The Influence of Program Context and Order of Ad Presentation on Immediate and Delayed Responses to Television Advertisements

Valerie Starr, University of Connecticut
Charles A. Lowe, University of Connecticut
ABSTRACT - Differences in program context significantly influence the effectiveness of advertisements appearing at different points within the same program. Specifically, recall for brands and products advertised was found to be greater when a low-involvement point of the surrounding story was interrupted. Also within a relatively uninvolving program context, the first advertisement to appear in a series of advertisements had a significant advantage in recall over the second and third ads, perhaps because viewers became impatient for the program to resume and paid less attention to the ads. These results have implications for optimal placement of advertisements to enhance their effectiveness.
[ to cite ]:
Valerie Starr and Charles A. Lowe (1995) ,"The Influence of Program Context and Order of Ad Presentation on Immediate and Delayed Responses to Television Advertisements", in NA - Advances in Consumer Research Volume 22, eds. Frank R. Kardes and Mita Sujan, Provo, UT : Association for Consumer Research, Pages: 184-190.

Advances in Consumer Research Volume 22, 1995      Pages 184-190

THE INFLUENCE OF PROGRAM CONTEXT AND ORDER OF AD PRESENTATION ON IMMEDIATE AND DELAYED RESPONSES TO TELEVISION ADVERTISEMENTS

Valerie Starr, University of Connecticut

Charles A. Lowe, University of Connecticut

ABSTRACT -

Differences in program context significantly influence the effectiveness of advertisements appearing at different points within the same program. Specifically, recall for brands and products advertised was found to be greater when a low-involvement point of the surrounding story was interrupted. Also within a relatively uninvolving program context, the first advertisement to appear in a series of advertisements had a significant advantage in recall over the second and third ads, perhaps because viewers became impatient for the program to resume and paid less attention to the ads. These results have implications for optimal placement of advertisements to enhance their effectiveness.

INTRODUCTION

Millions of dollars are spent annually on the creation and testing of television advertisements. Historically, marketers have assumed that audience attention to their message relied heavily, if not exclusively, upon the size and characteristics of the viewing audience. Specifically, the larger the audience and the greater the similarity of the audience to the target consumer, the more effective (at least potentially) the advertisement. Understandably, decisions of when and where to schedule television advertisements often have been based on estimates of audience size (e.g., Nielsen ratings) and audience characteristics (e.g., demographic or psychographic profiles) (Engel, Blackwell, & Miniard, 1990).

Recently, advertising researchers have begun to consider the context in which an advertisement appears, and the possible influence of this context on advertising effectiveness (Schultz, 1979). Those aspects of program context previously investigated include humor (Goldberg & Gorn, 1987), emotional content or mood (Pavelchak, Antil, & Munch, 1988; Schuman, 1986), similarity or congruence of the program and ad (Conard & Lowe, 1988), and the degree to which the program story line "involves" the audience (Soldow & Principe, 1981; Kennedy, 1971).

These several program context effects on advertising effectiveness should be of considerable interest to marketers concerned with promoting their products, since these effects suggest that size and similarity of the audience may not be of primary importance. Thus, the prevailing desire to have one's advertisements appear during popular (i.e., "prime-time") programs which have large audiences may be quite inappropriate even when these large audiences are dominated by the appropriate target consumers. For example, Pavelchak et al. (1988) found that emotionally involved viewers of Super Bowl XX could recall fewer advertisements shown during the game than did neutral viewers, even though the former watched more of the game and, presumably, saw more advertisements. The implication is that unless marketers consider program context effects, the financial costs of placing advertisements on specific television programs may be money spent unnecessarily.

Pavelchak et al. (1988) identified two competing predictions regarding the influence of program context on advertisement effectiveness. One prediction is that the experience of positive emotions would facilitate information processing because pleasant emotional states could be expected to increase the elaboration and activation of (more) cognitive categories than would unpleasant emotional states (Isen, 1984). This increased availability of cognitive categories would enable more encoding and recall of persuasive information (i.e., advertisements) presented while viewers are experiencing positive rather than negative emotions. In fact, some research has shown that viewers did exhibit better recall for advertisements which were presented during a "happy" program compared to a "sad" program (Goldberg & Gorn, 1987).

A second prediction is that emotional arousal should inhibit information processing because any "intense" feeling, regardless of its valence, would impair viewers' ability to attend to and/or to process information presented while these intense feelings are being experienced. Thus, advertisements presented during emotionally involving programs, regardless of the valence of the emotions elicited, would be screened out (Pavelchak et al., 1988). In short, any emotionally involving (e.g., very happy or very sad) program would impair viewers' ability to attend to and hence to recall advertisements that appear on the program.

Similar detrimental effects of arousal on recall have been predicted by Steiner (1963), who utilized the Gestalt principle of "need for closure." This principle, as applied to television programming, suggests that viewers experience a need for completion (or closure) when the plot of a program remains unresolved. Further, the more emotionally involving or suspenseful the story line, the greater viewers' need for closure. This need for closure regarding the story plot may cause the advertisements to be perceived as "noise," and as such, viewers may pay little attention to advertisements presented during an involving or suspenseful program (Kennedy, 1971). The net result of these advertising distractions would be lower recall for advertisements which interrupt a suspenseful program context.

Despite the intuitive appeal of this line of reasoning, Kennedy (1971) found little support for the prediction that advertisements presented during high-involvement programs would not be recalled as well as the same advertisements presented during low-involvement programs. Specifically, unaided brand recall showed a trend (p<.1) in the predicted direction, but aided brand recall, recall of product attributes, and recall of the advertising message did not differ as a function of the program context.

More recent research by Soldow and Principe (1981) investigated the effects of involvement on advertising effectiveness by presenting subjects with a series of three advertisements in the context of an involving or suspenseful program ("Baretta"), a non-involving or non-suspenseful program ("The Brady Bunch"), or in the absence of any program context (i.e., a control group). Their results revealed that compared to advertisements presented in isolation or in a low-involvement context, advertisements presented in a high-involvement context scored lower on several measures of advertising effectiveness including unaided brand recall, sales message recall, purchase interest, and positive attitude-toward-the-ad. Further, Soldow and Principe (1981) reported an interesting pattern of results due to ordinal position for this latter measure. While viewers of the high-involvement program reported less positive attitudes toward all advertisements than did viewers in the other two conditions, the high-involvement viewers exhibited more positive attitudes toward the first commercial compared to the second or third (last) commercials.

These differences in attitude-toward-the-ad were explained by the aforementioned principle of closure. According to Soldow and Principe (1981), their high-involvement condition created a greater need for closure, so that ads shown during the high-involvement program were more disruptive and disturbing than were the same ads shown during the less involving program. Further, Soldow and Principe suggested that the first ad in the sequence was accepted by viewers as an "inevitable" event, but that the subsequent ads served to increase viewers' annoyance because they prolonged viewers' need for closure with regard to the story line of the program.

Although the principle of closure provides an explanation which appears to account for the data, the order of presentation of the three commercials used by Soldow and Principe was not varied. Therefore, unique characteristics of the advertisement which always appeared in the first position, or properties of that ad which may have had a unique relationship to the high-involvement programs, cannot be eliminated as possible causes for the differences in audience response as a function of the ordinal position and program context.

The present study sought to determine the potential effects of program context and ordinal position of the advertisement on several measures of advertising effectiveness assessed at two different times. Several researchers have found that even within the same general type of program (e.g., situation comedy, suspense, etc.), different programs may not have the same influence on the effectiveness of the ads appearing within them (Yuspeh, 1979). To avoid such effects due to potential program differences, the present study manipulated context by presenting a series of three advertisements at either a relatively uninvolving or a relatively involving segment of the same program. To avoid the potential confounding of ordinal position with a particular advertisement and to lend generalizability across specific advertisements, each of three different advertisements was presented first, second, and third in a sequence of three advertisements. In addition, the present study included both "immediate" and day-after assessment of the advertisements' effectiveness. While most researchers have employed only immediate assessment, day-after assessment has become the standard measure by which marketing professionals evaluate the effectiveness of their advertisements (Haller, 1983).

Hypothesis 1: Drive for closure was expected to influence viewers' affective reactions to advertisements. Specifically, since drive for closure should be stronger during highly involving program segments than during less involving program segments, advertisements interrupting a highly involving program segment should be seen as more interfering, should create more annoyance, and should result in less positive attitudes toward the ad and in lower purchase interest in the product advertised than should advertisements interrupting a less involving program segment.

Hypothesis 2: Based on previous results demonstrating interference effects on recall of advertisements due to the emotional content of the program, it was expected that advertisements which interrupted highly involving program segments would be recalled less well (on each of several recall measures) than would advertisements which interrupted less involving program segments. In addition, all measures of recall were expected to be influenced by time of assessment such that viewers would show lower recall for day-after compared to immediate assessment.

Hypothesis 3: With regard to the ordinal position of advertisements, while viewers may have become conditioned to accept the inevitability of a "commercial break" (Soldow and Principe, 1981), lengthy commercial breaks which contain several advertisements should increase viewers' experienced need for closure regarding the program story line. Thus, we could expect that within a sequence of advertisements, ads which appear later would be seen as more annoying and as more interfering, and consequently, would be recalled less well and elicit fewer positive reactions (e.g., liking and purchase interest) than would ads which appear earlier.

METHOD

One hundred forty-eight students (84 females and 74 males) enrolled in an undergraduate psychology course at a large eastern university served as subjects in the actual experiment. An additional 48 students participated in the pretesting sessions. The study employed a mixed design, with two between groups variablesCinvolvement (high vs. low) and time of recall (immediate vs. day-after), and one within groups variableCordinal position of the advertisement (first vs. second vs. third).

To select the stimulus advertisements, twelve different advertisements which had been aired on television in the past two years to advertise a wide range of products relevant to college-age consumers were shown to 29 subjects. For each of these twelve advertisements, subjects indicated their liking [0=not at all, 6=very much] and the importance to them that they purchase a particular brand of the product advertised [0=not at all important, 6=very important]. Based on these responses (means are shown in Table 1), three advertisements were selected for the experiment (GE light bulbs, Hansen's fruit juice, and Aim toothpaste) because they did not differ in likability [F(2,26)=2.06, n.s.] or in brand importance [F(2,26)=2.16, n.s], were moderately likeable (M=2.64), and described products for which particular brands were moderately important (M=2.61).

To define the involvement manipulation, 19 subjects watched a videotaped episode of "Miami Vice" and rated the suspensefulness of the plot [0=boring, 6=very suspenseful] at four different points in the episode. Two of these points of interruption were selected for use in the experiment because they had different levels of suspense [M=3.11 vs. M=0.74, F(1,18)=49.93, p<.0001] and because they were separated by only 2.5 seconds of programming. Both points of interruption (i.e., high- and low-involvement points) were approximately 20 minutes into the "Miami Vice" episode.

Since the three advertisements could be arranged in six possible orders, twelve stimulus tapes were constructed by inserting the selected advertisements into either the low suspense or high suspense points of the "Miami Vice" episode. By employing twelve different tapes, each advertisement appeared an equal number of times in each ordinal position (first, second, or third) and in each sequential position relative to the other two advertisements. To conform to the cover story, the program was extended for approximately 12 minutes beyond the presentation of the three advertisements. Since the points of interruption for the high- and low-involvement conditions were separated by only 2.5 seconds, the "Miami Vice" segments which followed the advertisements were nearly identical in terms of total length and program content, included both suspenseful and non-suspenseful segments, and ended at precisely the same point.

Subjects were run in groups of 8 to 12 at staggered times throughout the day. The experimental room was arranged like a living room and contained couches, chairs, and a television monitor. The cover story informed subjects that as part of a study of "person perception," they would be shown a short program segment and would then be questioned about characters from the program. Each experimental group was then shown one of the twelve stimulus tapes containing the "Miami Vice" episode and the three advertisements. Both involvement (high or low) and order of presentation (first, second, and third) were manipulated by randomly assigning tapes to experimental sessions, with the restriction that each of the six orders had been presented to an equal number of subjects in each involvement condition.

TABLE 1

RATINGS OF LIKING AND BRAND IMPORTANCE FOR ADS SELECTED

Immediately following the conclusion of the "Miami Vice" episode, subjects attending each session were randomly assigned to one of two time of assessment conditions. "Day-after" assessment subjects were given a sealed envelope, told not to open the envelope until the next day (approximately 24 hours later), and dismissed. These envelopes contained the experimental questionnaire together with instructions for completing the questionnaire and returning it (sealed in the envelope provided) within 24 to 36 hours after the experimental session to a central location in the psychology building. Compliance was encouraged by making partial course credit contingent on following these instructions regarding when to open the envelope and how to complete and return the questionnaire. "Immediate" assessment subjects completed the experimental questionnaire once the "day-after" subjects had left the experimental room.

Measures: The experimental questionnaire included items which assessed the effectiveness of the involvement manipulation, subjects' reactions to the "Miami Vice" episode, and several dependent variables. As checks on the involvement manipulation, subjects were asked to indicate their perceived level of suspense at the point when the advertisements interrupted the show (0 to 6) and their experienced annoyance in response to this interruption (0 to 6). Assessment of the "Miami Vice" episode included asking subjects whether or not they had seen the particular episode before (0="no," 1="maybe," and 2="yes") and how much they had enjoyed the episode (0 to 6). The dependent variables included items which assessed (a) the perceived interference (0 to 6) of each advertisement, (b) liking (-3 through 0 to +3) for each advertisement and purchase interest in each product advertised (0 to 6), and (c) unaided recall of product type, brand name, message content and image content for each advertisement (open-ended responses). The open-ended responses for recall of product type and brand name were scored as either correct (0) or as incorrect (1), and the open-ended responses for message and image content were scored as no recall (0), partial recall (1) or total recall (2).

RESULTS

The data were analyzed using 3-way mixed-design ANOVAs which had involvement (high versus low) and time of assessment (immediate versus day-after) as between groups variables, and ordinal position (first, second, third) as the within groups variable. Minimal attrition occurred as a function of the time of assessment. Only 3% of the subjects in the day-after assessment conditions failed to return their completed questionnaires more than 24 hours or less than 36 hours from the initial viewing of the stimulus tape, and these early and late returns were not used in the analyses. Considerably more attrition occurred in responses to the liking and purchase interest questions, both of which offered subjects a "don't know" option. Overall, 44% of the subjects (42% in high involvement conditions and 46% in low involvement conditions) chose this option, thus accounting for the reduced degrees of freedom reported for analyses on liking and purchase interest ratings.

Manipulation Checks

The only significant effect obtained for subjects' ratings of the level of suspense at the point of interruption was a main effect for involvement [F(1,144)=15.53, p<.001]. Specifically, subjects in the high-involvement conditions (M=4.0) judged the point at which the advertisements interrupted the program to be more interesting or suspenseful than did subjects in the low-involvement conditions (M=3.6).

Although 20% of the subjects reported that they "thought" they previously had seen the particular episode of "Miami Vice" used in the present study, this percentage, as well as subjects' ratings of enjoyment (M=3.5), remained constant across all experimental conditions. Thus, any effects due to possible previous exposure to the program or to the level of viewer enjoyment of the program were constant across experimental conditions.

Effects Due to Involvement

Affective Reactions: Hypothesis one predicted that viewers' drive for closure would be stronger when the advertisements interrupted a highly involving program segment, and that such advertisements would be seen as more interfering, as more annoying, as less likeable and would result in lower purchase interest compared to the same advertisements when they interrupted less involving program segments. These predictions were partially supported. Subjects felt that the three advertisements did interfere more with their enjoyment of the program [M=3.1 vs. M=2.5, F(1,144)=5.42, p<.02] and were marginally more annoying [M=4.0 vs. M=3.6, F(1,144)=3.51, p<.06] when the ads appeared during a more suspenseful rather than a less suspenseful point in the program. In addition, subjects experienced more interference with their enjoyment of the program as the advertisements continued (see ordinal position effects reported below). By contrast, subjects' liking for each of the three advertisements [F(1,80)=1.62, n.s.] and subjects' purchase interest in the products advertised [F(1,80)<1, n.s.] were not influenced by involvement.

Recall: Hypothesis two predicted that advertisements which interrupted highly involving program segments would be recalled less well on each of the several recall measures than would advertisements which had interrupted less involving program segments. Again these predictions were partially supported. Main effects due to involvement for both brand name recall [F(1,144)=5.32, p<.02] and product type recall [F(1,144)=2.67, p<.1] indicated that recall was lower for advertisements which interrupted the program episode at a highly involving point rather than at a less involving point, although this effect was marginal for recall of product type (for brand, M=0.22 vs. M=0.34; for product, M=0.36 vs. M=0.45). Differences due to involvement were not obtained for recall of the advertisements' message contents or image contents (F's<1.00 on both measures).

TABLE 2

DIFFERENCES IN RECALL AND LIKING OVER TIME

TABLE 3

EFFECTS OF ORDINAL POSITION ON PERCEIVED INTERFERENCE AND RECALL

Time of Assessment

As shown in Table 2, time of assessment had the expected effects on each of the four recall measures. Recall of product type [F(1,144)=15.98, p<.001), brand name [F(1,144)=8.40, p<.01], advertisement message [F(1,144)=7.18, p<.01] and advertisement images [F(1,144)=8.01, p<.01] were each lower when they were assessed the "day-after" rather than "immediately." In addition, liking for the advertisements was similarly influenced by time of assessment, with day-after assessment (M=3.4) showing significantly less liking [F(1,80)=5.67, p<.01] than "immediate" assessment (M=3.8). Although the means for purchase interested followed this same pattern (for day-after, M=3.3, for immediate, M=3.61), this difference did not reach statistical significance [F(1,80)=2.19, p<.15].

Effects of Ordinal Position

Hypothesis three predicted that advertisements which appeared later in a sequence would be seen as more annoying and more interfering, and consequently, would be recalled less well than would advertisements which appeared early in a sequence. The relevant means for which significant ordinal position effects were obtained are presented in Table 3, together with the results of specific comparisons conducted between the means for each dependent variable.

As expected, the perceived interference of the advertisements increased as the advertisements continued to appear [F(2,288)=8.80, p<.001]. As shown in Table 3, specific comparisons revealed that the third advertisement was seen as interfering significantly more (p<.05) with subjects' enjoyment of the program than either of the first two advertisements. Ordinal position did not influence recall of the advertisements' message contents [F(2,288)=2.09, p<.13] or image contents [F(2,288)=1.14, p<.32] but did influence recall of both product type [F(2,288)=2.88, p<.06] and brand name [F(2,288)=3.51, p<.03]. Specific comparisons indicated that both product type and brand name were recalled better for the first advertisement shown than for the third advertisement shown. However, these results for brand recall were qualified by a marginally significant involvement by ordinal position interaction [F(2,288)=2.45, p<.09]. As shown in figure 1, the relative advantage of the first ad over the third ad occurred only under conditions of low involvement.

The results for the purchase interest measure are especially interesting, since two significant interactions involving ordinal position were obtained. First, a significant time of assessment by ordinal position interaction [F(2,154)=5.20, p<.01] revealed that the time of assessment effect held only for the first advertisement shown (see figure 2). For both the second and third advertisements, time of assessment had no effect on purchase interest. Second, a significant involvement by ordinal position interaction [F(2,154)=3.18, p<.04] indicated that involvement influenced purchase interest only for the second advertisement shown (see figure 3). Second or middle advertisements which interrupted low involvement program segments produced significantly more purchase interest (p<.05) than did (the same) advertisements when they interrupted highly involving program segments.

DISCUSSION

The results of the study suggest that when advertisements interrupt an involving or suspenseful part of a television program, viewers are annoyed and feel advertisements hinder enjoyment of the program more than when advertisements appear at a relatively uneventful point in the story. These reactions, or at least memories of these reactions, persistCwhen questioned the day after exposure to the ads, viewers remembered that the interruption had been unpleasant, even if they had forgotten the brands or products advertised.

FIGURE 1

THE INFLUENCE OF ORDINAL POSITION AND INVOLVEMENT ON BRAND RECALL

FIGURE 2

THE INFLUENCE OF ORDINAL POSITION AND TIME OF ASSESSMENT ON PURCHASE INTEREST

Perhaps more importantly, recall of the advertising message was similarly affected. Regardless of the specific ads used, brands and products advertised at a more involving point in the surrounding program are generally remembered less well than those advertised at a less suspenseful point--a disadvantage which again persists over time (i.e., day-after measures). The "need for closure" explanation proposed by Steiner (1963) would seem to explain this effect. Advertisements inserted at high suspense points create a greater need for closure than do the same advertisements when inserted at low suspense points in the story line. This heightened need for closure could lead to heightened arousal (i.e., feelings of annoyance and interference with enjoyment), which in turn could cause viewers to pay less attention to and thus remember less well the advertisements' messages.

Interestingly, while advertisements inserted at a suspenseful point in the program may "delay" closure, this delay did not affect viewers' overall enjoyment of the program. Viewers claimed to enjoy the program equally, regardless of when the advertisements appeared. Thus, Isen's (1984) suggestion that differences in recall result from more efficient processing due to increased "positivity" of viewers' emotions does not account for the obtained results.

While the above discussion pertains to differences in an advertisement's effectiveness due to its insertion in the same program at points which differ in their level of suspense, the same could hold true for an advertisement inserted into different programs which differ in their overall level of suspense. Specifically, an advertisement presented during a highly suspenseful program might be less effective than the same advertisement presented during a less suspenseful program. To the extent that "suspenseful" programs attract larger audiences, the paradox arises that while advertisements shown during these programs would be seen by more people, viewers are likely to pay less attention to the advertising messages.

The ordinal position of an advertisement within a sequence of advertisements also seems to influence recall of the advertisement's message. Regardless of time of assessment, or the point of program interruption, the first advertisement in a sequence achieved better recall for both product type and brand name than did either of the following ads. In addition to these recall differences, the results of the present study also suggest that ads appearing later in a commercial break may be perceived as "more interfering" with viewers' enjoyment. Again, the need for closure explanation would seem to account for these results. Viewers' need for closure could increase as the advertisements continue to appear and viewers see later ads as more interfering, which could cause viewers to pay less attention to and remember less well the advertising messages presented in ads appearing later in the sequence.

FIGURE 3

THE INFLUENCE OF ORDINAL POSITION AND INVOLVEMENT ON PURCHASE INTEREST

There was some indication that the influence of context may be more robust than the influence of ordinal position. While the context effects obtained on both affective and recall measures held true regardless of an advertisement's ordinal position, ordinal position effects did not generalize across context. For example, the advantage of the first ad over subsequent ads in brand recall disappeared under conditions of high involvement.

It should not be surprising that time of assessment had a substantial influence on viewers' reactions to the advertisements they were shown. Although viewers can recall several features of advertisements shown in sequence when they are asked to do so a short time later (e.g., approximately 20 minutes), their recall of these features, as well as their liking for the ads, is substantially reduced in as little as twenty-four hours. For example, in the present study, day-after recall across all four recall measures (type of product, brand name, message content, and image content) was reduced by an average of nearly 40%. It is not surprising, then, that marketers advocate the use of multiple exposures to reduce this effect, even when this strategy appears to have diminishing returns (Krugman, 1975).

To summarize, the results of the present study show that the effectiveness of an advertisement may be influenced by the point at which it is inserted into a program as well as its ordinal position vis-a-vis other advertisements. The implication of these findings is that decisions regarding the placement of advertising should not be based exclusively on the size and composition of the viewing audience, or on broad program categories. Ironically, the contexts of the programs which typically attract advertisers may negate the benefits expected from strategic development and purposeful placement of advertising.

REFERENCES

Conard, M. & Lowe, C. (1988). The effectiveness of television advertising: Program context and advertising orientation. Paper presented at the annual convention of the American Psychological Association.

Engel, J.F., Blackwell, R.D., & Miniard, P. (1990). Consumer Behavior (6th ed.) Chicago: Drydon Press.

Goldberg, M.E. & Gorn, G.J. (1987). Happy and Sad T.V. Programs: How they affect reactions to commercials. Journal of Consumer Research, 14, 387-403.

Haller, T. (1983). Danger: Marketing Researcher at Work. Westport, CT: Quorum Books.

Krugman, Herbert E. (1975), What makes advertising effective. Harvard Business Review, March-April, 96-103.

Isen, Alice M. (1984). The Influence of Positive Affect on Decision-Making and Cognitive Organization. In T.C. Kinnear (Ed.), Advances in Consumer Research, Vol. 11. (pp. 534-537). Provo, UT: Association for Consumer Research.

Kennedy, J.R. (1971). How Program Environment Affects T.V. Commercials. Journal of Advertising Research, 11, 33-38.

Pavelchak, M.A., Antil, J.H., & Munch, J.M. (1988). The Super Bowl: An investigation into the relationship among program context, emotional experience, and ad recall. Journal of Consumer Research, 15, 360-367.

Schultz, D.E. (1979). Media research users want. Journal of Advertising Research, 19, 13-17.

Schumann, D.W. (1986). The influence of program contexts on commercial effectiveness: A review of the literature. Unpublished Manuscript, University of Tennessee College of Business Administration.

Soldow, G.F. & Principe, V. (1981). Response to commercials as a function of program context. Journal of Advertising Research, 21(2), 59-64.

Steiner, G.A. (1963). The People Look at Television. New York: Knopf.

Yuspeh, Sonia (1979). The Medium versus the Message. In George Hafter (Ed.), Proceedings of the 10th National Attitude Research Conference, (pp. 109-138). Chicago, IL: American Marketing Association.

----------------------------------------