Issues Surrounding the Examination of Delay Effects in Advertising

J. Wesley Hutchinson, University of Florida
Danny L. Moore, University of Florida
ABSTRACT - This paper considers issues surrounding the persistence of cognitive and affective responses to advertising. A memory-based approach to persistence is outlined and some of the conceptual and methodological implications of this approach are discussed. Special attention is given to the relative contributions of encoding and retrieval factors over time.
[ to cite ]:
J. Wesley Hutchinson and Danny L. Moore (1984) ,"Issues Surrounding the Examination of Delay Effects in Advertising", in NA - Advances in Consumer Research Volume 11, eds. Thomas C. Kinnear, Provo, UT : Association for Consumer Research, Pages: 650-655.

Advances in Consumer Research Volume 11, 1984      Pages 650-655

ISSUES SURROUNDING THE EXAMINATION OF DELAY EFFECTS IN ADVERTISING

J. Wesley Hutchinson, University of Florida

Danny L. Moore, University of Florida

ABSTRACT -

This paper considers issues surrounding the persistence of cognitive and affective responses to advertising. A memory-based approach to persistence is outlined and some of the conceptual and methodological implications of this approach are discussed. Special attention is given to the relative contributions of encoding and retrieval factors over time.

INTRODUCTION

Consumers typically do not purchase a product immediately following exposure to advertising. Instead, there is usually some interval separating advertising from the purchase situation. During this interval associations between the product and ad become weaker, and consequently, the likelihood that information or preferences acquired from an advertisement will be salient in the purchase situation should decline as the amount of time since exposure increases. It follows from this line of reasoning that the decay of advertising effects over time should be related to the rate at which information is forgotten.

Despite the intuitive appeal of a memory-based approach to advertising, the causal role of memory in determining the persistence of communication effects is unclear. Two streams of research are relevant to this issue. First, persuasion research indicates that the learning and forgetting of message arguments is only weakly related to the persistence of attitudinal reactions (see Cook & Flay, 1978, and Sawyer & Ward, 1979, for reviews of this literature). This finding does not rule out memory as an important component of communication effects, however. Prior research on memory and persuasion has failed to distinguish recalled arguments that are accepted from rejected arguments. Hence, it is possible that the greater the proportion of acceptable message arguments recalled, the more positive attitudinal reactions. The second tradition relates memory to purchase behavior and emphasizes the importance of information accessibility in product choice. In this case, information acquired through advertising is expected to influence purchase behavior to the extent that it is accessible during purchase decisions (Bettman, 1979; Biehal & Chakravarti, 1983).

While these two traditions focus on slightly different problems, they point to a set of related issues concerning time-related phenomena in advertising. Foremost among these issues is whether memory for advertising information mediates the effects of marketing communications on brand attitude and purchase behavior. That is, does the type of information recalled during brand evaluations or purchase decisions influence or simply covary with attitudes and choice? A second related issue is whether there are different forgetting rates associated with various types of advertising information. For example, important attributes of a product may be forgotten more slowly than unimportant attributes, and copy-points may be remembered but the source of these points (i.e., advertising) may be forgotten. Finally, Wright & Barbour (1975) note that effectiveness in advertising may depend upon providing information that is congruent with consumers' decision strategies. This suggests that a further factor affecting the delayed effects of advertising is the degree of overlap between the context in which information is learned and the strategies consumers use to make purchase decisions. In short, there seem to be two key issues concerning the role of memory for marketing communications and purchase behavior. First, how does the differential forgetting of information influence advertising effects, and secondly, what effect do cues in the retrieval environment have upon the accessibility of advertising originated information?

A MEMORY-BASED APPROACH TO THE TIME-RELATED EFFECTS OF ADVERTISING

Advertising carry-over effects have received considerable attention in prior research (Sawyer & Warde 1979). We do not intend to provide an exhaustive review of the relevant research. However, it is worth noting that the memory-related research in this area has focused primarily on events that occur during the encoding or learning of information. Sawyer and Ward (1979), for example, review verbal learning literature and conclude that if learning for two different pieces of information is equal, then there should be equal rates of forgetting. They also conclude that events occurring immediately prior to, during, and immediately following exposure to a communication are extremely important in determining the rate of forgetting. Although we would agree with these conclusions, current memory research points to the importance of the retrieval environment in determining the accessibility of information (e.g., Anderson & Pichert, 1978; Hasher & Griffin, 1978). Thus, the memory-based approach advocated in this paper shares ma .v characteristics with other approaches; however, it differs in that interactions are expected between encoding and retrieval contexts.

There are two working assumptions underlying our approach to time-related phenomena in advertising. First, exposure to advertising is expected to result in multiple memory traces that are associated with each other and with previous related traces. This implies that advertising does not produce a unitary trace, and hence, simply determining whether or not consumers remember being exposed to an advertisement for a particular brand is not sufficient to assess the communicative value of the ad. Tentatively, we assume that the types of traces produced by advertising include, but are not limited to, ad specific traces (e.g., affective tone of the ad, medium, etc.), brand related information (e.g., affect toward the brand, visual representations of the product, and product attributes), and recipient-generated thoughts (cognitive responses). Each trace may be characterized by its strength and a forgetting parameter. The strength of a memory trace should increase with repetition and with the amount of elaborative processing, i.e., with the number of connections formed between prior information memory and the available advertising information (Anderson, 1983; Craik & Tulving, 1975). Once a memory trace is placed in long-term memory, it is not expected to be lost; however, the strength of the trace should decay over time. Thus, for each memory trace resulting from exposure to advertising there is a decay function characterizing the strength of the trace at any given time.

The second working assumption is that the influence of encoding and retrieval processes on accessibility may differ for each trace. Encoding of advertising may occur in varying contexts. Two likely sources of encoding differences are the level of involvement and the processing goals of the consumer. In high involvement situations more effort may be devoted to processing message arguments and relating these arguments to prior knowledge (Krugman, 1965; Petty & Cacioppo, 1983). In low involvement situations, attention may be directed toward non-message information such as the source of the communication (Petty & Cacioppo, 1983). Note that this does not imply that consumers do not encode product information when involvement is low. Instead, the effort devoted to processing message versus non-message information, and the manner in which this information is encoded may differ between low and high involvement situations. Petty and Cacioppo (1983) report that post-communication attitude responses for high involvement subjects are influenced more by the quality of message arguments than source factors and that the reverse pattern of response occurs for low involvement subjects. It is interesting to- note that even though recall rates were low (1.75 out of 5 arguments were recalled) in Petty and Cacioppo's study, there was no difference between the low and high involvement groups on the recall measure.

The processing objectives of consumers at the time of encoding should also influence the manner in which the information is stored in memory. Lichtenstein and Srull (in press) find that correlations between recall and brand evaluations differ depending upon whether print ads are processed in order to arrive at an evaluation of the product or to simply comprehend information contained in the ad. Presumably, when evaluations are formed during exposure, subsequent evaluations may be made by retrieving prior evaluations (cf. Lingle & Ostrom, 1981). However, when no evaluation is formed during exposure, a later request to evaluate the product should correspond to information that is retrieved from memory. Thus, the relationship between recall of advertising information and attitude may depend upon the processing objectives of consumers during exposure to an ad. Lichtenstein and Srull's findings, in conjunction with research on involvement, suggest that both processing objectives and level of involvement may influence the future accessibility of information. Involvement should be closely tied to the extent and depth to which message arguments are processed, and consumer processing objectives may influence the storage of information, i.e., whether or not information is stored as evidence for some overall evaluation or simply as associations to the brand.

The critical aspect of our second working assumption is that the retrieval environment should interact with encoding to determine what information is retrieved from memory. Overall, we expect that information that is highly context dependent should show higher levels of recall when cues that are present during encoding are also present during retrieval (Tulving & Thompson, 1973). At this point it is useful to note that the three types of information alluded to earlier, i.e., ad specific, product related, and recipient-generated thoughts, differ with respect to their dependency on context cues. Ad specific information, e.g., the source of the message, is likely to be highly dependent upon the overlap between encoding and retrieval cues. However, if product related information and recipient-generated thoughts are stored as associations to the brand more frequently than to the ad, reinstating the cues available at encoding may be less important for recall of brand information than in the case of ad specific information. Hence, we expect the information recalled during product evaluations and purchase decisions to depend not only on the strength of memory traces but also on the overlap between encoding and retrieval contexts. This distinction is important since the cues used to elicit recall of advertising information may selectively activate information in memory and alter brand evaluations.

In sum, the memory-based approach to time-related phenomena we are advocating stipulates that the accessibility of memory traces resulting from advertising depend upon two factors: 1) The strength of the trace and 2) the extent to which cues at retrieval are linked to a particular memory trace. The remaining sections of this paper discuss the methodological and conceptual implications of these assumptions for research on time-related phenomena in advertising.

CONCEPTUAL AND METHODOLOGICAL ISSUES

Delay effects can arise from at least two qualitatively different types of sources. First, there are many external events that can intervene between ad exposure and brand evaluation. These would include subsequent advertising for the same or competing brands, corrective advertising and word-of-mouth. In contrast to these external sources which all involve the assimilation of new information, there are various internal sources which mainly affect the accessibility of information that is (or was) in memory. These include the various factors that have been shown to be influential at the time of encoding (i.e., at or near the time of exposure), as well as those that exert there influence mainly at the time of retrieval (i.e., brand evaluation). It is these latter sources that we will focus on here.

Most previous theorizing and empirical research on what we are calling internal factors has centered on encoding phenomena. In their comprehensive review, Sawyer and Ward (1979), for instance, conclude that "by far the most important variable in the persistence of learning is the level of the original -learning itself," and.

Cognitive processes at the time of exposure and immediately following exposure are likely to be extremely important factors affecting persistence. These processes may include rehearsal of counterarguments, supports arguments, source derogation, curiosity, and connections to personal life.

Our interpretation of the literature is, for the most part, in agreement with theirs, and we are quite sympathetic to their approach, particularly with regard to their attention to cognitive processes. However, we believe that the approach can be profitably extended. There are significant phenomena that operate well after exposure and are relatively independent of the level of initial learning; and, these have rather clear implications about the persistence of persuasion and advertising effectiveness. Here, we concentrate on two types of phenomena -- differential forgetting curves and latent events.

Differential Forgetting Curves

There are three qualitatively different aspects of persistence -- initial degree of learning, rate of forgetting and the final, or asymptotic, level of retention. In the context of memory phenomena, all three have been shown to be important. It is clear, however, that neither rate of forgetting nor asymptotic retention are completely determined by initial learning (or any other aspects of encoding), rather interference from subsequent traces (during the retention interval, and particularly at retrieval) is the most likely cause of forgetting. Initial learning does affect the resistance to interference, but so do other phenomena (the simplest example is retroactive interference; see Crowder, 1976, Chapter 8, for an excellent review of interference theory). A particularly likely determinant of the amount of interference in ad recall, for example, is ad clutter and degree of competitive advertising.

In the attitude context, the distinction between initial learning, rate of forgetting and asymptotic retention are also important; particularly when memory processes are specifically hypothesized. Although rate of forgetting is often mentioned (e.g., information about the source decays faster than message arguments, which in turn decay faster than conclusions -- thus accounting for some sleeper effects), typically the level of retention or persuasion after some period of delay, not rate of forgetting or asymptotic retention, are measured.

The general methodological and theoretical importance of distinguishing between the three aspects of persistence is best illustrated with a specific mathematical example. Perhaps the most frequently used forgetting curve in the marketing literature is the exponential decay -- which is typically implemented in discrete format as a simple percent retained per time period model (e.g., Little and Lodish, 1969; Zielske, 1959; Zielske and Henry, 1980). In Figure 1 we have used a similar function (given below) to demonstrate the effects of initial learning (IO), rate of forgetting (b) and asymptotic retention (a) on retention (I ). Thus retention at time t is defined as a three parameter, recursive function of retention at time t-1, which implies an exponential decay.

It = a + b(It-1 - a)    (1)

FIGURE 1

DIFFERENTIAL FORGETTING CURVES

Panel A of Figure 1 illustrates the genera] shape of this type of forgetting curve. Panels B, C and D each shown the relative changes that result when two curves differ in one of the three parameters. In each case, one of the curves dominates the other in that the values of one curve are greater than the values of the other, regardless of time. That is, the amount of retention associated with one of the curves is greater than (or equal to) the retention associated with the other curve, regardless of the time period that is examined. (The sets of curves do differ in terms of how they converge and diverge over time; however, such aspects are beyond the scope of the present discussion.) In such situations a single measurement of retention after a delay of any magnitude would suffice to establish which curve was the greater; and, it makes some sense to refer to this as greater persistence. Panels E, F, G and H illustrate situations where measurement and interpretation are not so simple. Here two or more of the parameters are permitted to vary and, in these cases, relative dominance is a function of time. Obviously, there is no single time period in which measurements of the two curves will indicate fully the nature of further (or previous) forgetting. Thus, the conclusions about persistence that have been offered based on single (or even multiple) delayed measured have implicitly assumed certain constraints on initial learning, rate of decay and asymptotic retention. Several options for future research are available. First, forgetting curves can be tracked over time. This option is most natural for panel studies or large studies in which sample size is large enough to be distributed over many separate measurement delays. Unfortunately this option is too costly, or methodologically infeasible for many purposes. A second option is to measure retention for several delays and then fit the data to some plausible function such as Equation 1. In this case, the estimated parameters would be assumed to measure the three aspects of persistence. Potential problems could arise, however, if different functional forms yielded qualitatively different sets of parameters. Perhaps the simplest option is to seek independent support for the hypothesis that only one parameter was likely to vary between conditions. For instance, it is possible to match various conditions with respect to initial learning and bring other evidence to bear about asymptotic retention In some cases it may be possible to argue that all conditions have the same (e.g., zero) asymptote. When this is possible, a single delayed measure suffices.

Latent Events

In addition to the relatively continuous process of forgetting that was discussed in the previous section, there is considerable evidence that discrete internal events can also have a significant impact on retention. We refer to these as latent events since their efficacy is in some sense continually present, but a "triggering" event is required for their potential effects to be actualized. A familiar example of such an event can be found in the dissociative cue hypothesis. Under this hypothesis, some source factor such, as low credibility, prevents a message argument from having a persuasive impact. If the information about source credibility is dissociated from the message while the message itself is retained (the latent event), then the persuasiveness of the message argument is suddenly actualized. This is quite similar to the Mutual ExcLusion Hypothesis discussed by Moore and Hutchinson (in press) in which retention of certain ad information prevents any (ad-induced) increment in brand name familiarity from having a beneficial effect on brand attitude.

Another type of latent event that is particularly interesting is the shifts of perspective, or frame of reference that Krugman (1965) suggested as a potential mechanism for the effects of television advertising. Krugman hypothesized that repeated exposure to advertising slowly shifts the relative salience and perceived importances of the various product attributes. Furthermore, he asserted that "... the full perceptual impact (of these shifts) is delayed," and that "... the purchase situation is the catalyst that reassembles or brings out all the potentials for shifts in salience that have accumulated up to that point." The results of relatively recent cognitive research on the role of importance in text comprehension and memory makes Krugman's speculations especially pertinent.

The effects of perspective and frame of reference on information processing have come to be referred to as schematic aspects of memory (see Alba and Hasher, 1983). Generally speaking, a schema is higher order structuring of knowledge that defines what types of information are important in a particular context. They are meant to include, but are not limited to, the relative saliences of object attributes -- in this case, product attributes. A wide variety of empirical studies have supported the hypothesis that schematic knowledge guides the encoding of new information (e.g., Bransford and Johnson, 1973) and affects the retrievability of retained information (e.g., Anderson and Pichert, 1978; Hasher and Griffin, 1978). Essentially, these studies have shown that ideas that are important within the context of an activated schema will exhibit enhanced recall.

Some schema effects have been shown to operate primarily, or exclusively, at encoding. That is, when schema activation is operationalized and experimentally manipulated, only schema activated at or before encoding affect recall (e.g., Bransford and Johnson, 1973). These schema effects rely on manipulations in which text is difficult or impossible to comprehend unless the appropriate schema is activated. Other studies, however, have demonstrated that schemas can be successfully "switched" at any point between encoding and retrieval, given that the text was comprehensible (e.g., Anderson and Pichert, 1978; Hasher and Griffin, 1978). for example, Anderson and Pichert had subjects read a brief narrative about what two boys did at one of the boy's homes while they were skipping school. Initial instructions directed subjects to adopt either the perspective of a homebuyer or, alternatively, a burglar. After a brief, task-filled delay recall was for story ideas was measured. After this first recall subjects were asked to attempt a second recall using either the same or the alternate perspective. The change of perspective instructions resulted in incremental recall of information relevant to the new perspective. Hasher and Griffin used story titles to manipulate schema activation and showed that post-encoding schema changes affected recall even when the schema manipulation and recall were separated by as much as a week.

It would seem that these results indirectly support Krugman's conjecture that the purchase situation might activate latent effects of advertising. Specifically, if advertising shifts the saliences of various attributes, this is tantamount to changing the schema that is likely to be activated during decision making. Moreover, the schema shift will also guide the encoding and subsequent recall of information that is acquired after the shift. It is not necessary to agree with Krugman's theory of advertising effectiveness, however, to appreciate the broad implications of schema shift phenomena for consumer behavior. Clearly, the schemas activated during ad exposure can be very different from those activated during the purchase decision, regardless of whether, or not, the ads themselves have had an impact on attribute salience. Thus, copy test procedures may systematically underestimate or give a biased estimate of the accessibility of product information presented in ads. This would be expected to be particularly true when the schemas activated by the ad do not assign great importance to product information.

IMPLICATIONS AND CONCLUSIONS

The foregoing discussion implies that the impact of persuasive communications on attitudes and purchase behavior may differ as a function of time. Clearly, this is not a particularly novel idea. Silk and Vavra (1974) argue that the probability that affective reactions to advertising are spontaneously associated with a brand should decline with time. If so, then a "sleeper" effect may occur such that irritating ads produce greater attitude change after some delay than immediately following exposure. The critical assumption in Silk and Vavra's work is that the strength of associations between a brand and message context (e.g.. liking or disliking for the ad, source credibility) decays faster than the strength of associations between the brand and message content (e.g., product attribute claims, visual representations of the product, etc.). Ray (1973) and Moore and Hutchinson (in press) have advanced similar hypotheses concerning differential forgetting curves. Sawyer and Semenik (1978) examine the effects of differential forgetting on the impact of corrective advertising. They propose that corrective ads may have an immediate effect on beliefs in a deceptive advertising claim; over time belief in the deceptive claim may recover because asymptotic recall of the deceptive claim is greater than that for disclaimers. Thus, research on both sleeper effects and corrective advertising points to the possibility that the immediate and delayed effects of advertising on beliefs. attitudes, and purchase behavior may differ.

Although there seems to be general agreement that time since exposure is an important component of marketing communication effects, few attempts have been made to specify the psychological processes mediating time-related phenomena. Sawyer and Ward (1979) review several hypotheses that explain delayed effects of persuasive communications by reference to various memory phenomena. Unfortunately, there is little empirical work that allows one to assess the adequacy of these explanations. Moreover, most memory-based explanations have-focused on differential forgetting or encoding phenomena. In this respect, our discussion of internal events and retrieval effects adds a new perspective to pre-existing speculations Specifically, the cues that are available during purchase and product evaluation may selectively activate information stored in memory. This, in turn, may be sufficient to produce differences between immediate and delayed measures of advertising effectiveness.

Regardless of the enthusiasm we have expressed for a memory-based approach to time-related phenomena, the cause-effect relationships between memory, attitudes, and purchase decisions are largely uncharted. Attitudes, evaluations, and inferences seem to persist even though the information producing such reactions is not accessible (Lichtenstein & Srull, in press; Lingle & Ostrom, ]981). Moreover, the ability to recall the content of a persuasive communication does not seem to predict attitudes (Greenwald, 1968; Petty & Cacioppo, ]983). Instead, the number of favorable cognitive responses recalled relative to the number of unfavorable responses appears to be correlated with attitudes (Greenwald, 1968). Of course, our discussion of the memory-based approach applies equally to cognitive responses and message content.

Given the equivocal relationships between memory and persuasion, it might seem premature to postulate that the persistence of advertising effects depends upon the accessibility of ad-related information. There are, however, conditions under which the ability to recall advertising information should play an important causal role. When attitudes or evaluations are not formed during exposure to advertising or when uncertain, weak opinions are produced by an ad, the impact of memory on attitudes and purchase decisions should increase (Lichtenstein & Srull, in press). Krugman (1965), and more recently Smith Swinyard (1983), hypothesize that advertising often elicits minimal affective reactions to a brand either because consumers are not involved or because they tend to discount advertising claims. Even when attitudes are formed during exposure to advertising, it is possible that they can be modified by retrieval cues. For example, during exposure to a wine ad, a person might evaluate the brand negatively as a table wine. However, if that person later shops for a cooking wine, the new perspective may facilitate recall of product attributes that were unimportant (or negative) from the first perspective.

In summary, we have proposed two memory processes that may mediate carry-over effects in advertising: 1) differential forgetting, and 2) latent events. Methodologically, these processes imply that precisely when and how advertising effectiveness measures are collected is critical. Immediate measures of product evaluations or purchase intentions may be based on different considerations than delayed measures. Moreover, the cues available at any given measurement delay should be important in determining the precise information recalled Finally, the memory processes we h.eve discussed point to several interesting areas of research. In particular, it would be useful to estimate forgetting parameters for various types of ad information and for affective and cognitive responses to ads. Also, it remains to be seen whether, or not, the notion of schema shifts can be operationalized in a way that is meaningful in the advertising context.

REFERENCES

Alba, J. & Hasher, L. (1983) Is memory schematic? Psychological Bulletin, 93, 203-221.

Anderson, J. R. (1983) A Spreading Activation Theory of Memory. Journal of Verbal Learning and Verbal Behavior, 22, 261-295.

Anderson, R. C. & Pichert, J. W. (1978). Recall of previously unrecallable information following a shift i perspective. Journal of Verbal Learning and Verbal Behavior. 17. 1-12.

Bettman, J. R. (1979). An information processing theory of consumer choice. Reading MA: Addison-Wesley Publishing Company.

Biehal, G., & Chakravarti, D. (1983). Information accessibility as a moderator of consumer choice. Journal of Consumer Research, 10, 1-14.

Bransford, J. D. & Johnson, M. K. (1973). Considerations of some problems of comprehension. In W G. Chase (Ed.), Visual Information Processing. New York: Academic Press. 1973.

Cook, T. D., & Flay, B. R. (1978). The persistence of experimentally induced attitude change. In L. Berkowitz (Ed.), Advances in Experimental Social Psychology: Volume 11 (pp. 1-57). New York, NE: Academic Press.

Craik, F. I. M. & Tulving, E. (1975). Depth of processing and the retention of words in episodic memory. Journal of Experimental Psychology: General, 104, 268-294.

Crowder, R. G. (1976). Principles of Learning and Memory. Hillsdale, NJ: Lawrence Erlbaum.

Dutta, S., & Kanungo, R. N. (1967). Retention of affective material: A further verification of the intensity hypothesis. Journal of Personality and Social Psychology, 5, 476-481.

Hasher, L., & Griffin, M. (1978). Reconstructive and reproductive processes in memory. Journal of Experimental Psychology: Human Learning and Memory, 4, 318-330.

Krugman, H. E. (1965). The impact of television advertising: Learning without involvement. Public Opinion Quarterly, 29 (Fall), 349-356.

Lichtenstein, M & Srull, T. K. (in press). Conceptual and methodological issues in examining the relationship between consumer memory and judgment. In L. F. Alwitt 2 A. A. Mitchell (Eds.), Psychological Processes and Advertising Effects: Theory, Research, and Application. Hillsdale. NJ: Lawrence Erlbaum Associates.

Lingle, J. H., & Ostrom, T. M. (1981). Principles of memory and cognition in attitude formation. In R. E Petty, T. M. Ostrom, & T. C. Brock (Eds.), Cognitive Responses in Persuasion, Hillsdale, NJ: Lawrence Erlbaum Associates.

Little, 3. D. C. & Lodish, L. M. (1969). A media planning calculus. Operations Research 17, 1-35.

Moore, D. L., & Hutchinson, J. W. (in press). The influence of affective reactions to advertising: Direct and indirect mechanisms of attitude change. In L. F. Alwitt & A. A. Mitchell (Eds.), Psychological Processes and Advertising Effects: Theory, Research, and Application. Hillsdale, NJ: Lawrence Erlbaum Associates.

Petty, R. E., Cacioppo, J. T., & Schumann, D. (1983). Central and peripheral routes to advertising effectiveness: The moderating role of involvement. Journal of Consumer Research. 10. 135-147.

Ray, M. L. (1973). Psychological theories and interpretations of learning. In S. Ward & T. S. Robertson, Consumer Behavior: Theoretical Sources (pp. 45-117). Englewood Cliffs, NJ: Prentice-Hall, Inc.

Sawyer, A., & Semenik, R. J. (1978). Carryover effects of corrective advertising. In H. K. Hunt (Ed.), Advances in Consumer Research (pp. 343-351), Vol. 5, An Arbor, MI: Association for Consumer Research.

Sawyer, A., & Ward, S. (1979). Carry-over effects in advertising communication. In J. N. Sheth (Ed.), Research in Marketing (pp. 259-314). Greenwich, CT: JA Press Inc.

Silk, A. J., & Vavra, T. G. (1974). The influence of advertising's affective qualities on consumer response. In D. Hughes & M. L. Ray (Eds.), Buyer/Consumer Information Processing |PP. 157-186). Chapel Hill, NC: University of North Carolina Press.

Tulving, E. & Thomson, D. M. (1973). Encoding specificity and retrieval processes in episodic memory. Psychological Review, 80, 352-373.

Zielske, H. A. (1959). The remembering and forgetting of advertising. Journal of Marketing, 23, 239-243.

Zielske, H. A., & Henry, W. A. (1980). Remembering and forgetting television ads. Journal of Advertising Research, 20 (April), 7-13.

Wright, P. L., & Barbour, F. (1975). The relevance of decision process models in structuring persuasive messages. Communication Research, 2 (July), 246-259.

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