How Promotional Content Changes Ratings: The Impact of Appeals, Humor, and Presentation
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Journal of Applied Communication Research Vol. 31, No. 3, August 2003, pp. 238–259 How Promotional Content Changes Ratings: The Impact of Appeals, Humor, and Presentation Susan Tyler Eastman, Gregory D. Newton, and Paul D. Bolls ABSTRACT This study assessed the impact of content—as opposed to structural—fac- tors on television program ratings, seeking to locate clusters of components that would identify effective on-air promotion and allocate content a better-defined place within theoretic models of media priming. Stepwise multiple regression analyses of 1,547 on-air promos for 155 prime-time programs demonstrated that 5–9% of ratings variance was accounted for by content appeals, humor, and presentation in promos for comedy programs. The influence of content variables was greater for familiar than unfamiliar comedies, and humor and presentation in promos contributed to variance in ratings for mid-rated but not high- or low-rated comedies. KEY WORDS: Comedy promotion, content appeals, humor, priming, program pro- motion, ratings O n-air promotion’s importance to prime-time ratings has been beyond ques- tion for the last decade. Estimates are that the major broadcast networks collectively air as many as 30,000 promos yearly (Ferguson, 2002), using up irreplaceable air time. Touting prime-time programs has become a big-budget item for the television industry, occupying more than $4 billion in air time that otherwise could be sold for commercials and that necessitates high in-house production costs (Eastman & Newton, 1998b). Competition from cable networks and the internet, hugely increased license fees for series programs, and prolifer- ating user technologies such as remote control devices, personal video recorders, and video-on-demand have made effective on-air promotion crucial to network programming strategy. Nonetheless, scholarly research in this arena has been limited despite decades of industry practice—leaving questions about how and why on-air promotion impacts program ratings. Only a few programs, such as the Super Bowls, Academy Awards, and Susan Tyler Eastman is Professor in the Department of Telecommunications at Indiana University in Bloomington, Indiana; Gregory D. Newton is Assistant Professor in the School of Telecommunications at Ohio University in Athens, Ohio; Paul D. Bolls is Assistant Professor in the Edward R. Murrow School of Communication at Washington State University in Pullman, Washington. Direct correspondence to the first author at the Department of Telecommunications, Indiana University, 1229 E. Seventh Street, Bloomington, IN 47405-5501 or eastman@indiana.edu. Copyright 2003, National Communication Association DOI: 10.1080/0090988032000103458
239 JACR AUGUST 2003 Olympics, are considered self-promoting, that is, able to attract a mass audience without help from on-air spots (promos) or newspaper and program-guide advertising.1 Most of the prime-time schedule depends on a steady stream of promotional announcements to elicit audience interest in watching shows as well as to convey reminders of the day and time they appear. Capturing an audience for a new series program is almost wholly dependent on having both a strong lead-in and extended on-air promotion, but continuing programs must also promote upcoming episodes, guests, and plot twists to maintain audiences (Eastman, 2000). Investigating on-air promotion has importance for two reasons. First, promos can be viewed as a form of high-stakes advertising. On-air promos are, in essence, advertisements for a television network’s own product, entertainment programs. As with advertisements, an effective on-air promo will present the product in a way that will make its target audience want to sample it. The degree to which on-air promos are successful at advertising programs is presumed to affect the programs’ ratings which, in turn, substantially impact the amount of money the network can charge for advertising time during those programs. Second, promos are among the factors that influence viewers’ evaluation of program content; they affect viewer decision making. Promos entice viewers to sample unfamiliar shows and new episodes of familiar programs. Moreover, list-style promos and combined spots promoting multiple programs supply a memory framework for understanding an evening’s program schedule and the sequence of the individual programs within it which, in turn, influences tuning behavior. Clearly, such factors as previous experiences with a program (or genre or even a particular network) and a viewer’s intended viewing situation (with family or friends, for example) may also impact viewers’ reactions to specific promos. Nonetheless, previous studies—using thousands of promos and pro- grams over long periods of time—have demonstrated that promotion appears to have a modest impact on viewing above and beyond those factors (Billings, Eastman, & Newton, 1998; Eastman & Newton, 1998b, 1999; Eastman, Newton, & Pack, 1996: Walker, 1993). In on-air promotion, sets of appeals frame the value to viewers of watching programs, often called the benefits in industry jargon (Eastman, 2002). When a program is promoted as “very touching,” for example, the benefit is that the viewer gets to feel the tender emotion of poignancy and experience the better side of human nature. Similarly, when a program is promoted as suspenseful, the benefit is the enjoyable experience of being scared—within the wholly safe context of a television program. Such appeals are equivalent to what researchers Vakratsas and Ambler (1999) identify as affect in commercial messages. Thus, emotional appeals are shorthand for a program’s content, or at least for that portion of the content picked up for inclusion in promos. Although content appeals are related to the idea of program quality, the former are identifiable and measurable, whereas quality (or whatever makes programs popular) generally is not. In practice, the appeals appearing in promos are selected by creative producers on a largely intuitive basis from the unwieldy mass of program content, often under pressure from tight deadlines (Ferguson, 2002). Isolating appeals that significantly affect ratings should contribute to theoretical and practical models of the process of building audience size. Viewer decision making is, after all, the central concern in programming research.
240 PROMOTIONAL CONTENT EASTMAN ET AL. As previous research has focused almost exclusively on structural factors in promotion affecting program ratings (such as location, associated clutter, length, and so on), this study distinguishes between content and structural components and separates their relative influences. The purpose of this study is to look for the factors in promos having measurable impact on comedy viewing, as deter- mined by changes in Nielsen ratings. Another goal of this line of research is to locate clusters of components that suggest empirically-derived guidelines for effective promotion. Applicable Theories The industry invests billions of dollars annually—even in tight economic times—in promoting television programs on the presumption that audience size will be affected. At the same time, explaining promotion’s impact on a theoreti- cal level necessitates accounting for three characteristics of promotional spots: (a) entertainment promos typically beget very low level, mundane reactions rather than highly arousing experiences; (b) promos’ impacts on ratings are generally modest in size; and (c) promos’ impacts commonly occur in spite of intervening activities and at substantial distances in time from the programs being promoted, thus any effects occur across hours, days, and weeks. The size of any collective impact from a large body of promos must necessarily be small because of the nearly overwhelming impact of inherited viewing (lead-in ratings) and program popularity (carriage and promoted program ratings), and the ines- capable fact that most new programs have such low ratings that they will be canceled within a year. Moreover, most ratings are understood to contain significant amounts of sampling and data collection error. Nonetheless, decades of industry assumptions about the critical impact of promos have led to the widespread and costly practice of extensive promotion of television shows in advance of their airdates. One problematic issue for research has been identifying the parts of promo- tional messages that contribute to their effects. A growing body of research on television advertising addresses message effects from a cognitive and emotional perspective (Hazlett & Hazlett, 1999; Lang, Bolls, Potter, & Kawahara, 1999; Nelson, Shavitt, Schennum, & Barkmeier, 1997; Rossiter & Silberstein, 2001; Yoon, Bolls, & Muehling, 1999). If promos are similar to ads, then cognitive and emotional processes probably act as the mechanisms underlying the effects of promotional messages. Two likely candidates for providing a theoretical link between parts of promotional messages and their cognitive and emotional effects on viewers are excitation transfer and associative priming effects. From a theoretical viewpoint, excitation transfer theory holds that content able to stir viewers’ emotions may lend those emotions to subsequent viewing (Mundorf, Zillmann, & Drew, 1991; Zillmann & Weaver, 1999). Indeed, research has produced evidence of emotional effects transferring from programs to com- mercials as well as from commercials to adjacent programs. Mattes and Cantor (1982) showed that watching highly arousing programs led to higher levels of enjoyment of commercials, and Mathur and Chadpadhyay (1991) showed that program-induced moods transferred to reactions to commercials. Looking for humor’s transference effects, Perry, Jenzowski, King, Hester, and Yi (1997) found that the presence of more humorous commercials in a show increased viewer
241 JACR AUGUST 2003 enjoyment and the program’s perceived entertainment value, but found no influence in the opposite direction. In other words, humor in programs was found to have no influence on commercial enjoyment (Bjorna, Karsal, Vicary, Wagner, & Perry, 2001). Nonetheless, a form of excitation transfer effect has been proposed as a possible explanation for how advertising works on a target audience. Associa- tional learning in consumer behavior occurs when a favorable response evoked by one stimulus is directed or transferred to another (Mullen & Johnson, 1990). Advertising that is successful at achieving associational learning evokes positive emotions and thoughts that get transferred to the product itself. In the context of on-air promotion, this means that emotions evoked during exposure to a promo could be transferred to the program and have some influence over viewing decisions. It is important to note that proposing excitation transfer as a possible cognitive and emotional mechanism behind the effects of on-air promos requires a major extension of the original theory. In studies of television content, excitation transfer has only been demonstrated for very short lengths of time, and in laboratory experiments arousal levels return to normal in seconds. This time- frame excludes promos’ apparent impact over much longer time periods. Apply- ing excitation transfer to the effects of promos would require the emotions evoked by a promo either to (a) maintain some level of activation in working memory until a viewing decision is made, which could be several days after exposure to the promo, or (b) reside in long-term memory where emotions are associated with the program. Clearly, the low-level emotions evoked by promos are unlikely to remain intensely active over the course of days. Recent models of human cognition, however, suggest that the emotional characteristics of stimuli get stored in memory in a manner that enables the activation of the same emotions when a cue related to the original stimulus is encountered again, even days later (Damasio, 1994). This means that emotions evoked by a promo could maintain some level of background activation, and be transferred to the pro- moted program when a viewer is cued to think about the program. Zillmann (1971) has defined the excitation transfer effect as a matter of incomplete decay of arousal, a definition that excludes promos’ apparent impact over much longer time periods. It seems, nonetheless, that the association of emotions with a program during exposure to a promo may be explained as a re-experiencing that occurs when a viewer is cued to think about a program. Thus, the process could constitute an incomplete decay of arousal that triggers another excited state that is then transferred to the program. We have named this the secondary excitation transfer effect. While this notion goes beyond tra- ditional views of the excitation transfer effect, it seems premature to rule out a multistep excitation transfer process as the cognitive and emotional mechanism behind the effects of promos until further experimental research can be conduc- ted. Associative priming theory also provides part of the conceptual explanation of the promotional process. Research on associative priming, as Jo and Berkowitz (1994, p. 46) have pointed out, shows that “ideas having emotional significance are linked associatively” to subsequent behavior. They also note that thinking about a behavior increases the likelihood of carrying out that behavior, and that visual stimuli such as video generate relatively high recall. Roskos-Ewoldsen,
242 PROMOTIONAL CONTENT EASTMAN ET AL. Roskos-Ewoldsen, and Carpentier (2002) summarize research showing that me- dia influence later behavior by priming aggression, stereotyping, and political judgments, and thus, it follows that an arousing television promo that urges viewers to tune in at a later time to experience a program might well have an impact on some viewers. While the content may be important because of its arousing nature, however, any impact on viewing occurs long after particular jokes and plot moments are forgotten. It may be that secondary arousal transfers to behavior when the proper occasion (another promo or perhaps merely the right date and time) occurs. Nonetheless, the explanations falter inasmuch as situation comedy promos typically generate only the most modest of emotional reactions of any kind, yet they apparently provoke subsequent viewing (by some people some of the time). Roskos-Ewoldsen et al. hold that “people have myriad mental models stored in long-term memory” (2002, p. 112) and that the media can prime these memories, meaning that media content will make these memor- ies readily accessible. Extending this mental model view of priming into promo- tional priming leads to the proposition that promos evoke viewers’ mental models of television programs or, conversely, that the beginning of a program (or its title, music, teaser) evokes previously viewed promos that are stored in long-term memory. The model of promotional salience developed by Eastman and Newton (1998b, 1999) adds another level of explanation to this discussion. Salience theory proposes that on-air promos with specific attributes—those that make promos prominent in their environments—are more likely to have an impact on viewing than promos with fewer of those attributes (or that have the attributes but in a non-salient condition). The idea is that certain conditions of variables—first or last position in breaks, within-program location, little clutter in breaks, and a dozen others—are structural factors that enhance attention to promos and conse- quently generate their impact on program ratings. By extension, salience theory suggests that aspects of content, such as the appeals, humor, and presentation (execution in industry jargon) that are the focus of this study, can effectively create expectations about programs under some conditions and, therefore, have an impact even without high levels of emotional arousal. Thus, associative priming and secondary transference may together account for the relationship between the promo and the program, while salience theory may identify the conditions of attributes creating the expectations. A first step in testing priming along with secondary transference as possible cognitive and emotional mechanisms behind promotional effects is to find out whether a relationship between promos’ content components and program ratings can be isolated. However, programming scholars already know that the single biggest component of program ratings is viewing inherited from the preceding program (Webster & Phalen, 1997).2 This study looks only at situation comedies in order to include the component of humor in promos, but when aired in prime time, situation comedies are nearly always blocked in groups for the first hour or two. The consequence is that inheritance should be particularly strong from comedy to comedy, irrespective of the amount and type of pro- motion received by the individual programs. While inheritance and other struc- tural characteristics of promos have previously been shown to affect program ratings (see Eastman & Bolls, 2000 for a summary), this study provides the first
243 JACR AUGUST 2003 direct test of the assumption that certain aspects of content also impact audience size. H1: In addition to structural components, promos’ content components will significantly affect the ratings of promoted comedies. Turning to the factor of program popularity as measured by Nielsen ratings, Eastman and Newton (1998a, 1998b) showed that promotion’s impact was greatest on the mid-rated shows, probably because the top-rated hits were already about as high in audience size as they could go, and the bottom-rated losers would be promoted less and less until they were canceled. The Eastman and Newton findings led to the practice of assessing promotion’s impact on the two mid quartiles (that is, the middle half) of programs as subdivided by their ratings. The consistency of previous findings for structural variables suggests a second hypothesis. H2: Content-related promotional variables will account for more variance in ratings for mid-rated comedies than for high-rated or low-rated comedies. Practitioners concerned with the effects of advertising have been asking how ads work for decades. Based on a meta-analysis of advertising studies, Vakratsas and Ambler (1999) proposed a three-dimensional model of advertising effective- ness with components of cognition, affect, and experience. They concluded that the cognitive aspects of ads were more important than their affective dimensions for high-involvement products, but that the affective aspects were more import- ant for low-involvement products; they also found that experience mattered most for mature, familiar products. Although their model was derived from studies of ads for consumer goods, it can be adapted to program promotion: If promos are indeed low-involvement messages, then they are most likely to be most effective when they emphasize affect. Nevertheless, structural factors must be accounted for, leading to this hypothesis: H3: After the structural variables of lead-in ratings and carriage program ratings, appeals in comedy promos will contribute more to ratings variance than presentation or humor. Walker (1993), however, showed that the familiarity of the program was key to effective promotion, and that more frequent promotion of continuing (but not new) programs was positively related to their ratings. Taking Vakratsas and Ambler’s (1999) and Walker’s conclusions a step further, when promos are for familiar programs, affect will be the crucial component, whereas in promos for unfamiliar programs, structural variables will dominate and, consequently, con- tent variables should have relatively little impact. The question then shifts back to the sources of affect in program promos. It may be that the appeals in on-air promos, especially those alleging benefits for viewers, will most impact viewing of familiar programs. Paraphrased, these extensions lead to this hypothesis: H4: The ratings for familiar comedy programs will be more affected than the ratings for unfamiliar comedies by the affective appeals contained within the promos.
244 PROMOTIONAL CONTENT EASTMAN ET AL. Prior Research About Humor Another factor, related to appeals and likely to be crucial when evaluating promos for situation comedies, is comedy or humor—a widely-used indicator of arousal. A reasonable presumption in most studies was that more humor implies greater likely impact. According to Speck (1990), scholars have identified several basic processes in humor, three of which apply to television commercials. The first of these, arousal-safety (or Ahhh humor), is a mechanism that allows relief from strain or from the need to repress feelings. Laughter usually occurs when a person experiences increased arousal but evaluates the stimulus (typically another person) as safe, cute, or inconsequential, as in much family humor. The second type of humor, identified as incongruity-resolution (or Ah-Ha humor), occurs when two ideas cannot be assimilated using the same schema or, in other words, when the outcome of a story or event doesn’t fit with the audience’s expectations, as occurs in puns, punch lines, comic reversals, understatement, and exaggeration. The third type of humor Speck calls humorous disparagement (or Ha-Ha humor), referring to censure, detraction, denunciation, or in present- day slang, dissing. Much hostile or aggressive humor falls into this category. Because writers and producers generally want the audience to feel good after jokes or laugh at put-downs, they avoid getting mean or ugly in promos, and because clips are very rarely short (usually excerpted from the program) and directed at broad audiences, promos are very subtle or witty. All three types of humor (and perhaps several others) probably appear within comedy programs, but only certain types will be selected by the producers or programming and promotion executives to appear within promos. In preparation for this study, a preliminary survey of adult reactions to sitcom promos (Eastman & Bolls, 1999) showed that virtually all the humor in a large promo sample was identified by viewers as one of these three types. Because measuring the quantitative amount of humor proved unreliable in pilot studies, this analysis focuses on the type of humor and investigates the relationship between type and impact. An additional factor is the presentation of the promo. Eastman and Bolls (2000) analyzed open-ended responses from nearly 2,000 adult television view- ers about why they might or might not watch a new program based solely on viewing a promo. Analysis showed that more than one-third of comments related to the promos (37%), rather than to the actors, storylines, or networks, and focused on humor and presentation (music, effects) within the promo. The advertising literature contains many studies of execution that look at the way commercials (and by extension promos, if considered as advertisements for programs) are implemented. Some studies have examined how the viewer feels about the presentation of the product being pitched. Lang (1990), for example, has demonstrated that physiological responses to structural features of commer- cials are intensified by mild emotional content. Two feelings can be dis- tinguished: (a) how the viewer feels about the product based on purchase, use, price, quality, and so on, and (b) how the viewer feels the product is being presented in a commercial (Laskey, Fox, & Crask, 1994; Stewart & Furse, 1986). Although viewers’ feelings about television shows are part of the equation leading to ratings, this study focuses on the second meaning of presentation (or execution) to assess that characteristic of on-air promos. Lang, Geiger, Strick- werda, and Sumner (1993) and Yoon, Bolls, and Lang (1998) examined aspects of
245 JACR AUGUST 2003 presentation in advertising research that include pacing in terms of length and number of editing cuts, narrative structure, recognition of background scenes, and the degree to which the message informs users about the things they want to know. When their findings are applied to promos, the use of quick cuts or relatively slow-paced camera changes (apparent pacing) might be expected to transfer to viewers’ expectations about promoted programs and thus impact ratings to some small degree. As Eastman (2002) points out, all measurable impacts are likely to be small but significant because tiny shifts in ratings affect millions of dollars in annual advertising revenue. The focus of this study was not the appeals, humor, and presentation in the programs themselves. Instead, it was expected that within 5-, 10-, 15- and 30-second promos the inclusion of certain kinds of appeals or representations of certain kinds of humor might have close association with certain types of presentation and thus, as a group, affect ratings (in addition to the well-recog- nized elements of inherited viewing and the structural variables in promotion identified in previous studies). If such combinations could be identified, they would further refine the current models of advertising and promotional effective- ness and suggest empirically-based guides for industry practitioners. Because so little research has been published about the role of content in on-air promotion, in addition to the four hypotheses already identified, this study posed one broad research question. It serves as a first step towards forming a model to guide practitioners and subsequent research into promotion: RQ1: Can combinations of certain kinds of presentation, specific appeals, and types of humor be identified within promos that associate with significant increases in comedy program ratings? Method Because prime time is the most watched daypart and because half-hour programs maximized the number of programs in the database, this study looked only at prime-time promotion. It analyzed only on-air promos for half-hour situation comedies (live or animated); humor was included and genre-related variation excluded. The prime-time hours on all six broadcast networks were videotaped by the researchers for six weeks: two weeks in October 1998, two weeks in May 1998, and two weeks in April/May 1999.3 These purposive samples encompass the highly-produced promos for new program introductions (October), where many programs would be unfamiliar, big-budget cliffhangers comprising the culminating episodes of many series (April), and the end-of-sea- son program specials (May); they also avoid the extreme fluctuations in audience size occurring between winter and summer and reflect a mix of sweeps and nonsweeps months. Promotional messages of less than three seconds in length were excluded, and no promos longer than 30 seconds were reported. Coders In addition to the three researchers, three classes of student coders worked on this project. One class pretested the coding sheet in two iterations, resulting in the rewording of confusing or redundant appeals, reducing the number of
246 PROMOTIONAL CONTENT EASTMAN ET AL. options for narrative structure, and eliminating measures of the quantity or degree of humor because reliabilities were low. Two subsequent classes of 22 and 31 students analyzed the videotaped promos, and all 218 tapes were coded at least twice, each time by different students. The first class analyzed the four weeks of 1998 tapes; the second class reanalyzed the four weeks of 1998 as well as the two weeks of 1999 tapes. The students undertook the coding for a grade as part of an upper-division seminar in electronic media promotion, and they were required to write a separately-graded paper about the research process. Two class periods were devoted to training each class of coders, focusing on distin- guishing promos for network sitcoms from local rerun sitcom spots, identifying appeals and types of humor, utilizing the semantic-differential scales, and counting shot changes. Stopwatches were supplied to measure promo length. In the training and written instructions, coders were repeatedly directed to give their attention to the promo and its content and presentation, and not to consider their reactions to whatever program was being promoted. Because they were enrolled on a course specifically about media promotion and marketing, not programming, focusing on the promos was a reasonable expectation. Coding Instrument The coding sheets requested nine items of structural and sorting data, includ- ing (a) identification of the type (or absence) of humor, (b) an assessment of the promoted program’s familiarity to the coder, (c) 14 ratings of specific appeals, and (d) four measures of presentation of the promo’s content. Following specification of the network, date and time, and name of the program containing the promo (the carriage program), coders were asked to code the name, date and time of the sitcom being promoted, the construction of the spot (operationalized as the number of named programs included in the promo), and the distance between the promo and its program (operationalized as the next show; later tonight; tomorrow night; later in the same week; next week—between seven and 14 days away; and much later—with space to supply a date if one was given). Coders classified the predominant type of humor used in the promo by choosing one of three options: relief/cute (followed by a definition and the words “arousal-safety or Ahhh!”), surprise/mismatch (definition and “unexpected in- congruity or Ah-Ha!”), and put-down/sarcasm (definition and “dissing and satire or Ha-Ha!”). Pretesting using promos not in the sample had demonstrated that the coders could consistently distinguish between types of humor and identify one dominant type even when more than one type appeared in 30-second spots. In a few cases, promos for sitcoms had no humor component, which was recorded. The coding instrument listed 14 appeals to rate on a nine-step semantic differential, anchored by not very or weak on the low end and very or strong on the high end. Following the heading question “How does the promo make the show seem to look?” were seven phrases anchored from negative to positive, as in not very funny to very funny, along with not very realistic, not very new, not very uplifting, not very hip/trendy, not very critically acclaimed, and not very suspenseful. Following the heading question “What does the promo make the show seem to have?” were seven more appeals: a not very puzzling situation (to a very puzzling situation, and so on), weak star appeal, weak sex
247 JACR AUGUST 2003 appeal, characters not easy to identify with, not very realistic characters, a not very intriguing situation, and no network popularity. Presentation was operationalized in four ways: as background scene, narrative structure, informational value, and the pacing of the promo. For the primary background scene, the options were bar/restaurant, workplace, school, home/ family, singles’ apartment, and other. Based on the results of pretesting, three options for the promo’s narrative structure were provided: clip, with or without a voice over (a short scene from the show with dialogue); announcement (a cast member speaking directly to the audience, sometimes accompanied by a brief voiceover); and other. The informational value of each promo was assessed with the question: “How informative was this promo?” Coders responded from not at all informative to very informative on a nine-step semantic differential. Pacing was assessed in two ways. Coders were asked to count the number of cuts (shot changes) in a promo (or in a segment dealing with one program during a spot promoting multiple shows) and to record the clock length in seconds of the promo (or individual segment). Pacing was then calculated as the number of cuts divided by length. Coders also made their own assessment of the pacing on a five-step semantic differential, ranging from very slow to very fast. Finally, for sorting purposes, a global measure of the coder’s familiarity with the promoted program was also obtained using a nine-step semantic differential anchored by the terms not very familiar and very familiar. Subsequent to coding, four additional kinds of information were added to the database by the researchers. The frequency of promotion of each program was calculated from the database, and the published ratings for each carriage and promoted program were added, in the latter case utilizing only the rating for the episode closest in time to the promo’s appearance. To account for the impact of inheritance, the rating for the lead-in program was also entered. When unavail- able because the sitcom was scheduled as a prime-time lead-off, the mean of all lead-ins in that two-week period was substituted for the missing data, thus maintaining sample size and utilizing all identified promos and situation come- dies.4 Ratings Analysis For the analyses dealing only with mid-rated comedies, researchers identified the range of high to low ratings separately for each network because the range of ratings varies even within the “Big Three” (ABC, CBS, NBC) and widely outside them (Fox, UPN, WB). One network’s ratings, for example, might have a high of 17.5 and low of 5.5, while another has a high of only 4.9 and a low of 1.9. The middle half of each network’s ratings was calculated independent of the other networks by computer subroutine. Coding Reliability The researchers looked for coding errors at five stages in the research process. During an initial examination of completed coding sheets, they discarded those with blank or off-scale responses, and scheduled the tapes for recoding. In addition, absence on a training day, and in-class responses or project papers showing misunderstandings of the protocols, triggered recoding of a student’s
248 PROMOTIONAL CONTENT EASTMAN ET AL. tapes. After the second tentatively approved coding, the two analyses were compared on administrative items and factual items about the carriage and promoted programs, spot construction, distance, pacing, length, background scene, and narrative structure. Any discrepancies in these items led to a third (or fourth) coding. During data entry by the researchers, the coding sheets were again examined for missing or inconsistent data. Whenever the ratings of the 14 appeals and one scaled information item differed by more than two steps on the nine-step scales, a recoding was undertaken (blind to the original evaluations). Because the number of students in the second participating class was not an exact multiple of the number of tapes to be coded, students were available for repeated recodings for class credit. In addition, one student was hired to recode the promos in 16 of the 218 tapes. When two codings of the appeals agreed within two degrees (within two steps on each of the 15 items) and on all other factual items, one of the codings, selected at random, was subsequently retained in the database. Finally, during statistical analysis, high measures of inconsis- tency on some appeals flagged two students’ work and led to recoding of all four of their tapes. As a final check, comparisons were made of 12 randomly selected pairs of final coding sheets. Following Holsti’s (1969) method, calculations showed 93% reliability on the 14 appeals (in other words, an average of five of the 126 decision steps differed by up to two degrees per pair). Results Altogether, 1,547 promos for 155 different situation comedies (for more than 300 different episodes) were located. Results showed that two-thirds (68%) of the promos were multiple spots promoting more than one program (in the patterns of two to four sitcoms together), leaving just one-third as single spots (32%). More than one third of the promos (37%) were for programs scheduled later the same week; 20% were for the same night; 17% for the next night; and 23% for the following week (generally the next episode); only 3% were for a program more than one week away. The promos’ narrative structure was similar from program to program as well as from network to network because of the current style of using a clip from the program with a voice-over narration, although occasionally, a star spoke directly to the audience (LL Cool J of In the House, and Wayans Brothers). Thus, lacking variance, no distinctions in impact could emerge for the narrative aspect of promo presentation. About one third (34%) of the comedy promos had a home or family situation as the primary situation, the second most frequent location was the workplace (31%), with a mix of bars/restaurants (15%), singles’ apartments (9%), and mixed other loca- tions (11%) having progressively smaller proportions of promo backgrounds. The number of cuts in the promos varied from less than three (33%), four to five (26%), six to eight (21%), to more than eight (20%). In general, coders assessed the promos’ pacing as fast (36%) or in the middle (36%), rather than very fast (15%) or slow or very slow (13%). The three kinds of humor were fairly well distributed across the sample, with 38% using cute type, 36% using surprise, 24% using sarcasm, and just 2% not funny. The variable of promo length was measured as the total clock time (in seconds) devoted to a single program. A 30-second multiple spot might have portions of differing lengths devoted to its programs. The distribution of lengths was as
249 JACR AUGUST 2003 follows: 27% were of 5 seconds or less, 39% were of 6–10 seconds, 15% were of 11–15 seconds, 10% were of 16–20 seconds, and 9% were of 21–30 seconds.5 Promo Content to Program Impact The first and most important question driving this study was whether reac- tions to content significantly impacted program ratings, and as Tables 1 to 3 show, they certainly appeared to do so. Although multiple regression analyses revealed that the well-established inheritance and structural variables related to promotion predicted the greatest amounts of variance in most analyses, several content-related variables also appeared to contribute significantly to subsequent program ratings. For example, Table 1 reports the analysis of the 1,547 promos for the total database of all 155 programs airing in 1998 and 1999. In the overall analysis, the combination of inheritance, structural variables (carriage program rating, number of promos aired), and five appeal and presen- tation variables accounted for 78% of the variance in program ratings (adj. R2 ⫽ .781). The content variables—fostering character identification, surprise humor, (not) seeming new, suspenseful, and promo length—produced changes in R2 totaling .062, a relatively small but still significant contribution to the total target program ratings in a business where tenths of a ratings point are treated as important because they are financially valuable. This result supports the first hypothesis—that content matters significantly, in addition to structure, in com- edy promotion. Program Popularity The second hypothesis posited a greater impact of content-related promotional variables for mid-rated comedies compared to high-rated or low-rated comedies. Table 2 shows the results of a stepwise regression analysis for the two middle quartiles of promoted programs, containing 85 mid-rated programs and 837 promos. Table 3 reports the results of a similar analysis of the 70 programs and 710 promos in the high- and low-rated quartiles. The overall difference between the two groups was substantial, as 7% more variance was explained for the mid-rated programs (adj. R2 ⫽ .840) than for the high- and low-rated shows (adj. R2 ⫽ .762). Moreover, more content variables, including the elements of humor and presentation, were significant factors in the mid-quartile ratings than in the extreme quartiles. However, contrary to Hypothesis 2, the content variables alone accounted for more of the ratings variance for the high- and low-rated shows (9% came from relating to characters, suspenseful, and seems new) while predicting less for the mid-quartile programs (just 6% came from star appeal, hip, cute humor, and length). Thus, Hypothesis 2 was not supported. Looking at the content variables, however, reveals some interesting differences between mid-rated programs and (1) those much more or much less successful and (2) as compared with the overall analysis reported in Table 1. Fewer separate content variables were significant predictors for the high- and low-rated shows, but one appears to be particularly important. Fostering identification with characters explains more than 4% of the variance for those programs, nearly twice as much as any other content variable in any of the regressions. The
250 PROMOTIONAL CONTENT EASTMAN ET AL. TABLE 1 Hierarchical (Stepwise) Regression Analysis of Appeal, Presentation, and Humor Variables for All Programs Variable B SE B B R2 Adj. R2 Step 1 .661 .658 Lead-in rating .948 .055 .813* Step 2 .044 .700 Lead-in rating .810 .059 .694* Carriage program rating .257 .054 .240* Step 3 .019 .717 Lead-in rating .799 .057 .685* Carriage program rating .250 .053 .233* Relate to characters .399 .125 .137** Step 4 .014 .729 Lead-in rating .761 .058 .652* Carriage program rating .237 .052 .221* Relate to characters .358 .123 .123** n of promos 6.788E-02 .024 .124** Step 5 .014 .742 Lead-in rating .745 .057 .639* Carriage program rating .238 .051 .222* Relate to characters .332 .121 .114** n of promos 7.344E-02 .024 .134** Surprise humor 1.061 .370 .119** Step 6 .009 .749 Lead-in rating .746 .056 .640* Carriage program rating .219 .051 .204* Relate to characters .377 .121 .130** n of promos 6.864E-02 .024 .126** Surprise humor 1.074 .365 .120** Program seems new ⫺ .282 .123 ⫺ .096*** Step 7 .011 .759 Lead-in rating .743 .055 .638* Carriage program rating .210 .050 .196* Relate to characters .280 .124 .096*** n of promos 7.474E-02 .023 .137** Surprise humor .913 .362 .102*** Program seems new ⫺ .343 .122 ⫺ .117** Program seems suspenseful .326 .121 .116** Step 8 .009 .768 Lead-in rating .758 .054 .650* Carriage program rating .172 .051 .160* Relate to characters .266 .122 .091*** n of promos 8.328E-02 .023 .152* Surprise humor .822 .358 .092*** Program seems new ⫺ .382 .121 ⫺ .130** Program seems suspenseful .335 .119 .119** Promo length .550 .220 .104*** Note: n of programs ⫽ 155; n of promos ⫽ 1,547; *p ⬍ .001; **p ⬍ .01; ***p ⬍ .05 negative coefficients for both appeals to hipness and cute humor suggest that contemporary audiences (at least younger ones, the avowed target audiences for most comedies) may respond best to spots with some edge but not to overt messages such as “look how cool this show is.” The results also provide some support for Hypothesis 3, which predicted that appeals would matter more than presentation or type of humor. Based on either the number of times appeals appear in the analyses, or on total predictive power, certain appeals (especially fostering character identification) explain more vari-
251 JACR AUGUST 2003 TABLE 2 Hierarchical (Stepwise) Regression Analysis of Appeal, Presentation, and Humor Variables for Mid-Rated Programs Variable B SE B B R2 Adj. R2 Step 1 .641 .636 Lead-in rating .883 .073 .800* Step 2 .130 .765 Lead-in rating .623 .070 .564* Carriage program rating .257 .038 .431* Step 3 .025 .788 Lead-in rating .544 .071 .493* Carriage program rating .255 .036 .428* n of promos 7.44E-02 .024 .174** Step 4 .016 .802 Lead-in rating .561 .069 .509* Carriage program rating .228 .036 .383* n of promos 6.809E-02 .023 .159** Star appeal .212 .082 .132*** Step 5 .017 .817 Lead-in rating .531 .067 .481* Carriage program rating .204 .036 .342* n of promos 6.3551E-02 .022 .149** Star appeal .288 .084 .179* Program seems hip ⫺ .331 .120 ⫺ .146** Step 6 .014 .830 Lead-in rating .539 .065 .488* Carriage program rating .179 .036 .300* n of promos 5.871E-02 .021 .137** Star appeal .379 .088 .236* Program seems hip ⫺ .392 .118 ⫺ .173* Cute humor ⫺ .776 .292 ⫺ .131** Step 7 .011 .840 Lead-in rating .562 .063 .510* Carriage program rating .152 .036 .256* n of promos 6.107E-02 .021 .143* Star appeal .366 .085 .288* Program seems hip ⫺ .387 .114 ⫺ .171* Cute humor ⫺ .719 .284 ⫺ .122*** Length .434 .180 .113*** Note: n of programs ⫽ 85; n of promos ⫽ 837; *p ⬍ .001; **p ⬍ .01; ***p ⬍ .05 ance than the type of humor employed or the presentation (where only length may be significant). Nonetheless, humor and presentation did contribute significantly to the variance in promos for mid-rated program ratings but not to the promos for highly popular or unpopular programs. Impact of Familiarity The fourth hypothesis predicted that appeals would impact the ratings for familiar comedies more than for unfamiliar comedies. Using familiarity as a binary (familiar/unfamiliar) sorting variable, the results reported in Table 4 and Table 5 indicate that although all of the variables accounted for nearly 20% more variance in the ratings for familiar (adj. R2 ⫽ .927) than for unfamiliar (adj. R2 ⫽ .759) programs, the contributions made by appeals were very similar. However, the effective appeals were very different for the two categories of programs: Structural variables and appeals contributed most significantly to the
252 PROMOTIONAL CONTENT EASTMAN ET AL. TABLE 3 Hierarchical (Stepwise) Regression Analysis of Appeal, Presentation, and Humor Variables for High-and Low-Rated Programs Variable B SE B B R2 Adj. R2 Step 1 .653 .648 Lead-in rating .967 .085 .808* Step 2 .046 .690 Lead-in rating .911 .082 .761* Relate to characters .760 .239 .219* Step 3 .027 .713 Lead-in rating .767 .097 .641* Relate to characters .680 .232 .196** Carriage program rating .394 .156 .207*** Step 4 .017 .727 Lead-in rating .746 .096 .623* Relate to characters .610 .229 .175** Carriage program rating .426 .153 .224** Program seems suspenseful .422 .203 .133*** Step 5 .026 .750 Lead-in rating .735 .091 .614* Relate to characters .628 .218 .181** Carriage program rating .420 .146 .221** Program seems suspenseful .554 .200 .175** Program seems new ⫺ .615 .229 .167** Step 6 .014 .762 Lead-in rating .707 .090 .591* Relate to characters .544 .218 .157*** Carriage program rating .394 .143 .207** Program seems suspenseful .601 .197 .190** Program seems new ⫺ .604 .223 ⫺ .164** n of promos 7.583E-02 .038 .127*** Note: n of programs ⫽ 70; n of promos ⫽ 710; *p ⬍ .001; **p ⬍ .01; ***p ⬍ .05 ratings for familiar programs. Presentation (length) was a factor only for unfam- iliar shows, whereas type of humor (surprise) was significant only for familiar show ratings. Table 4 shows that for the 45 familiar programs promoted in 492 promos, two structural variables (inheritance and carriage program rating) accounted for TABLE 4 Summaries of Hierarchical (Stepwise) Regression Analysis of Appeal, Presentation, and Humor Variables for Familiar Programs Variable B SE B B Lead-in rating .729 .053 .668* Carriage program rating .102 .025 .203* Network popularity .329 .075 .194* Surprise humor 1.050 .224 .195* Program seems hip ⫺ .405 .109 ⫺ .174* Star appeal .189 .075 .114*** R2 ⫽ .773 for Step 1; .079 for Step 2; .024 for Step 3; .031 for Step 4; .021 for Step 5; and .011 for Step 6. Adjusted R2 ⫽ .767 for Step 1; .844 for Step 2; .866 for Step 3; .897 for Step 4; .917 for Step 5; and .927 for Step 6. Note: n of programs ⫽ 45; n of promos ⫽ 492; *p ⬍ .001; **p ⬍ .01; ***p ⬍ .05
253 JACR AUGUST 2003 TABLE 5 Summaries of Hierarchical (Stepwise) Regression Analysis of Appeal, Presentation, and Humor Variables for Unfamiliar Programs Variable B SE B B Lead-in rating .720 .071 .610* Carriage program rating .289 .087 .204* Identify with characters .376 .159 .120*** Program seems suspenseful .461 .154 .152** Program seems new ⫺ .432 .160 ⫺ .134** N of promos 7.943E-02 .030 .137** Length of promo .625 .280 .108*** R2 ⫽ .654 for Step 1; .043 for Step 2; .026 for Step 3; .012 for Step 4; .016 for Step 5; .012 for Step 6; and .011 for Step 7. Adjusted R2 ⫽ .651 for Step 1; .692 for Step 2; .716 for Step 3; .725 for Step 4; .739 for Step 5; .749 for Step 6; and .759 for Step 7. Note: n of programs ⫽ 110; n of promos ⫽ 1,055; *p ⬍ .001; **p ⬍ .01; ***p ⬍ .05 approximately 84% of the ratings variance. Three appeals (network popularity, seeming hip, and star power) added an additional 5%. The remainder (3%) was attributed to the use of surprise humor. In contrast, Table 5 shows that the conventional structural variables account for only 70% of the ratings variance for the 110 unfamiliar programs, and presentation (length) adds an additional 1%. The significant appeals for unfam- iliar shows were identifying with characters, suspense, and (negatively) seeming new, which collectively added 5%. Thus, Hypothesis 4 was not supported because the size of variance accounted for by the appeals was identical for both familiar and unfamiliar program ratings. The differences lay in the contributions of humor and presentation. Nonetheless, appeals and humor together accounted for 8% for familiar programs, whereas appeals and presentation accounted for just 6% for unfamiliar programs. Discussion First, and most important, these findings strongly support the presumption that promos have a significant and measurable impact on program ratings. Second, the content—not just the structure—of promos was demonstrated to impact program ratings in this study. Third, affective appeals contributed sub- stantially more to promotional impact than the forms of humor or execution in the promos. Fourth, the impact of promo content was greater for familiar than unfamiliar programs. Fifth, the impact of humor and presentation were greater for promos promoting mid-rated than for top- or bottom-rated comedies. At the same time, the results also confirm the expected powerful impact of inherited viewing on program ratings and the influential role of scheduling of promotional messages to reach a wide audience. The consistency of such findings with the long tradition of programming research adds credence to these results. Particularly new in this study was the identification of some of the specific appeals that made a difference. For television comedies, the key content ele- ments appear to be: having characters that can be identified with; having dramatic suspense and realism; using big stars; avoiding self-identified trendi-
254 PROMOTIONAL CONTENT EASTMAN ET AL. ness and newness in appeals; and avoiding cute humor. While these results may not generalize to other program genres, and some key appeals may have been omitted, these findings provide the first examination of promotional content outside the boundaries of sexual and violent content. The importance of foster- ing identification with characters in hit programs will be no surprise to program producers and writers, but what emerges here is its importance to successful promotion. The ability of audiences to relate to characters, and to feel some element of suspense related to the program episode, appear to be significant factors in the effectiveness of promotion for both high-rated and low-rated programs. Successful promotion of mid-rated shows, in contrast, seems to be driven to a greater degree by structural factors, by appealing to the star power of the actors, and by allotting sufficient time to the spots (or to the individual programs in a multiple spot) to get the promotional message across (promo length). Further, these findings offer preliminary evidence that the content with which programs are promoted affects viewing decisions. This significant rela- tionship should give experimental researchers the confidence to test such specific mechanisms as excitation transfer and priming that could be behind the relationship. On the theoretical level, these findings add strength to the presumption that even the very modest interest stirred by on-air promotion measurably primes behavior related to program ratings. This occurs despite the low level of arousal—if that is the appropriate word for such a mundane effect—created by promos. It occurs despite the regular presence of intervening activities including other programming and the extended time lapse between the airing of promos and the airing of promoted programs. For familiar shows in particular, the affective and humor components are effective (and more important), probably because they aid viewers’ recall of positive experiences and build anticipation through their desire for subsequent similar experiences. To be maximally effec- tive, then, promos should contain cognitive and emotional elements that trigger viewer recall of previous experiences with a program and emphasize the success- ful nature of the show. As examples, one might point to the very successful “must-see-TV” theme and the widespread use of variants of “Last Sunday, 30 million Americans saw. . . .” Limitations and Future Research As previously mentioned, executional (presentational) variables are commonly used in laboratory and marketing studies. Their infrequent appearance in these analyses may be measurement artifacts or may represent real differences between advertising and promotional messages. The variable of promo length, an aspect of pacing, becomes of particular interest because it was the only presentational variable to show up in the overall and mid-quartile regression analyses as a significant factor in promos’ impact on program ratings. Another qualification is that the regression method employed lacks the rigor of laboratory experimentation for deducing causal explanations, while simul- taneously being too demanding for situations where the variance is small (as in the smaller networks’ ratings). Nonetheless, while more extended tests of specific findings are highly desirable, along with using evaluators (coders) reflecting a wider range of demographics, these results have remarkable simi-
255 JACR AUGUST 2003 larity to the findings of some previous studies of structural variables in pro- motion (Eastman & Newton, 1998a, 1998b). Thus, despite limitations, the findings confirm some general outlines that should be expected to continue to hold up in subsequent research into promotion. In addition, given that the affective appeals and content pacing were rated by college students and were necessarily somewhat subjective (or at least, time bound), generalization from these findings to a wider range of age groups is not warranted. While the specific appeals probably differ for demographic groups other than young adults, these finding do provide the first steps toward isolating the powerful content elements in promos. Moreover, college students are part of the demographic groups that programmers want to attract as viewers (adults 18–34 and 18–49) and are an ideal target for building loyalty (bonding) to specific programs, one of the other key goals of promotion. While the collective contribution of the content variables to the variance accounted for in this study remains small, it must be kept in mind that the end-of-season differences in ratings from the number one to number two network may be as small as one-tenth of a ratings point. This was indeed the case in May 1999, the end of that year’s 36-week season. CBS was ahead by one-tenth of a ratings point (9.0 to 8.9, season-to-date) and was equal in share points (at 15), a situation that was unchanged for much of the year. ABC, the third-ranked network, was usually less than one ratings point behind throughout that year. Thus, the amount of variance available to be found is small but vital to the industry because it represents millions of dollars in advertising revenue, and important to scholars in demonstrating the applicability of priming and salience theories. Future applied research needs to assess the kinds of content factors salient to appeals in promos for other genres of programs, as well as the ideal and maximum distance of promos from various genres of programs. It can be reasonably expected, for example, that effects would occur over greater lengths of time for specials and movies but only over relatively shorter lengths of time for episodic promos for series. Future theoretical research in laboratory experi- ments needs to pinpoint the exact mechanism that accounts for the secondary excitation transfer effect or priming effect. Practical Applications The importance of on-air promotion when assessing the sources of prime-time ratings lends practical value to this study. The findings have implications for both television programming executives and scholars interested in teaching and researching program promotion because they provide empirical evidence to guide the design of effective comedy promos. First, the results illustrate how similar program promotion is to product advertising. As in advertising, the type of appeal used matters a great deal. Another similarity is that, as in product advertising, brand (or program) familiarity and popularity impact the roles of appeals in determining promotion’s effectiveness. The pattern of results in this study should give television executives added confidence in drawing on the knowledge gained from advertising research when designing effective strategies for program promotion. Second, the study shows the need to distinguish pro- grams by their ranking in the ratings when creating promos and probably to
You can also read