Ad pod effects in TV advertising: order, adjacency, and informational\emotional appeal.

Manning, Stacey ; Larsen, Val

INTRODUCTION

The effects of various contexts on responses to advertising, especially television advertising, have received considerable attention from researchers. The relationship between programs and the emotions they evoke on the one hand and ad responses on the other has been especially well studied. For example, Goldberg and Gorn (1987) have explored the effect of happy and sad TV programs on reactions to commercials. They found that a happy program produced a more positive mood as viewers watched commercials, greater perceived commercial effectiveness, more affectively positive cognitive responses, and to some extent, better recall. Studying the power of feelings in understanding advertising effects, Edell and Burke (1987) also found that antecedent negative and positive feelings are an important predictor of ad effectiveness and that these feelings significantly influence beliefs about brand attributes and attitudes toward the brand.

However, important as the programs in which an ad is embedded and the feelings they evoke may be to ad effectiveness, other aspects of context may be still more important. The immediate context for most television commercials is not the television program in which it is embedded but rather, other, adjacent commercials also included in the pod of ads. Precisely what precedes and follows an ad is also, of course, a function of an ad's position within the pod of ads, so pod position effects are another potentially important aspect of ad context. Our objective in this study was to explore these more immediate context effects.

PRIMACY AND RECENCY EFFECTS

Two recently published studies on ad pod position effects have provided some support for the idea that serial position within an ad pod may drive ad responses. Testing the applicability of the large body of psychological research on primacy and recency (e.g., Kerr, Ward, and Avons, 1998), Zhao (1997) found that advertisements fared "better ... in an earlier position within the pod." Two other researchers, Pieters & Bijmolt (1997), found that "primacy and recency have only modest effect sizes [and that] placing a commercial first is better than placing it last." Important as these studies are in giving an idea of the effects that advertisers can expect from various serial positions in an ad pod, each has important limitations. Zhao's study was a quasi-experiment conducted in the aftermath of the Super Bowl. While the study has good external validity, its internal validity is questionable. Zhao was unable, for instance, to control for effects of different brand names and of various executional variables. Since all subjects saw the same set of ads in the same positions within the pod, the effects of order were confounded with execution and brand effects. And Super Bowl ads being notorious for their clever writing, stylistic innovation, and high production values, Zhao's sample of ads cannot be seen as representative of ordinary advertising. The Pieters and Bijmolt (1997) study was conducted in The Netherlands where ads occur only between programs and where the average pod is much longer than it is in the United States. Thus, the blocks of commercials they used in their study averaged 12.7 minutes in length. Findings on ad pods of that length may well be inapplicable to the much shorter ad pods typical in the United States, especially since American ad pods are generally much more embedded within television programs. In this study, ad pods were more representative of those typical in the United States, both in their length (6 minutes) and in being embedded in segments of a program. In addition, the data were collected in an experiment rather than a quasi-experiment, so internal validity should be higher than it was in Zhao's study.

INFORMATIONAL/TRANSFORMATIONAL AD EFFECTS

In an important article, Puto and Wells (1983) distinguished some years ago between two broad classes of ads: those that are transformational (primarily emotional in their appeal) and those that are informational (primarily rational in their appeal). Subsequent researchers have found that ads in the two classes have quite different effects on viewers. Hitchon and Thorson (1995), for instance, found that ads with a high emotional content (transformational ads) wear out over multiple exposures more slowly than those with a low emotional content (informational ads). In a more elaborate study that built upon this distinction between cognitive and affective ads, Singh and Cole (1993) found that informational ads produce higher ad claim recall than transformational ads across 0, 4, and 8 repetitions but lower brand name recall. Moreover, liking for the commercial increased as the length of the commercial increased from 15 to 30 seconds if the ad were transformational but decreased if the ad were informational. While these studies make it clear that the distinction between informational and transformational advertising should be of interest to advertising theorists and practitioners and that the two kinds of ads have different effects, they do not look at the context effects of these two classes of ads. In our study, we explored these context effects.

HYPOTHESES

The issues treated in this study are encapsulated in three sets of hypotheses that bear on order effects, adjacency effects, and on ad class effects. The serial position effects we hypothesize are expressed in hypotheses 1 and 2 which both focus on ad order.

[H.sub.1]: Ads positioned at the beginning of an ad pod will perform better on measures of ad effectiveness than will other ads.

[H.sub.2]: Ads positioned at the end of an ad pod will perform better on measures of ad effectiveness than will other ads.

Along with order effects, we proposed competing hypotheses on possible adjacency effects. On the one hand, it is possible that ads will perform better if they are placed adjacent to other similar ads. Specifically, informational ads may perform better when placed in a set of informational ads. Likewise, transformational ads may perform better when placed in a set of transformational ads. One might label these hypothesized results a resonance effect. Sharing a single cognitive or affective orientation, the ads in the set might reciprocally enhance ad responses to other ads in the set.

[H.sub.3]: Ads in an ad pod will perform better when they are placed next to other similar ads, i.e., informational ads being placed next to other informational ads and transformational ads being placed next to other transformational ads.

On the other hand, it is possible that the opposite might occur. Ads may perform better when they are placed next to a dissimilar ad, informational ads being placed next to transformational ads and visa versa. This effect would be related to Weber law on stimulus change and might be called a contrast effect since the heightened responses would be evoked by the contrast between the rhetorical styles of adjacent ads.

[H.sub.4]: Ads in an ad pod will perform better when they are placed next to other dissimilar ads, i.e., informational ads being placed next to transformational ads and visa versa.

Finally, we offer two competing hypotheses on the relative value of the two main classes of ads evaluated in this study--informational and transformational ads. The hypotheses suggest, alternatively, that one or another of the two main classes of ads identified by Puto and Wells (1983) may be more persuasive.

[H.sub.5]: Informational ads are more persuasive than transformational ads.

[H.sub.6]: Transformational ads are more persuasive than informational ads.

METHOD

The subjects in this study were 69 students in introductory business courses at highly selective university in the Midwestern United States. Among these subjects there were 32 females and 37 males. The experiment had a 2 (transformational first\informational first) x 2 (resonance\ contrast) design. In other words, the six ads used in the study were grouped as follows: TTTIII in cell 1, TITITI in cell 2, IIITTT in cell 3, and ITITIT in cell 4, where T and I signify transformational and informational ads. The three transformational ads were for Folgers Coffee, Theragran Vitamins, and the Visa Gold credit card. The three informational ads were for a GE diswasher, a Dogdge pickup truck, and for Advil pain reliever. These six ads were selected from a set of twelve ads shown to 30 student subjects in a pilot study. The subjects rated each ad on two scales, one measuring the degree to which it was informational, the other the degree to which it was transformational. The six ads in the study were those which rated highest and lowest on their respective scales. On its own dimension, each set of three ads were rated significantly higher than the alternative set of three ads, a fact that suggests our manipulation of the informational\transformational distinction was adequate.

In each data collection session for the main experiment, subjects were shown a video that began with a 10 minute segment of a 20/20 program from ABC on the marketing of Nike products. One of the four ad pods was then shown, followed by another 20/20 segment on Rollerblades. The 20/20 segments were selected because they were likely to be of interest to college age subjects and because they were consistent with the cover story in the study--that subjects would be quizzed on the topic of marketing ethics. After watching the program segments and the ad pod, subjects filled out a questionnaire that included items on product recall, brand recall, product recognition, brand recognition, ad claim recall, and brand attitude. All recall and recognition items except for ad claim recall were coded 1 if correctly recalled/recognized, 0 if not recalled/recognized. So numbers reported below on brand/product recall/recognition represent the percentages of the products correctly identified. The measure for ad claim recall was the number of ad claims correctly recalled. Brand attitudes were measured on a multi-item 5 point scale that ranged from 1, dislike very much, to 5, like very much. Cronbach's alpha for the attitude scales ranged from .95 to .96.

RESULTS

To test [H.sub.1] (the primacy hypothesis) and [H.sub.2] (the recency hypothesis), linear and quadratic contrasts in within-subjects ANOVAs were used. The tests compared the performance of ads in first through sixth positions in the ad pod on each of the six dependent variables. In the linear test, [H.sub.1]-ads positioned at the beginning of an ad pod will perform better on measures of ad effectiveness than will other ads--was unambiguosly supported across the first five dependent variables and ambiguously on the sixth: product recall (F = 8.420, p < .005), brand recall (F = 8.257, p < .005), product recognition (F = 7.174, p < .009), brand recognition (F = 6.831, p < .011), ad claim recall (F = 5.244, p < .026), and brand atttitudes (F = 23.284, p < .000). The means for the several dependent variables used to test H1 are reported in Table 1. On all the recall and recognition items, the first ad performed better than any other ad. On the sixth measure, brand attitude, the significant linear contrast supports the recency hypothesis rather than the primacy hypothesis. However, as reported below, the quadratic contrast was also significant, and that supported the hypothesis of a primacy effect.

Since the primacy effect was stronger than the recency effect on all dependent measures except brand attitude, [H.sub.2]--ads positioned at the end of an ad pod will perform better on measures of ad effectiveness than will other ads--was not supported by the linear contrasts with the exception of the last one, as discussed above. It was, however, somewhat ambiguously supported in the quadratic contrasts which were also significant across all six dependent variables: product recall (F = 7.976, p < .006), brand recall (F = 10.534, p < .002), product recognition (F = 19.880, p < .000), brand recognition (F = 4.191, p < .045), ad claim recall (F = 7.044, p < .010), and brand attitudes (F = 7.714, p < .007). An inspection of the means will reveal that the final ad performed better than at least one ad in the middle of the pod. This effect was strong enough to make all quadratic tests significant.

To test [H.sub.3] (the resonance hypothesis) and [H.sub.4] (the contrast hypothesis), the performance of ads in the first two between-subjects conditions TTTIII and IIITTT was compared with the performance of those in last two conditions TITITI and ITITIT. Across all six dependent variables, the results of this test were not significant. So [H.sub.3]--ads in an ad pod will perform better when they are placed next to other similar ads, i.e., informational ads being placed next to other informational ads and transformational ads being placed next to other transformational ads--was rejected. [H.sub.4]--ads in an ad pod will perform better when they are placed next to other dissimilar ads, i.e., informational ads being placed next to transformational ads and visa versa--was likewise rejected. Informational and transformational ads do not appear to produce resonance or contrast effects within ad pods.

To test [H.sub.5] (the information superiority hypothesis) and the competing hypothesis, [H.sub.6] (the affect superiority hypothesis), the performance of the three informational ads was compared with the performance of the transformational ads across all six dependent measures. On the product and brand recall and recognition measures, there were no significant differences in the responses to the transformational and informational stimuli. Thus, the class of the ad does not seem to affect basic recall and recognition. However, unsurprisingly, subjects were able to recall more ad claims on average from the informational ads (which contained more claims) than from the transformational ads (F = 11.103, p < .001). What is more surprising is that subjects were also more persuaded by the rational than by the emotional appeals (F = 22.377, p < .000). These results support H5 and lead to a rejection of H6. Means for these tests are reported in Table 2

MANAGERIAL IMPLICATIONS

Our study has several important implications for marketing managers. The support for the primacy and recency hypotheses suggests that various positions in an ad pod should be differently priced, slots at the beginning and end of an ad pod costing more than those in the middle. At present, prices do not differ, so marketing managers may be able to secure a competitive advantage if they can persuade advertisers to place their ads in initial or final position within a pod.

Our study also suggests that at least for ads to which consumers are likely to have only one or two exposures, informational ads are likely to be more persuasive than transformational ads. While transformational ads receive the bulk of the attention at award time and are more likely to be enjoyed by the public, our results suggest that informational ads may be more effective in actually selling products.

In applying the results of this study, practitioners should keep in mind several limitations. Since the subjects for the study were drawn from a student pool at a highly selective liberal arts university, they may have a higher need for cognition than the typical viewer of a television advertisement. Consequently, they may be more inclined to respond well to informational appeals. The study also contained one confound not entirely controlled by the design. Ads were tested in only three or four of the six possible positions within the ad pod. This means that the order effects were confounded with the effects of particular brands and ad executions to some degree. This problem is not as serious as that in the study by Zhao discussed above since ads were rotated through multiple positions, but it is an important limitation of the study nonetheless.

REFERENCES

Edell, J. A. & M. C. Burke (1987). The power of feelings in understanding advertising effects, Journal of Consumer Research, 14 (3), 421-433.

Goldberg, M. E. & G. J. Gorn (1987). Happy and sad tv programs: How they affect reactions to commercials, Journal of Consumer Research, 14 (3), 387-403.

Hitchon, J. C. & E. Thorson (1995). Effects of emotion and product involvement on the experience of repeated commercial viewing, Journal of Broadcasting & Electronic Media, 39 (3), 376-389.

Kerr, J., G. Ward & S. E. Avons (1998). Response bias in visual serial order memory, Journal of Experimental Psychology: Learning Memory, and Cognition, 24 (5), 1316-1323.

Pieters, R. G. M. & T. H. A. Bijmolt (1997). Consumer memory for television advertising: A field study of duration, serial position, and competition effects, Journal of Consumer Research, 23 (4), 362-372.

Puto, C. P. & W. D. Wells (1983). Informational and transformational advertising: The differential effects of time, Advances in Consumer Research, 9, 638-643.

Singh, S. N. & C. A. Cole (1993). The effects of length, content, and repetition on television commercial effectiveness, Journal of Marketing Research, 30 (1), 91-104.

Zhao, X. (1997). Clutter and serial order redefined and retested, Journal of Advertising Research, 37 (5), 57-73.

Stacey Manning, Truman State University

Val Larsen, Truman State University

TABLE 1

 Product Brand Product
 Recall Recall Recognition

First Ad .71 .65 .94
Second Ad .44 .41 .81
Third Ad .29 .29 .71
Fourth Ad .45 .36 .72
Fifth Ad .45 .39 .66
Sixth Ad .38 .38 .84

 Brand Ad Claim Brand
 Recognition Recall Attitude

First Ad .85 1.16 3.62
Second Ad .81 .92 3.34
Third Ad .71 .60 3.63
Fourth Ad .72 .89 3.51
Fifth Ad .63 .64 3.89
Sixth Ad .73 .83 4.09

TABLE 2

 Product Brand Product
 Recall Recall Recognition

Transformational 1 .75 .68 .89
Transformational 2 .30 .23 .67
Transformational 3 .32 .33 .74
Informational 1 .56 .46 .90
Informational 2 .55 .49 .72
Informational 3 .39 .29 .77

 Brand Ad Claim Brand
 Recognition Recall Attitude

Transformational 1 .88 1.06 3.37
Transformational 2 .67 .29 3.07
Transformational 3 .77 .90 3.88
Informational 1 .78 1.40 3.81
Informational 2 .78 .97 3.80
Informational 3 .89 .59 4.10