Effects of Variation in Message Execution on the Learning of Repeated Brand Information

Robert E. Burnkrant, The Ohio State University
Hanumantha R. Unnava, The Ohio State University
ABSTRACT - The-research reported here examines the effects of varying the executions to which subjects are exposed when presented with three ads for a brand of liquor. Based on the encoding variability hypothesis it is proposed that presenting three different executions of an ad will result in greater brand recall than presenting the same execution three times. The results support this hypothesis. Other results suggest that these effects were not due to differential attention levels. Media scheduling implications and future research possibilities are discussed.
[ to cite ]:
Robert E. Burnkrant and Hanumantha R. Unnava (1987) ,"Effects of Variation in Message Execution on the Learning of Repeated Brand Information", in NA - Advances in Consumer Research Volume 14, eds. Melanie Wallendorf and Paul Anderson, Provo, UT : Association for Consumer Research, Pages: 173-176.

Advances in Consumer Research Volume 14, 1987      Pages 173-176

EFFECTS OF VARIATION IN MESSAGE EXECUTION ON THE LEARNING OF REPEATED BRAND INFORMATION

Robert E. Burnkrant, The Ohio State University

Hanumantha R. Unnava, The Ohio State University

ABSTRACT -

The-research reported here examines the effects of varying the executions to which subjects are exposed when presented with three ads for a brand of liquor. Based on the encoding variability hypothesis it is proposed that presenting three different executions of an ad will result in greater brand recall than presenting the same execution three times. The results support this hypothesis. Other results suggest that these effects were not due to differential attention levels. Media scheduling implications and future research possibilities are discussed.

INTRODUCTION

A considerable body of literature has examined repetition effects (cf. Sawyer, 1981). Much of this literature has varied, in an experimental design, the number of exposures people receive. It has typically been found that increasing the number of exposures increases learning up to a point but beyond that point wearout occurs (e.g., Calder and Sternthal 1980).

Much less attention has been focused on the effects of varying the executions of repeated communications. For example, repetition studies that employ a 3 exposure condition typically show people the same commercial three times. This is fairly consistent with media scheduling because it is not at all uncommon to see the same commercial repeated on television even within the same half hour program. However, based on an encoding variability hypothesis (e.g., Melton 1967, Madigan 1969) to be discussed below, we would expect more learning to take place when people are exposed to three different executions of a commercial than when they are exposed to the same execution three times.

The objective of this research is to test these implications of the encoding variability hypothesis. In the paragraphs below we will first review encoding variability research and draw out its implications for scheduling repeated exposures to commercials. We will then generate and test the implications of hypotheses that follow from the encoding variability position.

Encoding Variability

The encoding variability hypothesis holds that when people are exposed to information more than once, the likelihood of retrieving that information subsequently is greater if the material is encoded differently (e.g., the to-be-remembered information is related to different stimuli or leads to different memory traces) on each repetition than if it is repeatedly encoded in the same way. The likelihood of an item of information being retrieved is believed to be directly related to the number of paths or ways in which the item is encoded.

Encoding variability research has proceeded in two different directions. One approach focuses on contextual variability and the other addresses semantic variability. According to the contextual variability position stimuli are said to be encoded and stored in terms of associations established between the various attributes of the word's context including nearby list words, task relevant thoughts, environmental distractions, etc. (cf. Anderson and Bower 1972, 1974). When a subject attempts to remember a word, he retrieves some of the contextual associations formed at encoding. The likelihood of recalling the to-be-remembered word is directly related to the availability of contextual information at the time the subject is tested.

Research showing that increasing the lag between exposures to a stimulus increases recall performance (Melton, 1970) is taken as support for this position. It is assumed that likelihood of contextual variability is directly related to the time interval between exposures.

Other research varies the stimuli associated with the to-be-remembered word. For example, Hintzman and Stern (1978) gave subjects the names of famous individuals to be remembered. The names were presented in brief sentences describing an action attributed to the individual. The sentences that occurred more than once were repeated either three or six times. At each frequency of repetition, the phrase accompanying the name either remained the same or was different each time. An incidental learning procedure was used with subjects required to rate the truth value of each sentence. It was fount that subjects recalled the target words significantly better in the different sentence condition compared to same sentence condition. The authors argued that this was because of the additional paths or memory traces generated in the condition in which the phrase accompanying the to-be-remembered word differed in each repetition.

Glanzer and Duarte (1971) also obtained support consistent with encoding variability using bilingual subjects and bilingual word lists. They found that, under massed conditions, recall was better when words were repeated in a different language than when they were repeated in the same language.

Paivio (1974) varied both time lag between exposures and the stimuli associated with the to-be-remembered stimuli. He argued that pictures repeated as words and words repeated as pictures are more likely to result in variable encoding than presenting the same picture or word twice. It was found that under standard free recall conditions, word-word and picture-picture conditions were less than additive in their effect on recall at zero exposure lag, but became additive at long exposure lags. Picture-word repetitions on the other hand, were additive at zero lag as well as longer lags. Both of these findings are consistent with the encoding variability position.

To summarize, the encoding variability research indicates that when the time lag between exposures is short, pairing a stimulus with two different cues is more effective (i.e., results in more learning) than exposing people to the same stimulus-cue combination twice. It appears that each succeeding repetition of the same stimulus strengthens an existing memory trace by a smaller amount. However, when different cues are paired with the same stimulus. each cue results in a different memory trace. The likelihood of retrieving the stimulus is believed to be directly related to the number of traces involving that stimulus.

Advertising Implications

The research reviewed above has at least two implications for advertising. First, it provides a theoretical explanation for the well known finding that distributed exposure to advertising is superior to massed exposure (e.g., Zielski 1959). Encoding variability research suggests that the superiority of longer delays between exposure to advertising stimuli is due to the increased likelihood that the stimuli will be encoded differently on each exposure occasion.

Second, encoding variability research suggests that if time between exposures to advertising stimuli is controlled or held constant, more learning will take place when different executions of the ads are used than when the same ad is repeated. This is because the variation in executions would be more likely to lead to varied semantic encodings of the repeated information than the straight repetition of the same ad. These varied encodings should make the brand name and other information more accessible subsequently than they would be without variable encoding.

It is important to note at this point that the encoding variability explanation is an alternative to the possibility suggested by others (i.e., Calder and Sternthal 1980) that varied executions may be more effective because people will pay more attention to varied executions than to repetitions of the same execution. The encoding variability hypothesis does not implicate attention. In accordance with this position, support has been obtained for the encoding variability hypothesis while controlling for processing time (D'Agostino and DeRemer 1973).

Calder and Sternthal (1980) did manipulate whether people were repeatedly exposed to the same or different executions of ads for two products. However, no evidence was reported of differential attention to varied versus same executions of the ads, and the execution manipulation had no effect on attitude toward the product. They did find that exposure to varied executions led to more total thoughts than repeated exposure to the same execution. This finding is not inconsistent with the encoding variability position. However, the effect may have been due to the additional information contained in the several executions to which people in the varied execution condition were exposed rather than to the variable encoding of the repeated information.

HYPOTHESES

The preceding discussion leads to the following hypotheses.

First, it is expected that exposing people to three different executions of an ad will lead to greater brand name recall than exposing people three times to any one of the three executions.

Second, it is expected that exposing people to three different executions of an ad will not result in any greater self reported attention to the ads than occurs in the groups exposed three times to the same ad execution.

Both hypotheses one and two follow directly from the encoding variability research discussed above. The exposure to different ad executions should increase encoding variability of the brand name and therefore make the brand name more accessible at the post test than it should be for the groups given three exposures to the same execution. It is expected that this effect will emerge without differential attention to the ads as noted in hypothesis two. This is an important additional hypothesis because in the absence of contradictory evidence, differential attention could be raised as an alternative explanation to the encoding variability explanation maintained here. To help control for attention the Dewars ads were separated by a minimum of five different filler ads, and subjects were instructed to evaluate each ad as it appeared. Exposure time was controlled and set to be sufficient to allow subjects to read all the copy contained in the ads.

Third, it is expected that attitude toward the brand will be greater when people are exposed to three different executions of an ad than when they are exposed to three repetitions of the same execution.

This hypothesis is based on McGuire's (1969) model of information processing which posits a direct relationship between learning (i.e., comprehension) and message acceptance (i.e., yielding). It is expected that material conveyed in the ads employed in this research will have favorable implications for the product. Therefore, similar effects are predicted for both learning (measured by brand recall) and attitude toward the product. While a number of studies have failed to find a direct relationship between learning and attitude (e.g., Petty, Cacioppo and Heesacker 1981) others have found such a relationship (cf. Eagly and Chaiken 1984). It is expected that a direct relationship will be obtained in this research.

METHOD

The hypotheses were tested using a completely randomized (i.e., single factor) experimental design. Seventy four Ohio State University undergraduate business students participated in this research to earn extra credit toward their course grades. All subjects were assigned to one of the four between subjects conditions of the study. In each condition, subjects were asked to evaluate a set of ads that they would be shown. They were then shown a set of 25 slides for 15 different consumer products. Each slide was held on the screen for 25 seconds. The twenty five second exposure time was necessary to allow the students to read the copy contained in the ads.

The critical ads in this experiment were those for Dewar's Scotch. All other ads were fillers. In all cases the Dewar's White Label Scotch ads were the sixth, twelfth and nineteenth of the twenty five slides shown. Filler ads included three repetitions of two other liquor brand ads, three repetitions of two other nonliquor ads and ten other consumer brand ads which were shown once. In condition one subjects saw the 25 slides including three exposures to a single execution (execution A) of an ad for Dewar's White Label Scotch. Subjects in condition two saw all the same slides in the same order except that they were exposed three times to execution B of the Dewar's White Label Scotch ad. Subjects in condition three saw all the same ads in the same position except that they were exposed three times to execution C of the Dewar's White Label Scotch ad. Subjects in condition four were exposed to all the same filler ads, and they were exposed once to each execution of the Dewar's White Label Scotch ad. In this last condition order was counterbalanced so that half the subjects saw A then B then C while the other half saw B then C then A.

The three Dewar's White Label Scotch ads contained the same illustration of the product and the same headline "Dewar's Profile." All three ads referred to the Dewar's brand name in the same way, with the same size print and used the brand name the same number of times. The ads differed only in terms of the person pictured (e.g., Gayle Sayers) and the description of the person's characteristics.

After subjects saw all 25 ads they were given a questionnaire in which they were asked to list the names of all the brands that were advertised. Another later question asked people to list the brands of liquor ads they remember having seen. They were also asked to rate Dewar's Scotch on a set of seven point bi-polar adjective scales. The scales were labeled good-bad, undesirable-desirable, rewarding-punishing. worthless-valuable and positive-negative respectively. Similar scales have frequently been employed to measure attitude (e.g., Ajzen and Fishbein 1980). Their reliability in this research will be assessed in preliminary analysis.

Finally, subjects were asked to respond to each of the following questions addressing attention on 7-point scales varying from strongly agree (1) to strongly disagree (7):

(1) "I paid as much attention to the second and third repetitions of the Dewar's ad as I did to the first,"

(2) "I found the third repetition boring, so I did not pay any attention to it."

(3) "For all ads that were repeated, I did not pay any attention to the second and third repetitions."

RESULTS

The Reliability of the bi-polar scales used to measure attitude toward Dewar's Scotch was assessed. The analysis yielded a coefficient alpha of 0.95. Based on this analysis it was concluded that the scale provides a reliable measure of the evaluation of Dewar's Scotch and mean scores across the five bi-polar adjective scales were calculated for each individual.

For both unaided and aided brand name recall, each subject was given a score of 1 if Dewar's Scotch was recalled and a score of 0 if it was not recalled. The aided and unaided recall data were then transformed using an arc sin transformation to stabilize variances and permit use of analysis of variance to test hypotheses.

Hypothesis Tests

Hypothesis one holds that brand name recall will be greater in the condition in which subjects are exposed once to each Dewar's execution than it will be when people are exposed three times to any one of the executions. This was tested first with unaided brand name recall and then with aided recall. The percent recalling the brand name for each condition is shown in Table 1.

After arc sin transformations were performed the unaided recall data were analyzed using a completely randomized (i.e., single factor) analysis of variance with four levels. The ANOVA yielded an F-ratio of 17.26 (df=3, 70) which was significant (p<0.001). The percentages shown in Table 1 indicate that the effects were in the expected direction. Pairwise contrasts were run using the Newman-Keuls multiple comparison test. The percent recalling Dewar's White Label Scotch in condition 4 (the condition in which people were exposed once to each execution) was significantly higher (p<0.05) than in any other condition. This supports hypothesis 1 with the unaided recall data.

The arc sin transformed aided recall data were also analyzed using analysis of variance. This analysis yielded an F-ratio of 14.86 (df=3, 70) which was significant (p<0.001). Again pairwise contrasts were run using Newman-Keuls multiple comparison test. Condition 4 (the condition in which people were exposed once to each execution) yielded a significantly higher percent correct brand name recall than any other condition. No other contrasts were significant. Again this supports hypothesis 1.

Hypothesis two held that there would be no difference between conditions in self reported attention to the ads. To test this hypothesis, separate analyses of variance were conducted on each attention question (see Table 1 for means). No significant differences were obtained for any of the three attention questions. These results support hypothesis 2 and show that the recall effects occurred in the absence of any differences in self reported attention.

TABLE 1

MEAN SCORES AND PERCENTAGES FOR DEPENDENT VARIABLES BY CONDITION

Hypothesis three was tested by performing an analysis of variance on product evaluation scores (see Table 1). Although the means were in the expected direction, the analysis of variance was not significant. Thus we must reject hypothesis three.

DISCUSSION AND CONCLUSION

The data analyzed here provide consistent support for the contention, based on encoding variability, that varying the executions of the ads to which people are exposed will increase the brand recall that results from that exposure. Failure to obtain significant differences on the attention measure suggests that the recall results are not due to the amount of attention paid to the ad. Thus, the results support the encoding variability position that varying executions will provide a richer network of memory traces involving the brand name and this will increase brand name accessibility.

Brand recall was used here not only because the encoding variability hypothesis makes clear predictions about this variable but also because the number of exposures to brand name was controlled in this experiment. Each execution referred to the brand name the same number of times. Hence it could not be argued that varying the executions biased the results obtained with this variable. However, had we used some other variable, such as the total amount of brand related information recalled, it could be argued that obtained differences were due to exposure of subjects to more information in the varied execution condition. The brand name recall variable employed here is not subject to this criticism. Another unbiased measure that could be used in future research would be recall of just those copy points that are repeated the same number of times in each execution.

It may, perhaps, be argued that we did not obtain significant differences with our attention measures because the measures were not sufficiently sensitive to pick up those differences. While this kind of rationale is always a possibility whenever results support a null hypothesis, we regard this argument as quite speculative. Other types of manipulations have been found to produce significant differences with similar types of self report measures (e.g., Petty and Cacioppo 1979; Petty, Wells and Brock 1976.) Nevertheless, it would be useful in further research to employ more sensitive measures of attention such as the secondary task technique, eye movement observations and/or various physiological measures (cf. Cacioppo, Petty and Shapiro 1983).

It was rather surprising, given the recall results, that no effect of the execution manipulation was obtained for attitude. Because the means were in the expected direction and a relatively small sample size was employed here, future research using larger samples may uncover such an effect. However, the failure to find a significant relationship between recall and attitude is certainly consistent with the contention made by others (e.g., Petty, Cacioppo and Schumann 1983) that there is no necessary relationship between recall and attitude.

The results of this research provide clear implications for media planners. They suggest that when repetitions are planned in close proximity to one another the executions of those repetitions should be varied. This should result in greater brand name recall for the reasons addressed earlier.

This does not necessarily mean, however, that a larger number of executions must be produced. The encoding variability research reviewed earlier indicates that if a sufficient time delay is provided between exposures to the same stimulus, results should be similar to those obtained here for different executions. This is because, as time increases, the context in which the information is taken in and the interpretation of the stimulus are likely to differ from the context and interpretation that occurred the first time the stimulus was presented. This suggests, then, that a media plan involving executions A, B and C would be more effective if scheduled in the pattern of A, B, C, A, B, C, A, B, C rather than the pattern A, A, A, B, B, B, C, C, C. While it remains for future research to establish the superiority of the former pattern, it is consistent with the encoding variability literature and the results of this study.

REFERENCES

Ajzen, I. and M. Fishbein (1980), Understanding Attitudes and Predicting Social Behavior, Englewood Cliffs: Prentice-Hall.

Anderson, J.R. and G.H. Bower (1972), "Recognition and Retrieval Processes in Free Recall," Psychological Review, 79, 97-1-3.

Cacioppo, J.T. and R.E. Petty (1979), "Attitudes and Cognitive Response: An Electrophysiological Approach," Journal of Personality and Social Psychology, 37, 2181-2199.

Cacioppo, J.T., R.E. Petty and D. Shapiro, eds. (1983), Social Psychophysiology: A Sourcebook, New York: Guilford.

Calder, B.J. and B. Sternthal (1980), "Television Commercial Wearout: An Information Processing View," Journal of Marketing Research, 17, p. 173-186

D'Agostino, P.R. and P. DeRemer (1973), "Repetition Effects as a Function of Rehearsal and Encoding Variability," Journal of Verbal Learning and Verbal Behavior, 12, 108-113.

Eagly , A.H. and S. Chaiken (1984), "Cognitive Theories of Persuasion," in Advances in Experimental Social Psychology, Vol. 17, ed. L. Berkowitz, Orlando: Academic Press.

Glanzer, M. and A. Duarte (1971), "Repetition Between and Within Languages in Free Recall," Journal of Verbal Learning and Verbal Behavior, 10, 625-630.

Hintzman, D.L. and L.D. Stern (1978), "Contextual Variability and Memory for Frequency," Journal of Experimental Psychology: Human Learning and Memory, 4, 539-549.

Madigan, S.A. (1969), "Intraserial Repetition and Coding Process in Free Recall," Journal of Verbal Learning and Verbal Behavior, 8, 828-835.

McGuire, W.J. (1969), "The Nature of Attitudes and Attitude Change," in Handbook of Social Psychology, Second Edition, Vol. 3, ed. G. Lindzey and E. Aronson, Reading: Addison-Wesley.

Melton, A.W. (1967), "Repetition and Retrieval From Memory," Science, 158, 532.

Melton, A.W. (1970), "The Situation with Respect to the Spacing of Repetitions and Memory," Journal of Verbal Learning and Verbal Behavior, 9, 596-606.

Paivio, A. (1974), "Spacing of Repetitions in the Incidental and Intentional Free Recall of Pictures and Words," Journal of Verbal Learning and Verbal Behavior, 13, 497-511.

Petty, R.E. and J.T. Cacioppo (1979), "Issue Involvement Can Increase or Decrease Persuasion by Enhancing Message-Relevant Cognitive Responses." Journal of Personality and Social Psychology, 37, 1915-1926.

Petty, R.E., J.T. Cacioppo and H. Heesacker (1981), "The Use of Rhetorical Questions in Persuasion: A Cognitive Response Analysis," Journal of Personality and Social Psychology, 40, 432-440.

Petty, R.E., T. Cacioppo and D. Schumann (1983), "Central and Peripheral Route to Advertising Effectiveness: The Moderating Role of Involvement," Journal of Consumer Re- search, 10, 134-148.

Petty, R.E., G.L. Wells and T.C. Brock (1976), "Distraction Can Enhance or Reduce Yielding to Propaganda: Thought Disruption Versus Effort Justification," Journal of Personality and Social Psychology, 34, 874-884.

Sawyer, A.G. (1981), "Repetition, Cognitive Responses and Persuasion," in Cognitive Responses in Persuasion, ed. RE. Petty, T.M. Ostrom and T.C. Brock, Erlbaum: Hillsdale, New Hersey.

Zielske, H.A. (1959), "The Remembering and Forgetting of Advertising," Journal of Marketing, 23, 239-243.

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