Glued to the Box?: Patterns of TV Repeat-Viewing

Glued to the Box?: Patterns of TV Repeat-Viewing by T. P. Barwise, A. S. C. Ehrenberg, and G. J. Goodhardt Only about half of those viewing a program one day view it again on any other given day, but this seemingly low level of repeat-viewing still reflects a marked degree of program loyalty. The extent to which people watch successive episodes of regular programs on U.S. television has rarely been studied. There are two main reasons for this. First, the level of repeat-viewing of programs with steady ratings is often assumed to be high (90 percent or more) and hence not worthy of detailed examination. Second, no routine data are available on the same people s viewing week-by-week-that is, there is no regular data base for examining individuals repeat-viewing of the once-a-week programs that are the main fare of prime time. Earlier studies in the United Kingdom found the level of repeatviewing to be about 55 percent (14). A comparable level was found in a U.S. pilot study for selected daily network programs (14, chapter 7). This article reports on a series of new findings in the United States, replicating and extending the earlier findings. The data analyzed here are mainly computer tapes of individual viewing records from the regular surveys of television viewing carried out by Arbitron (1). Viewing in sample households is measured for all individuals aged two and over, using a one-week diary. Each survey is T. P. Barwise is Senior Research Officer at the London Business School, where A. S. C. Ehrenberg is Professor of Marketing. G. J. Goodhardt is Cohen Professor of Consumer Studies, City University, London. The analyses reported here are based on data tapes from Arbitron Surveys of individual television viewing behavior. This article is based on one of a series of studies of U.S. television and radio at the London Business School supported by the Markle Foundation. 22

Patterns of TV Repeat-Viewing t conducted over a four-week period. The main analyses cover almost 18,000 individuals in New York City (January 1974 and October 1976), Los Angeles (October 1976), and San Francisco (February 1979). These data permit the study of repeat-viewing for programs that are stripped, i.e., shown five days a week. About half of those who view a daily (stripped) network program on one day will view it the following day. On weekdays in October 1976, about 2.9 percent of adults in the New York City area saw the game show Tattletales on WCBS at 1:00 p.m. each day. The rating showed little change from day to day, varying between a low of 2.7 percent and a high of 3.1 percent. These steady ratings do not mean, however, that the same three percent or so of people watched each day. For any two days, just over half of those who saw Tattletales on one day also saw it on the other. About 29 out of every 1,000 viewers saw Tattletales on either day of any given two consecutive days. But they were not the same 29 viewers, since some saw the program the first day but not the second, and vice versa. Of the 29 viewers on Day 1, about 16 also saw the show on Day 2, a repeatviewing rate of 55 percent. This is slightly above average for stripped daily programs on the networks. The 13 viewers who failed to repeat-view on Day 2 would not have been lost forever. They would include a high proportion of rather infrequent viewers of the program, whiie the 16 who watched on both 23

Journal of Communication, Autumn 1982 Table 1: Day-by-day repeat-viewing by viewers over age 18 of WNBC, New York City, October 1976 (averages over all pairs of weekdays) 1O:OO a.m. 11:OO a.m. 12:30 p.m. 1:30 p.m. 2:30 p.m. 3:OO p.m. 4:OO pm. 5:OO p.m. 6:OO p.m. 7:OO p.m. Sanford and Son Wheel of Fortune Gong Show Days of Our Lives The Doctors Another World Marcus Welby Newscenter 4 at 5 Newscenter 4 at 6 NBC Nightly News % of repeat-viewing Average 53 47 54 50 55 53 65 41 45 52 66 days would be fairly regular viewers of the program. But even they would not always watch: less than half would see all five episodes in one week, and hardly any would see all 10 episodes over a two-week period. Table 1 shows the repeat-viewing rates for all daily programs on WNBC in New York City in October 1976. The average rate is 53 percent, and the figures for individual programs range from 41 to 66 percent. The highest repeat-viewing level is for the NBC network news. Such a high value does not recur for the NBC news in the other cities or years that have been checked, nor for the corresponding early evening news on the other two networks. Indeed, with one exception, neither the news nor any other program type differs systematically in its day-by-day repeat-viewing levels-almost all program types average about 50 percent. The exception is daily soap operas, which average fairly consistently about 10 percentage points higher (60-65 percent repeat-viewing). Many observers have noted the audience s apparent loyalty to daytime soap operas (e.g., 5, 9, 15) and the importance of a continuing story in generating this loyalty (20,22,23). But despite such loyalty and continuity, almost 40 percent of the audience on any given day did not see the previous episode. The figures in Table 1 represent the repeat-viewing levels for each program across all possible pairs of weekdays. The repeat-viewing level varies slightly by how far the two days are apart: for consecutive days, Of those who saw Tattletales at all in a week, 47 percent saw it only once, 17 percent twice, 13 percent three times, 14 percent four times, and only 9 percent on all five days. This type of frequency distribution can be successfully modeled using a Beta- Binomial Distribution (14, 18). 24

Patterns of TV Repeat-Viewing Wednesday to Thursday for example, it is generally a few points higher than for Wednesday to Tuesday (six days later). But the effect is small and the figures in Table 1 represent the general level of repeat-viewing of daily network programs to within a few percentage points. Repeat-viewing of daily programs on independent stations, which may be shown both in the afternoon and during prime time, is generally lower, about 30 percent. Only about one in three of those seeing a stripped program on an independent station one day also see it the next day. This may be because some of these programs have been frequently rerun. The average repeat-viewing levels for daily programs of about 50 percent for networks and about 30 percent for independents are consistent in different years, different locations, and for different adult subgroups (men, women, blacks). Teenagers and children, however, show lower repeat-viewing levels than adults: roughly 30 percent for independent stations and if anything slightly lower for the networks, which are less popular with children. Heavy viewers of television are somewhat more regular in their repeat-viewing of programs day by day-about 60 percent for adults who averaged 35 or more hours of viewing per week. Conversely, light viewers are less regular-only about 25 percent repeat-viewing for those who watched less than 20 hours a week. Light viewers therefore do not seem to be especially selective : i.e., they generally do not pick out only a few programs which they then watch very regularly. The repeat levels here do not particularly vary with the rating levels of the different programs. This may be due to their being stripped programs and to their relatively low rating levels. There is certainly no evidence here that low-rated programs attract dedicated minority viewers who watch them with above-normal regularity. We can examine week-by-week repeat-viewing for prime-time programs using the special Los Angeles data. As shown in Table 2, repeatviewing of first showings of six typical prime-time programs was just Table 2: Week-by-week repeat-viewing by viewers over age 18 of six representative prime-time weekly network programs, Los Angeles, 1976 % of repeat-viewing Rhoda (KNXT, Monday, 8 p.rn.) 64 Muppets (KNXT, Saturday, 7 p.m.) 40 Price is Right (KNBC, Thursday, 7:30 p.rn.) 21 NFL Football (KABC, Monday, 7 pm.) 49 Baretta (KABC, Wednesday, 9 p.rn.) 54 Donny & Marie (KNBC, Friday, 8 p.m.) 57 Average 48

Journal of Communication, Autumn 1982 under 50 percent. (The average repeat-viewing rate for all regular weekly programs with steady ratings was 46 percent.) While the week-by-week results illustrated in Table 2 are based on a single experimental study, they are strongly supported by the daily repeat-viewing results and by our finding that in the United Kingdom repeat-viewing of weekly prime-time programs is about 50 to 55 percent (14). If only about half of those who see a program on one day view it again the next day, what ia the other half doing? Only about six to eight percent of those who view a program one day view a different program in the same time slot the next day. For WNBC, in 1976, for example, for every hundred viewers of the average episode of a stripped program, on another day at the same time 53 repeat-viewed the program, 6 watched another channel, and 41 were not watching television at all. The 50 percent overall level of repeat-viewing of network programs discussed here therefore arises from the combination of two effects. First, only about 60 percent of those who watch one episode of a program are viewing TV when the second episode is shown. Second, of those viewing at all, the great majority-about 90 percent-do in fact repeatview the same program. In general, the repeat-viewing level for total TV is about 60 percent. It varies only slightly for different times of day and holds within weeks and between weeks. Some specific variations are: 1. The total Tv repeat-viewing level is slightly higher during prime time (60-65 percent) than at other times (55-60 percent). But during the day it sometimes rises to about 65 percent, when soap operas are being shown. 2. During the daytime, but not noticeably for prime time, it is a few percentage points higher for consecutive than for non-consecutive days. 3. It is lower for lighter than for heavier viewers-30 percent for those watching fewer than 20 hours per week. But even for quite heavy viewers (35 hours a week or more) it does not exceed 70 percent. 4. It is lower for children and teenagers (about 40 percent) than for adults (60 percent). These minor variations do not affect the general conclusion that most people do not watch TV regularly at the same time of day every day. An important factor here seems to be audience availability. Evidence from the United Kingdom, where total TV repeat-viewing levels are 26

Patterns of TV Repeat-Viewing comparable to those in the U.S. (3), suggests that about half of those not watching TV at all at any given time are simply not available, i.e., they are either out or are asleep (2). Although the level of repeat-viewing of specific programs may seem low, there are two reasons to interpret it as nonetheless reflecting a marked degree of program loyalty. First, most of those who are not watching a program the next time are not watching TV at all. As discussed, for daily programs on WNBC, about 90 percent (an average of 53 out of 59) of those watching TV at all were repeat-viewing the same program. In other cases, such as for weekly prime-time programs, the proportion is somewhat lower than 90 percent. Nonetheless, viewers of a program are generally very much more likely to watch the same program again than any specific competitive program at that time. Second, the level of repeat-viewing is much higher than can be accounted for by any kind of channel loyalty effect (14). Thus, in the Los Angeles study, the overlap between a network s audience during the same prime-time slot in different weeks averaged almost 50 percent for different episodes of the same program (as illustrated in Table 2), but it was only 20 percent when the networks were showing different programs in the two weeks. These results throw some light on the wider issue of how people watch television. They are consistent with Comstock s view (10, p. 38) that typically, television is consumed as a medium and the decision to view ordinarily takes precedence over the selection of what to view. This certainly seems to be the case for most prime-time viewing and for people watching daytime programs on independent channels. It seems to be less clearly so for the viewing of daytime programs-especially soap operas-on the networks. But even in the case of daytime soap operas, over one-third of any day s viewers did not see the previous day s episode because they were doing something other than watching television. The first stage of Comstock s (10, 11) postulated two-stage decision process-the decision whether to watch-seems to be largely passive. People mostly seem to watch television when they have nothing else to do or perhaps nothing better they can be bothered to do. The size of the total TV audience usually seems to depend much more on audience availability than on what programs are on. But the second stage of the viewing decision-the decision what to watch-does not seem to be quite so passive. Viewers do not just pick a program at random. As the results described here show, they exhibit 27

Journal of Communication, Autumn 1982 marked program loyalty: if they are watching at all, they will usually (over 70 percent of the time during prime time, 90 percent of the time during daytime) watch the same program they watched last time. In addition to the obvious suggestion that they repeat-view because they like this program more than the alternatives, other studies have shown that there is a marked relationship between how much people say they like a program and how often they say they watch it (6, 12)-partly because viewers may learn to like a program over time. But it would be wrong to conclude that program choice is a very active process. One does not have to go as far as Paul Klein s least objectionable programming theory (16, reprinted in 9) to admit that the effort viewers are prepared to invest in their viewing is usually minimal, reflecting rather low involvement (17). Thus, they tend to avoid programs that are intellectually or emotionally demanding (7, 25); they sometimes do not bother to switch channels between programs (4, 14); and even the program loyalty evidenced here partly reflects the fact that it takes less effort to watch a familiar than an unfamiliar program. People now watch television for an average of over 25 hours per week (19). Does this mean they are glued to the box? The results here suggest not. Rather, the level of repeat-viewing of television-about 60 percent-suggests that television fills the gaps between other activities. The reason people watch so much is that they have so much time to kill. This interpretation is supported by other evidence. While people say they enjoy watching television, they tend to say they enjoy almost all other free-time activities more (21). They also are less likely to say they enjoy television than people were 10 or 20 years ago (8, 24). The 50 percent level of repeat-viewing described here may intuitively seem lower than expected. There are two reasons-apart from the impression given by the steady ratings-why it might subjectively appear that people s actual viewing is more regular than this. First, people probably tend to count only their favorite programs when they reflect on the regularity of their TV viewing. Perhaps they do not even think of the programs they watch occasionally and in which their involwement is minimal. Second, people may mentally exclude those occasions when they were not available to view: they think of how regularly they watch when they are at home and doing nothing else. Counting only these occasions, repeat-viewing is much higher, as noted earlier, reflecting people s loyalty to particular programs. (This would explain why people tend to over-estimate their frequency of viewing a program when they are explicitly asked about this [6, 12, 131.) But that loyalty typically does not motivate great efforts to ensure that they never miss an episode. 28

Patterns of TV Repeat-Viewing REFERENCES 1. Arbitron Company. Description of Methodology. Laurel, Md., 1981. 2. Aske Research Ltd. Availability to View. Report prepared for the Independent Broadcasting Authority, London, 1978. 3. Aske Research Ltd. The Regularity of Viewing Any Teleuision. Report prepared for the Independent Broadcasting Authority, London, 1978. 4. Aske Research Ltd. The Effort of Switching Channels. Report prepared for the Independent Broadcasting Authority, London, 1980. 5. Barthel, Joan. The World Has Turned More than 3200 Times, and 8 Million People Keep Watching. New York Times Magazine, September 8, 1968, Section IV, p. 66 et passim. 6. Banvise, T. P. and A. S. C. Ehrenberg. The Liking and Viewing of Regular TV Programs: Cincinnati Pilot Study. Report prepared for the Markle Foundation, New York, 1982. 7. Barwise, T. P., A. S. C. Ehrenberg, and G. J. Goodhardt. Audience Appreciation and Audience Size. Journal of the Market Research Society 21(4), 1979, pp. 269-289. 8. Bower, R. T. Television and the Public. New York: Holt, Rinehart & Winston, 1973. 9. Cole, Barry (Ed.) Television Today: A Close-up View. Oxford and New York: Oxford University Press, 1981. 10. Comstock, George. Television in America. Beverly Hills, Cal.: Sage, 1980. 11. Comstock, George, Steven Chaffee, Natan Katzman, Maxwell McCombs, and Donald Roberts. Television and Human Behavior. New York: Columbia University Press, 1978. 12. Ehrenberg, A. S. C. and G. J. Goodhardt. Attitudes to Episodes and Programmes. Journal of the Market Research Society 23(4), 1981, pp. 189-208. 13. Frank, Ronald and Marshall Greenberg. The Public s Use of Teleuision: Who Wutches and Why. Beverly Hills, Cal.: Sage, 1980. 14. Goodhardt, G. J., A. S. C. Ehrenberg, and M. A. Collins. The Television Audience: Patterns of Viewing. Farnborough, England: Cower Press; Lexington, Mass.: Lexington Books, 1980. 15. Headen, Robert S., Jay E. Klompmaker, and Roland T. Rust. The Duplication of Viewing Law and Television Media Schedule Evaluation. Journul of Marketing Research 16, August 1979, pp. 333340. 16. Klein, Paul L. Why You Watch What You Watch When You Watch. TV Guide, July 24, 1971. 17. Krugman, Herbert E. Point of View: Sustained Viewing of Television. Journal qf Advertising Research 20(3), June 1980, pp. 65-68. 18. Metheringham, Richard A. Measuring the Net Cumulative Coverage of a Print Campaign. Journal of Advertising Research 4(4), December 1964, pp. 23-28. 19. Nielsen, A. C. Television Audience 1981. New York: A. C. Nielsen Company, 1981. 20. Pritchard, William H. One Man s Soap. Channels, April-May 1981, pp. 83-86. 21. Robinson, John P. Television and Leisure Time: A New Scenario. Journal of Communication 31(1), Winter 1981, pp. 120-130. 22. Rose, B. Thickening the Plot. Journal of Communication 29(4), Autumn 1979, pp. 81-84. 23. Stedman, R. W. A History ofthe Broadcasting of Daytime Serial Dramas in the United States. Unpublished Ph.D. dissertation, University of Southern California, 1959. 24. Steiner, Gary. The People Look ut Television. New York: Knopf, 1963. 25. Stewart, W. Brian. The Audience Size of TV Programs. Unpublished report, 1981. 29