98 100 = Average-Minute Index Base
96 89 Index vs. A vg Minute 80 60 40 20 0 B ro a dc a st AQH C o mm er cia l M in u te C a b le Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM, P6+ Figure 14 also shows the index of AQH programming content against the base of commercial minute. Given the greater rounding-up factor for quarter-hour processing, we expected that the AQH index would be over 100. This is the case for Broadcast, with an index of 108. However, it is not the case for Cable programming which indexes at 96% against the average-minute base. We suspect this reflects the relatively brief discrete media-use episodes for a number of cable-channel formats. Again, these results point to the shortcomings of average-quarter-hour as a proxy for commercial minute � for both Broadcast and for Cable. When we display GRP indices against the base of AQH program viewing, we see the expected results inferred from the previous comparisons � and the limitations for today's U.S. local-market use of AQH to plan and buy � Copyright by ESOMAR� / The ARF Progress towards media mix accountability commercial time. Commercial minute indices for both Broadcast and Cable are well under 100: 91 and 93 against AQH the base of program viewing. The Broadcast average-minute index is also under 100 (at 93% of AQH). However, consistent with the previous index results, the Cable average-minute GRP indexes above 100, exceeding the AQH audience. Figure 15 COMMERCIAL-MINUTE GRPS INDEX LOWER THAN AVERAGE MINUTE AGAINST BASE OF AQH
1 20 10 4 1 00 21 100 = AQH Index Base
93 91 93 80 In dex vs. A Q H 60 40 20 0 B ro ad c a st Av era g e M in u te C o mm e rcia l M in u te C a b le Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM, P6+ Given the impact of program genre and channel on commercial avoidance (reported above), we expected these to make a difference on commercial index values. They do. For the remaining results, we focus on indexing against average-minute data. As expected, the broadcast channels with the highest levels of commercial avoidance do show lower indices against the average-minute audience (figure 16). For example, the Fox affiliate shows an index of 96 in contrast with the 98 index for all of the "Big Three" affiliates, NBC, CBS, and ABC. In other words, the bias of using average minute for placing commercials is probably not the same for different channel formats, nor for specific program content. � Copyright by ESOMAR� / The ARF 22 Roberta M. McConochie, Leslie Wood, Beth Uyenco, Chris Heider Figure 16 SLIGHTLY LOWER COMMERCIAL-TO AVERAGE-MINUTE INDEXES FOR NETWORKS WITH MOST COMMERCIAL AVOIDANCE
1 20 Broadcast
99 98 98 98 97 96 1 00 In de x v s . A v g. M in ute 80 60 40 20 0 AB C NB C C BS UPN WB Fox v s. Av era g e M in u te Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-12Mid, P6+ The same sort of expected results occur for the high-churn Cable-channel formats. For example, music channels MTV and VH1 show indices of 89 and 86 against their respective program minute audiences (figure 17). In contrast, Nickelodeon, A&E and Lifetime are close to or slightly higher than 100 index levels. � Copyright by ESOMAR� / The ARF Progress towards media mix accountability Figure 17 EVEN LOWER COMMERCIAL-TO AVERAGE-MINUTE INDICES FOR HIGH-CHURN FORMATS
120 10 2 .1 100 23 Cable
10 0 .1 9 9.3 9 8.5 9 8.0 9 7 .0 9 6 .7 96 .6 95 .4 9 3.6 8 8.6 8 6 .0 Ind ex vs. A vg. M inu te 80 60 40 20 0 N IC K CN N A& E LIF E TNT TBSC ESPN TL C FX BET MT V VH1 v s. Ave ra g e M in u te Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM, P6+ Further investigation confirms that, as expected, gender plays a part in explaining commercial-audience index levels. Results imply an interaction between gender and media form � Broadcast vs. Cable channels (figure 18). Men's viewing (vs. women's) produces slightly lower broadcast-channel indices for commercial minute against average minute (a 97 index vs. 98 for Women 18+). For Cable channels, the gender gap is greater: 86 for Men 18+ vs. 92 for women. � Copyright by ESOMAR� / The ARF 24 Roberta M. McConochie, Leslie Wood, Beth Uyenco, Chris Heider Figure 18 GENDER DIFFERENCES AFFECT COMMERCIAL-MINUTE INDEX LEVELS AGAINST AVERAGE MINUTE
C om m ec ial-M inute In dex 120 100 80 60 40 20 0 98 .6 91 .3 9 7.0 8 6.3 96 .2 82 .4 9 6.8 8 4 .2 96 .7 83 .5 9 7.6 8 3.0 MEN: Broadcast average = 97 Cable average = 86 1 8 -2 4 25 -3 4 3 5 -44 B ro a d ca s t 45 -5 4 C a b le 5 5 -6 4 6 5+ Com m e cial-M inute In de x 12 0 10 0 80 60 40 20 0 99 .3 91 .9 97 .0 88 .6 9 7.5 9 0.8 98 .5 92 .9 9 9 .1 9 1 .5 9 9.1 9 4.3 WOMEN: Broadcast average = 98 Cable average = 92 1 8 -24 2 5 -3 4 35 -4 4 B ro a d c as t 4 5 -54 C a b le 5 5-6 4 6 5+ Philadelphia PPM Evaluation, 10/20/03 - 10/26/03, Mon-Sun 7AM-1AM, P18+ We also expect from the commercial-avoidance/ churn results that younger demos might also show slightly lower index levels than older. However no consistent age effects are discernable in the results shown in figure 18. It may be that this one-week's worth of information on relatively small sample sizes may not be sufficient to discern age effects. We look forward to results from the PPM evaluation in Houston and to Project Apollo results to assess the impact of age on audience churn and commercial-minute index levels vis-a-vis average minute estimates. � Copyright by ESOMAR� / The ARF Progress towards media mix accountability 25 SUMMARY AND CONCLUSIONS
These results offer important early insights from PPM to both the sellers and buyers of media. Detailed PPM data confirm that the majority of commercial avoidance reflects channel switching, as opposed to other interruptions of TV use such as turning off the TV set or leaving the room where the TV is on. In other words, the majority of audience exits from a commercial arguably reflect a decision to look for alternative Television content. The extent of commercial avoidance depends upon media program content. Drama programs, especially the daytime soap operas, show far less avoidance than cable music channels and pre-game sports programs. Simply put, some contents appear to possess greater stickiness than others. Gender and age exacerbate program impact on commercial avoidance with Men and younger viewers showing somewhat higher avoidance. In contrast to the overall average of 7% commercial avoidance over all commercials on all encoded channels, high-churn content can produce three times as much avoidance. Given the variation in commercial avoidance, it comes as no surprise that there is also variation in the relationship of exact, commercial-minute audience levels to average-minute audience. High-churn formats produce lower indices with even lower levels for men than women. In other words, there is the potential for bias against particular media formats, program contents, and particular targets in today's reliance on currency proxy measures, average minute and AQH, to predict commercial audiences. PPM and Apollo estimates would help identify the bias potential and offer alternatives going forward. REFERENCES
Bosarge, John and Dupree, Linda. (2004). Media on the Move: How to Measure In- and Out-of-Home Media Consumption. Consumer Insight, ACNielsen, Winter. Campbell, Mike. (2005). Is ROI Dead? Admap, March. Ephron, Irwin. (2005). The Ephron Letter, January. Glock, Bernhardt. (2004). Keynote Address to ESOMAR/ARF. Geneva, June 18. Knowledge Networks. (2004). The People Look at Television, December. McConochie, R., Uyenco, B., Wood, L. and Heider, C. (2004). Toward Accountability. Proceedings of the ARF Conference, November. McConochie, Roberta. (2005). The Difficult Balance of Media Measurement. Admap, April. Papazian, E. (2004). TV Dimensions. Media Dynamics, New York. � Copyright by ESOMAR� / The ARF 26 Roberta M. McConochie, Leslie Wood, Beth Uyenco, Chris Heider
Pellegrini, Pasquale and Purdye, Ken. (2004). Passive vs. Button Pushing. Proceedings of the ESOMAR/ARF WAM Conference, June. THE AUTHORS
Roberta M. McConochie is Director, Portable People Meter Client Relations, Arbitron Inc., United States. Leslie Wood is President, Leslie Wood Research, United States. Beth Uyenco is US Director of Strategic Resarch and Analysis, OMD, United States. Chris Heider is Senior Research Analyst, Arbitron Inc., United States. � Copyright by ESOMAR� / The ARF