A review and critique of the hierachy of effects in advertising

Page 1 of 13

Published by NTC Publications Ltd Farm Road, Henley-on-Thames, Oxon RG9 1EJ, UK Tel: +44 (0)1491 411000 Fax: +44 (0)1491 571188

Vol 9 No 2 (1990)

A review and critique of the hierarchy of effects in advertising

Thomas E Barry and Daniel J Howard

Industry practitioners and social scientists throughout the world have long sought to determine the effectiveness of advertising in this multi-billion dollar industry. For close to a century, many social scientists have debated just how advertising works. One suggested explanation is the hierarchy of effects - that body of literature that posits that audiences go through a variety of stages, namely cognitive, affective, and conative, in responding to advertising, and other persuasive marketing messages. This paper reviews the hierarchy of effects from its beginning at the turn of this century to present-day developments. The paper also examines the controversies surrounding hierarchy models and provides a critique of the evidence supporting its position as an advertising model.

INTRODUCTION
For close to a century, advertising and marketing researchers and practitioners the world over have diligently sought to understand just how advertising influences buyers' purchase decisions. At stake is the effectiveness of strategies developed in this multi-billion dollar industry. At the core of current understanding is the body of literature referred to as the hierarchy of effects. This literature deals with the way in which target audiences process and ultimately use advertising information to influence product and brand choices and is considered a top priority research area for contemporary marketing and advertising researchers (Schmalensee, 1983). One testimonial to the importance of the hierarchy of effects is the number of diverse communities interested in the topic. Among those interested are advertising and marketing researchers and practitioners, sociologists, communication theorists, cognitive psychologists, social psychologists and others who have debated the realities of advertising effects on consumption behaviour. A second testimonial to the importance of the hierarchy literature is its longevity. The first published suggestion that a hierarchy of effects was operative in marketing communications appeared in 1898 and researchers and practitioners continue to contest the hierarchy notion today. Scores of hierarchy-of-effects models have been proposed. The vast majority of them merely offer changes in nomenclature to the traditional hierarchy-of-effects model which hypothesizes that audiences respond to messages in a cognitive, affective, and conative (behavioural) sequence. More recently, however, researchers have begun to debate the existence of a single hierarchy sequence and have proposed alternative-order hierarchy models. The purpose of this article is to review the large body of hierarchy literature that has accumulated over nine decades. The review first describes the traditional hierarchy models and is followed by a discussion of alternative-order hierarchy models. A critique of the evidence supporting the hierarchy proposition as a model of advertising effectiveness is then presented.

THE TRADITIONAL HIERARCHY OF EFFECTS: COGNITION - AFFECT - CONATION

Proponents of the traditional hierarchy framework claim that audiences of advertising and other marketing communications respond to those messages in a very ordered way: cognitively first ('thinking'), affectively second ('feeling'), and conatively third ('doing'). The most often cited hierarchy model was posited by Lavidge and Steiner (1961) and is presented in Figure 1. These writers believed that advertising was an investment in a long-term process that moved consumers over time through a variety of stair-step stages, beginning with product 'unawareness' and moving ultimately to actual purchase. Their view of the stages of the advertising hierarchy is implicitly a causal one. However, by recognizing that advertising is essentially a 'long-term' process, it suggests that a causal influence between stages must occur only in the long-run,

http://www.warc.com/print/5160p.asp

4/03/2004

A review and critique of the hierachy of effects in advertising

Page 2 of 13

although it may not be found in the short-run. The argument that a favourable response at one step is a necessary, but not sufficient condition for a favourable response at the next step is central to the idea of advertising hierarchy of response models to this day (Preston and Thorson, 1983). Predecessors to the Lavidge-Steiner Model Prior to the development of the Lavidge-Steiner model, there were many proponents of the cognition-affect-conation sequence. These proponents and their models appear in Table 1. The most well-known is AIDA (attention-interestdesire-action), generally attributed in the marketing and advertising literature to Strong (l925b). Actually, this model originated with E. St Elmo Lewis in the late 1800s and early 1900s (Strong, 1925a). Lewis theorized that sales people, in order to be successful, had to attract attention (cognition), maintain interest and create desire (affect), and then 'get action' (conation). Sheldon (1911) included 'permanent satisfaction' as a fifth step (AIDAS). This step was an early treatment of the now recognized importance of post-purchase reactions. In the selling process, an ordered process of getting the sale was critical to Sheldon: The great problem of salesmanship is so to master this fact that the customer, realizing his best interests are being served, is persuaded to make a purchase because you follow the right method. You do not try to make him take action before you have stimulated his desire: and you do not try to create desire until you have secured his interest. (p. 31). The basic stages of the AIDA model were adhered to by the vast majority of and advertising writers for 60 years after its publication. The banner of the traditional hierarchy was directly or indirectly supported by Scott (1903, 1908), Printers Ink (1910), International Correspondence Schools (1911), Hall (1915), Adams (1916), Eastman (1916), West Coast Life Insurance Company (1920), Ramsay (1921), Kitson (1921), Osborn (1922), Starch (1923), Jenkins (1935), Hawkins (Strong, 1938), Bedell (1940), and Devoe (1956). So important was the AIDA formulation at the turn of the century that Strong (l925b) estimated that 90 per cent of persons engaged in selling and the vast majority of advertising and selling textbooks fully endorsed the Lewis-Sheldon hierarchical framework. These early writers played a key role in shaping the minds of advertising and selling practitioners and researchers. While the authors of many of these early 'models' merely changed the nomenclature of their predecessors or added or deleted stages, and while there was no empirical validation of any kind, the way to a development of effective marketing communications was clearly thought to be the route of cognition, affect and conation and only in that order.

TABLE 1: A SUMMARY OF POPULAR HIERARCHY MODELS PRECEDING THE LAVIDGE-STEINER MODEL

Year 1898 AID Model Attention, Interest, Desire Developer E St Elmo Lewis Comment Developed as a sales guide for salesmen to be successful in moving a prospect to buy. Added the action stage as necessary to convince salesmen to move buyer prospects through complete selling process.

Circa 1900

AIDA

Attention, Interest, Desire, Action*

E St Elmo Lewis

1910

AICA

Attention, Interest, Conviction, Action

Printer's Ink Editorial The first mention of the hierarchy model for advertising use; a complete advertisement must follow this model of persuasion.

http://www.warc.com/print/5160p.asp

4/03/2004

A review and critique of the hierachy of effects in advertising

Page 3 of 13

1911

AIDAS

Attention, Interest, Desire, Action, Satisfaction

Arthur F Sheldon

Added 'permanent satisfaction' as a necessary part of the persuasive and long run selling process; this final stage not carried through to contemporary literature. The necessary steps in writing a good persuasive advertisement. Mentioned this model at the beginning of his book on how to write effective, direct advertising although the model is not developed in the book. Used this model in writing about how the mind of the buyer works. Writing about the creative/persuasive process in advertising. For advertising to sell, it must follow these 'proved selling stratagems' as formulated by Kitson in 1921. Referred to the importance of different psychological sequences in constructing advertisements (AIDCA and AIDMA) but does not develop these in his book.

1915

AICCA

Attention, Interest, Confidence, Conviction, Action

Samuel R Hall

1921

AIDCA

Attention, Interest, Desire, Caution, Action

Robert E Ramsay

1921

AIDCA

Attention, Interest, Desire, Conviction, Action

Harry D Kitson

1922

AIJA

Attention, Interest, Judgement, Action

Alexander Osborn

1940

AIDCA

Attention, Interest, Desire, Conviction, Action

Clyde Bedell

1956

AIDMA

Attention, Interest, Desire, Memory, Action

Merrill DeVoe

*Italics indicate change in stage/nomenclature from previous model(s)

Recent traditional hierarchy support While the Lavidge-Steiner model appeared to be developed independently of its 'predecessors', it clearly advocated the traditional ordering notion. However, its authors went further than earlier writers in holding that the hierarchy steps were not necessarily equidistant from each other. Furthermore, consumers could move 'up' several steps simultaneously. Lavidge and Steiner (1961) were also really the first to refer to the concept of respondent 'involvement'. According to the authors, the consumer's psychological or economic commitment would have an important bearing on his or her succession to the conation stage of the hierarchy with the more committed purchasers taking longer to go through the process. Nevertheless, the more committed or 'involved' consumers would go through

http://www.warc.com/print/5160p.asp

4/03/2004

A review and critique of the hierachy of effects in advertising

Page 4 of 13

the same ordered procedure as less committed or 'involved' consumers. Only the speed of the progression might differ. Several more recent publications also advocate the traditional hierarchy framework. The most important are illustrated in Table 2. Among those is the work of Colley (1961) who in his well-known DAGMAR (Defining Advertising Goals for Measured Advertising Results) suggested that a hierarchy of advertising objectives should be used by managers to measure advertising effects rather than focusing on sales alone. At the same time, the Advertising Research Foundation (1961) developed a separate, but similar hierarchy model suggesting the stages of exposure, perception, communication (knowledge), communication (attitude), and action.

TABLE 2: A SUMMARY OF MORE RECENT HIERARCHY MODELS SUPPORTING THE TRADITIONAL COGNITIVE-AFFECTIVE-CONATIVE ORDERING
Year 1961 ACCA Model Awareness, Comprehension*, Conviction, Action Developer Russell H Colley Comment Proposed this model as important to the development of specified advertising goals and measuring advertising effectiveness. The model supported by the foundation of advertising practitioners and researchers to be used for developing more effective advertising campaigns. Illustrated how business used the hierarchy concept as a guideline to develop advertising strategy. Proposed the first application of a hierarchy-type model to the process of new product adoption. The first to suggest that probabilities could be associated with the stages of the hierarchy models to show ultimate behavioural impact of advertising.

1961

EPCCA

Exposure, Perception, Advertising Research Communication Foundation (Knowledge), Communication (Attitude), Action

1962

AAPIS

Awareness, Harry D Wolfe, James Acceptance, K Brown, C Clark Preference, Intention, Thompson Sale Provocation

1962

AIETA

Awareness, Interest, Evaluation, Trial, Adoption

Everett M Rogers

1969

PACYRB

Presentation, Attention, Comprehension, Yielding, Retention, Behaviour

William J McGuire

http://www.warc.com/print/5160p.asp

4/03/2004

A review and critique of the hierachy of effects in advertising

Page 5 of 13

1971

ACALTA

Awareness, Comprehension, Attitude, Legitimation, Trial, Adoption

Thomas S Robertson

|Expanded on the adoption hierarchy of Rogers, this model more based on Howard and Sheth buyer behaviour model (attention, comprehension, attitude, intention, purchase). Proposed more comprehensive hierarchy model that preserved the traditional order: distribution, vehicle exposure, advertising awareness, advertising elements awareness, association evaluation, product perception, integrated perception, products evaluation, prior evaluation, integrated evaluation, product stimulation, prior stimulation, integrated stimulation, search, search perception, search valuation, search stimulation, trial, trial perception, trial stimulation, adoption, adoption evaluation, adoption stimulation.

1982 1983 1984

The Association model and the expanded Association model

Ivan L Preston

Esther Thorson

*Italics indicate change in stage/nomenclature from previous model(s)

Further evidence that the cognition-affect-conation ordering process was popular came in the form of Rogers' (1962) adoption model. Rogers proposed that consumers followed a hierarchical process of awareness, interest, evaluation, trial, and adoption in the new product-adoption process. This model was expanded upon by Robertson (1971) when he proposed an awareness, comprehension, attitude, legitimization, trial, and adoption hierarchy. The first to attach probabilities to the traditional sequence was McGuire (1969; 1978). According to McGuire, the probability that one would engage in the purchase of a brand as a result of advertising was conditional upon one's retention of a message, which was conditional upon yielding to the message, which was conditional upon comprehension, which, itself, was conditional upon attention to a presented message. Ultimately, purchase of a brand as a result of advertising had a very low probability because of multiplicative conditional probabilities. The most recent conceptualization which supports the traditional hierarchy-of-effects framework comes from the work of Preston and Thorson (Preston, 1982; Preston and Thorson, 1983, 1984) and their 'association model'. This model focuses on a comprehensive advertising process that takes into account advertising research techniques (eg, syndicated data, surveys, experimentation) and concepts consistent with behavioural intentions models (Fishbein and Ajzen, 1975).

ALTERNATIVE-ORDER HIERARCHIES OF EFFECTS

http://www.warc.com/print/5160p.asp

4/03/2004

A review and critique of the hierachy of effects in advertising

Page 6 of 13

While there is little disagreement among researchers regarding the importance of the three stages of the hierarchy, there has been significant disagreement regarding the order of the three stages. This has been the area of the most intense criticism and debate concerning the hierarchy of effects. The first alternative model, formally recognized as such, followed the suggestions of Krugman (1965, 1966) and was labelled a 'low involvement' hierarchy. In this perspective, consumers are seen to be passive and disinterested recipients of many advertising messages. With little motivation to filter those messages, massive repetition of advertisements eventually leads to a modified cognitive structure in those consumers, who may purchase a product on that basis alone, and decide afterwards whether they like it. Thus, Krugman appears to suggest a cognitionconation-affect sequence. Zajonc (1980a, 1980b, 1984, 1986) and Zajonc and Markus (1982), on the other hand, suggest that preferences do not require a cognitive basis, but instead are primarily affectively based. The position that preferences can be decided on the basis of affect alone presents the potential for an affect-behaviour path. If an individual later saw a need to justify or further reflect on a preferred choice (c.f., Mayer et al., 1980), an affect-conation-cognition sequence could arise. Citing Bem (1972) and Kelly (1973), Ray et al. (1973) suggested another alternative sequence that would have relevance to marketing communications. In this situation, a person first behaves, attitudes are then formed to bolster choice, and selective learning follows to further support that action. Thus, a conation-affect-cognition sequence was indicated. However, a conation-cognition-affect sequence also appears plausible. Kiesler (1971) states that a behaviour often wields a power of commitment which results in the reorganization of cognitions to be consistent with that commitment. Affective formation or change may then follow commitment with both behaviours and cognitions. In a marketing context, for example, purchase of a product may cause one to think about it in a manner that supports the action and then feelings are developed consistent with those thoughts and the behaviour. The work of Vaughn (1980, 1986) of Foote, Cone and Belding presents an applied interpretation of the conation-affectcognition and conation-cognition-affect orderings noted above. However, this author offers another possibility: affect-cognition-conation. This sequence is thought to typify the responses of 'feeling' consumers who respond more to emotion than information in making purchase decisions resulting from advertising messages. Vaughn posits that this hierarchy is applicable to consumers when buying 'emotional products' such as fashions, jewellery, and cosmetics. The possibility of an affect-cognition-conation sequence has also been suggested by Zajonc (1980a, b), and is consistent with his position that affect usually, if not always, temporally precedes cognition. In summary, theoreticians and researchers appear to have provided a foundation for at least six different hierarchical models with the potential for explaining consumption-related activities, in general, and responses to advertising specifically. The models that have been presented are: 1. 2. 3. 4. 5. 6. cognition-affect-conation cognition-conation-affect affect-conation-cognition conation-affect-cognition conation-cognition-affect affect-cognition-conation

These various models have been presented here for the sake of completeness (ie, they represent all possible combinations). The relevance of these different models with respect to understanding how advertising works is a separate issue that needs to be considered.

A CRITIQUE OF THE HIERARCHY OF EFFECTS

What can be made of these proposals with respect to understanding how advertising works? For guidance, a review of the empirical literature was conducted. One immediate problem encountered is that there does not appear to be an accepted means of distinguishing between cognition and affect which is devoid of criticism by workers in the hierarchy area. This seems attributable to two related concerns: 1. uncertainty over how to define cognition and affect precisely (which, in turn, may determine one's conclusions concerning temporal or causal relationships between them); and 2. the difficulty in using measures which fully incorporate all possible dimensions of the constructs (which, again, may determine one's conclusion concerning temporal or causal relationships between them). Nevertheless, cognition is typically defined as 'mental activity' as reflected in knowledge, beliefs or thoughts that someone has about some aspect of their world. For the present purposes, measures of cognition will be distinguished by their non-valanced nature. For example, advertisers have historically relied on measures of memory, such as various recall, recognition and key comprehension scores, to operationalize cognition.

http://www.warc.com/print/5160p.asp

4/03/2004

A review and critique of the hierachy of effects in advertising

Page 7 of 13

Peterson et al. (1986) note that affect is typically treated as feelings and emotions which are physiologically based or have some physiological component. However, advertisers often use affect synonymously with the concept of 'attitude'. One possible difficulty here, as alluded to by Lazarus (1984), is that someone's self-reported attitude may simply be an intellectual choice (cognition) and not solely a feeling-based preference. Nevertheless, for the present purposes, any measure distinguished by its valenced nature, or feeling and emotional measurements in a more general sense, is included in the affective component of the hierarchy. For the sake of simplicity, workers in the hierarchy area have used the term 'conation' to refer to either intentions to perform a behaviour or performance of the behaviour per se, and this definition was likewise used for the present purposes. The type of behaviour most commonly discussed in the hierarchy literature is product purchase, although other consumption-related behaviours might include recommending a product, initiating information search, facilitating product usage, and so forth. Using the above guidelines, the literature was reviewed for empirical evidence supporting the different models. Several criteria were used for study selection. First, the study had to directly examine the effects of advertising. Second, measures of all three constructs had to be present. Third, each construct had to be modelled as a separate and distinct component with definitions corresponding to those presented above. Fourth, either temporal precedence or causal relationships among the three constructs had to be examined, but not necessarily found. Use of these criteria seemed reasonable. The first corresponds to the historical premise of the hierarchy of effects, that is, that it describes the process by which advertisements come to influence recipients of those advertisements. The extent to which the hierarchical process associated with advertising differs from the process associated with noncommercial messages, or consumer decision-making in general, is questionable, but that issue is incidental to the objectives here. This critique focused on investigations using advertising messages to assess the degree to which some hierarchical process involving cognitive, affective, and conative measures has been demonstrated in the advertising area. The second criterion seemed necessary since bivariate relationships may provide misleading indications of trivariate relationships. The third and fourth criteria simply correspond to a key premise underlying the hierarchy of effects: each stage is causally related (or at least temporally prior) to the next. Using these criteria, two studies were found (Zinkhan and Fornell, 1988; Batra and Vanhonacker, 1986). Zinkhan and Fornell (1988) provided some causal evidence consistent with the traditional hierarchy. Batra and Vanhonacker's (1986) time-series investigation provided a complex set of results difficult to reconcile with any given model. On the basis of the evidence presented to date, it is difficult to make a clear statement concerning any of the hierarchy-of-effects models in advertising. This conclusion remains even given other studies that have been frequently cited in the advertising hierarchy literature (eg, Ray et al., 1973; Palda, 1966; Assael and Day, 1968; O'Brien, 1971) which did not fit the criteria utilized here. For obvious reasons, the results of our empirical review led us to conclude that something was amiss in the hierarchy debate. Part of the problem may lie in the inherent difficulty in being able to make firm conclusions concerning the situations in which different models do and do not operate. A second aspect of the problem may relate to uncertainty over why it is important to address the hierarchy problem. These issues, and their related implications, are addressed in the following sections. The hierarchy sequence debate One issue that needs to be mentioned is whether each of the six models reviewed above is equally relevant to understanding the effects of advertising. Two of the models, conation-cognition-affect and conation-affect-cognition, appear less relevant than the others. Strictly speaking, these models suggest that the advertising influence process begins with purchase-related behaviour with no prior cognition or affect. On that basis alone, we feel their utility is questionable, although it would be difficult to support that contention in an absolute sense. One potential value of the above two models is the recognition that behaviour can influence both cognitive and affective processing. It is clear that the use of a product in itself can have strong influence on what one thinks and how one feels about the product. However, in order for the influence of a behaviour on cognition and/or affect to be relevant to the advertising hierarchy of effects, that influence must be shown to be different for consumers exposed to advertising than for those not exposed. The same point might be argued for other relationships in the hierarchy as well. The demonstration of the influence of advertising on any process requires appropriate and controlled comparison points. The remaining four models reviewed were: cognition-conation-affect, affect-conation-cognition; cognition-affectconation; affect-cognition-conation. In one sense, the hierarchy sequence debate is easy to summarize with respect to

http://www.warc.com/print/5160p.asp

4/03/2004

A review and critique of the hierachy of effects in advertising

Page 8 of 13

these models since so little is known with any certainty. However, the principle issue that is central to the controversy surrounding these models is concerned with the independence and sequential ordering of cognition and affect. This debate, in various forms, has already been heatedly contested for years and shows little promise of being quickly resolved. Much of the debate is currently motivated by the replicable and seminal work of Zajonc. Zajonc has demonstrated (eg, Moreland and Zajonc, 1977, 1979; Kunst-Wilson and Zajonc, 1980) that increased liking of a stimulus can occur without recognition memory for that stimulus and argues for the partial independence of cognitive and affective systems in the human species. Zajonc, (1980a, b) also contends that affective reactions usually, if not always, precede cognitive processing. Others persuasively argue that affect is post-cognitive (Mandler, 1982; Lazarus, 1981, 1982, 1984; Tsal, 1985). For example, Tsal (1985) points out that recognition failure does not necessarily indicate the absence of prior cognitive mediation of affect given the fallibility of memory processes. Measures of either recognition or recall can only serve as indirect indicators of what was learned. On a theoretical (or empirical) level it appears that there are no clear grounds to dismiss any of the four models noted above. On a practical level, the value of the debate is unclear. This is discussed next. Perspectives on the hierarchy sequence debate The cognitive and affective systems in the human species are closely interwoven. The question of whether (or when) a cognitive or affective response 'comes first' is not easily answered. As summarized by Peterson et al. (1986), 'from a practical perspective this distinction is probably arbitrary and partially depends on how cognition and affect are defined' (p. 158). Debates on this issue (eg, Zajonc, 1980a, 1984; Lazarus, 1982, 1984; Watts, 1983) appear to ultimately regress to differentiations between neural and physiological processes separated by milliseconds. However, it seems unlikely that millisecond level differentiations will play a dominant role in eventual consumer decision-making. It is even more unlikely that the most commonly employed advertising research tools are capable of distinguishing such effects. Yet, once cognitive and affective processing begin to interact, even at the millisecond level of initial stages of information processing, the quest for clearly defining a hierarchy based on the idea of sequential ordering of effects loses meaning. If it is contended that we can only be concerned with the ordering of cognitive and affective processes that we are able to 'reliably determine', then the notion of a sequential hierarchy becomes a contingent one that will vary as a function of the sophistication of the research tools employed and the sensitivity of the chosen measurements. If it is argued that the 'cognition' and 'affect' of interest to practitioners are different, in some way, from the meaning of those constructs when debated at the basic research level, then a sequential hierarchy in advertising will require a unique set of theoretical and measurement assumptions which has yet to be provided. If it is conceded that, on a managerial level, the real issue is one of degree of response and not absolute order of response, then the notion of a sequential hierarchy of effects becomes vague and undefined. However, by embracing the idea of 'degree of response', the question concerning 'which comes first' is replaced by a question concerning how the two operations can together be most usefully considered in their influence on some criterion. This issue is discussed next. An integrated perspective Fishbein and Ajzen (1975) view beliefs as a measure of cognition in which the evaluative implications of those beliefs are immediately assessed by an individual when the beliefs are formed such that 'as a person forms beliefs about an object, he automatically and simultaneously acquires an attitude towards an object' (p. 216). This expectancy-value formulation is conceptually based on the premise that cognition determines affect (ie, beliefs determine attitude), although the sequential ordering and causal relationship between the two is considered a secondary issue: 'although we have argued that a person's salient beliefs determine his attitude, the model itself it not predicated on an assumption of causality but merely deals with...the way in which different beliefs (and the evaluations of the associated attributes) are combined or integrated to arrive at an evaluation of the object' (Fishbein and Ajzen 1975, pp. 222-223). It is the explicit integration of non-valanced (beliefs) and valenced (evaluative implications of those beliefs) information which distinguishes the Fishbein-Ajzen behavioural intention model from most discussions of hierarchyof-effects models in advertising. This expectancy-value is often called cognitive attitude'. In the advertising area, Smith and Swinyard (1982, 1988) argue for the benefits of expectancy-value formulations (along with the concept of 'trial' versus 'commitment' behaviour) in helping to resolve sequence controversies in the hierarchy of effects. Preston's (1982) 'association model' includes as its final steps 'integrated evaluation' (where recipients of advertising are thought to engage in expectancy-value type processing) which is postulated to lead to 'integrated stimulation' (purchase intentions) which leads to 'action' (typically, purchase behaviour). The incorporation of an expectancy-value framework in advertising hierarchy-of-effects models amounts to a recognition that the joint consideration of cognitive and affective processing has particular utility with respect to understanding preferences. The need to know? From a practical perspective, why do we need to know about the sequential hierarchy of effects in advertising?

http://www.warc.com/print/5160p.asp

4/03/2004

A review and critique of the hierachy of effects in advertising

Page 9 of 13

One reason is that determining hierarchical processes allows us to predict behaviour (Preston and Thorson, 1983). Yet, the easiest way to predict what someone will do is simply to ask them what they intend to do (ie, the intentionbehaviour relationship). Determining temporal precedence or a causal relationship between cognition and affect is not necessary if one's interest is simply the prediction of behaviour. The most immediate determinant of behaviour is behavioural intentions. A second reason is that understanding the hierarchy of effects provides information on what advertising strategy to emphasize. The notion of different hierarchical sequences appears to have influenced management by the assumption that if a given component comes first, it dominates what follows and thus can have important implications for advertising strategies (Vaughn, 1986). The problem is that such an event has yet to be empirically demonstrated. It is unclear whether primacy of an initial cognitive or affective response has a significant influence on the processing of advertising-related information. Most importantly, even if such an influence could be demonstrated, its magnitude is likely to be relatively minor and 'washed out' prior to behaviour. No evidence currently exists supporting the contention that the sequential ordering of cognitive versus affective responses to advertising communications 'ultimately masters' in terms of what people purchase or consume. A third reason is that the hierarchy of effects has proven valuable for helping to organize planning, training, and conceptual tasks within a firm. This appears valid. As previously reviewed, the advertising literature has historically utilized the idea of a hierarchy of effects for identifying important concepts which need to be addressed. As a heuristic tool, the hierarchy concept has survived because it has an intuitive appeal. The fact that there is a difference between the use of a model as a heuristic tool and its empirical realization may not be a critical issue in the minds of many practitioners. From an empirical perspective, however, the primary difficulty appears to lie in defining when one stage ends and another begins, or when one stage starts, but another does not. This issue is frustratingly complicated by the fact that any such sequential determinations will be very sensitive to how the hierarchical components are operationalized and the research methods employed. Add to this the uncertain relevance of results to management. Concerning the cognition-affect controversy, Peterson et al. (1986) suggest: The question 'did the person think first or feel first' is not very meaningful. Individuals are always in a stream of thinking or feeling; therefore, it is irrelevant to say 'Are there any thoughts preceding an affect' or 'is there affect preceding cognition. Both of these activities are continually occurring....Mental activities are dynamic, not static. The important issue to be addressed is how affect and cognition interact to influence behaviour. To this date in the advertising hierarchy literature, the views of Smith and Swinyard(1982) come closest to addressing this issue. Hierarchical prospects The empirical evidence one can offer to support the hierarchy of effects depends, of course, on how one wishes to define the various components and other characteristics of a study one considers desirable. Nevertheless, using the criteria selected here, little evidence was found to support the existence of an advertising hierarchy. However, the conclusion that a hierarchy does not exist would be contrary to the principal point we wish to raise. We believe the lack of published research may reflect the opinion of many researchers that there is uncertain value in undertaking such research. Two possible reasons for this could be offered: one is the question of whether (or when) cognition precedes affect, or vice versa. This shows little promise of being resolved in the immediate future and may vary as a function of how the constructs are arbitrarily defined. Lazarus (1984) argues that this 'primacy' issue is fundamentally indeterminant. If he is wrong, this sensitive question probably shows the least promise of being conclusively resolved using a complex stimulus such as an advertisement, especially since for most products consumers already possess product-related cognitions and affect in memory, further complicating the search for a sequential order of response. Secondly, even if the first issue were resolved, researchers would still be faced with the difficulty of demonstrating that the information provided by conducting a hierarchical investigation was not only valuable to management, but could not be obtained more easily in other ways. The onus is on future research in the hierarchy area to address the above two concerns before other issues are considered. First, efforts must be made to assure that the various hierarchies can be confirmed or disproved, even if this involves limiting the meaning of the terms 'cognition' and 'affect'. The framework required will include a set of welldefined theoretical and measurement assumptions and an acceptable (as well as unacceptable) means of testing those assumptions. For example, are advertisers content with assuming that if a consumer is unable to recall or recognize an advertisement then 'cognition' did not occur? If that assumption is unacceptable (eg, the information was learned, but was lost in the period between acquisition and attempted retrieval) then a precise means of resolving the issue must be

http://www.warc.com/print/5160p.asp

4/03/2004

A review and critique of the hierachy of effects in advertising

Page 10 of 13

delineated. Simply suggesting that a given assumption is unacceptable is of little benefit unless one can provide an assumption that is acceptable. Are advertisers content with assuming that if attitude change is not found after exposure to advertising then 'affect' did not occur? If that assumption is unacceptable (eg, affective processing did occur, the end-product of which happened to be 'no change' in the criterion score) then a means of resolving the issue must be presented. Since the effects of advertising are often 'long-term' (Lavidge and Steiner, 1961), can it be assumed that investigations of 'short-term', especially single-advertisement exposures, provide insufficient evidence concerning the hierarchy? Consider, for example, a finding of cognition-affect-conation causal linkages, which would be consistent with the traditional hierarchy of effects. However, without defining what is meant by 'long-term' effects of advertising, there is no requirement that a lack of causality obtained at any given point in time be considered inconsistent with that hierarchy. The problem is that a theory in which causality is consistent, but a lack of causality is not inconsistent, with its underlying assumptions is potentially non-falsifiable. Assuming that the first concern is addressed, the contributions of results stemming from hierarchy investigations must be vigorously compared to other means of obtaining the same information. A choice can then be made about which method is most practical. For example, Vaughn's (1986) extensive study arrived at conclusions thought to be consistent with different hierarchies, but were not obtained by empirically examining those different hierarchies. Accomplishing the first two steps should then allow workers in the area to distinguish, if necessary, between the hierarchy of effects as a heuristic model which provides guidance to management in a general sense, as opposed to an empirical model which can be relied upon to provide consistent and interpretable results which deserve to be specifically addressed.

CONCLUSION
The concept of an advertising hierarchy of effects has been advocated for at least 80 years. The major challenge for those researching the possibility of such a hierarchy in the future is the conceptualization of a framework which, when tested, allows clear and unambiguous inferences to be made concerning competing sequences. This must be followed by coming to grips with whether the information is both valuable to management and cannot be more easily obtained in other ways. If success with both of these tasks cannot be attained, viewing the hierarchy of effects simply as a heuristic model which may have utility for general planning and guidance purposes may be most appropriate.

ACKNOWLEDGEMENTS
The authors appreciate the financial support of the Research and Development Council and the Center for Marketing Management Studies as well as comments of Professor John Slocum, Cox School of Business at SMU.

REFERENCES
1. Adams, H. E. (1916) Advertising and its Mental Laws. New York: The Macmillan Company. 2. Advertising Research Foundation (1961) Audience Concepts Committee. Toward Better Media Comparisons, New York. 3. Assael, H. & Day, G. S. (1968) Attitudes and awareness as predictors of market share. Journal of Advertising Research, 8, 3-10. 4. Batra, R. & Vanhonacker, W. R. (1986) The hierarchy of advertising effects an aggregate field test of temporal precedence. New York: Columbia Business School. (Avis Rent a Car System Working Paper Series in Marketing, March). 5. Bedell, C. (1940) How to Write Advertising that Sells. New York: McGraw-Hill. 6. Bem, D. J. (1972) Self-perception theory. Advances in Expenmental Social Psychology. Berkowitz, L. (Ed.) New York: Academic Press. 7. Colley, R. H. (1961) Defining Advertising Goals for Measured Advertising Results. New York: Association of National Advertisers. 8. Devoe, M. (1956) Effective Advertising Copy. New York: The Macmillan Company. 9. Eastman, G. R. (1916) Psychology of Salesmanship. New York: A. F. Fenno and Dayton, OH Service Publishing. 10. Fishbein, M. & Ajzen, I. (1975) Belief, Attitude, Intention and Behavior. Reading, Massachusetts: AddisonWesley. 11. Hall, S. R. (1915) Writing an Advertisement: Analysis of the Methods and Mental Processes that Play a Part in the Writing of Successful Advertising. Boston: Houghton Mifflin Co. 12. International Correspondence Schools (1911) The Conducting of Sales. 13. Jenkins, J. G. (1935) Psychology in Business and Industry. New York: John Wiley & Sons, Inc. 14. Kelley, C. A. (1973) The process of causal attribution. American Pyschologist, 28, 107-128. 15. Kiesler, C. A. (1971) The Psychology of Commitment: Experiments Linking Behavior to Belief. New York:

http://www.warc.com/print/5160p.asp

4/03/2004

A review and critique of the hierachy of effects in advertising

Page 11 of 13

Academic Press. 16. Kitson, H. D. (1921) The Mind of the Buyer: A Psychology of Selling. New York: The Macmillan Company. 17. Krugman, H. E. (1965) The impact of television advertising: learning without involvement. Public Opinion Quarterly, 29, 349-356. 18. Krugman, H. E. (1966) Answering some unanswered questions in measuring advertising effectiveness. Proceedings, 12th Annual Meeting, Advertising Research Foundation, New York, 18-23. 19. Kunst-Wilson, W. R. & Zajonc, R. B. (1980) Affective discrimination of stimuli that cannot be recognized. Science, 207, 557-558. 20. Lavidge, R. C. & Steiner, G. A. (1961) A model for predictive measurements of advertising effectiveness. Journal of Marketing, 25, 59-62. 21. Lazarus, R. S. (1981) A cognitivist's reply to Zajonc on emotion and cognition, American Psychologist, 36, 222223. 22. Lazarus, R. S. (1982) Thoughts on the relations between emotion and cognition. American Psychologist, 37, 1019-1024. 23. Lazarus, R. S. (1984) On the primacy of cognition. American Psychologist, 39, 124-129. 24. Mandler, G. (1982) The structure of value: Accounting for taste. In Cognition and Affect. (Ed.) Fiske, S. & Clarke, M. New York: Academic Press. 25. Mayer, F. S., Duval, S. & Duval, V. H. (1980) An attributional analysis of commitment. Journal of Personality and Social Psychology, 39, 1072-1080. 26. McGuire, W. J. (1969) An information-processing model of advertising effectiveness. Paper presented to Symposium on Behavioural and Management Science in Marketing, University of Chicago. 27. McGuire, W. J. (1978) An information-processing model of advertising effectiveness. In Behavioral and Management Science in Marketing (Ed.) Davis, H. L. & Silk, A. pp. 156-180. New York: Ronald Press. 28. Moreland, R. L. & Zajonc, R. B. (1977) Is stimulus recognition a necessary condition for the occurrence of exposure effects Journal of Pasonalily and Social Psychology, 35, 191-199. 29. Moreland, R. L. & Zajonc, R. B. (1979) Exposure effects may not depend on stimulus recognition, Journal of Personality and Social Psychology, 32, 1085-1089. 30. O'Brien, T. (1971) Stages of consumer decision making. Journal of Marketing Research, 8, 282-289. 31. Osborn, A. E. (1922) A Short Course in Advertising. New York C. Scribner's Sons. 32. Palda, K. S. (1966) The hypothesis of a hierarchy of effects: a partial evaluation. Journal of Marketing Research, 3, 13-24. 33. Peterson, R. A., Hoyner, W. D. & Wilson, W. R. (1986) Reflections on the role of affect in consumer behavior. In The Role of Affect in Consumer Behavior: Emerging Theories and Applications. Lexington, Massachusetts: Lexington Books, pp. 141-159. 34. Preston, I. L. (1982) The association models of the advertising communication process. Journal of Advertising, 11, 3-15. 35. Preston, I. L. & Thorson, E. (1983) Challenges to the use of hierarchy models in predicting advertising effectiveness. Proceedings of the Annual Convenhon of Amencan Academy of Adverdsing, 27-33. 36. Preston, I. L. & Thorson, E. (1984) The expanded association model: keeping the hierarchy concept alive. Journal of Advertising Research, 24, 59-65. 37. Printers' Ink (1910) December 1, 74. 38. Ramsay, R. E. (1921) Effective Direct Advertising: The Principles and Practice of Producing Direct Advertising for Distribution by Mail or Otherwise. New York: Appleton and Company. 39. Ray, M. L., Sawyer, A. G., Rothschild, M. L., Heeler, R.M., Strong, E.C. & Reed, I. B. (1973) Marketing communications and the hierarchy of effects. In New Models for Mass Communication Research. (Ed.) Clarke, P., pp. 147-176. Beverly Sage, CA: Publishing. 40. Robertson, T. S. (1971) Innovative Behavior and Communication. New York Holt, Rinehart, Winston. 41. Rogers, E. M. (1962) Diffusion of Innovation. New York Free Press. 42. Schmalensee, D. H. (1983) Today's top priority advertising research questions. Journal of Advertising Research, 23, 49-60. 43. Scott, W. D. (1903) The Theory of Advertising: A Simple Exposition of the Principles of Psychology in Their Relation to Successful Advertising. Boston: Small, Maynard and Company. 44. Scott, W. D (1908) The Pyschology of Advertising: A Simple Exposition of the Principles of Psychology in Their Relation to Successful Advertising. Boston: Small, Maynard and Company. 45. Sheldon, A. E (1911) The Art of Selling. Chicago: The Sheldon School. 46. Smith, R. E. & Swinyard, W. R. (1982) Information response models: an integrated approach. Journal of Marketing, 46, 81-93. 47. Smith, R. E. & Swinyard, W. R. (1983) Attitude-behavior consistency: the impact of product trial versus advertising. Journal of Marketing Research, 20, 257-267. 48. Smith, R. E. & Swinyard, W. R. (1988) Cognitive response to advertising and trial belief strength, belief confidence and product curiosity. Journal of Advertising, 11, 3-14. 49. Starch, D. (1923) Principles of Advertising. New York: McGraw-Hill Book Company. 50. Strong, Jr., E. K. (1925a) The Psychology of Selling and Advertising. New York: McGraw-Hill. 51. Strong, Jr., E. K. (l925b) Theories of selling. The Journal of Applied Psychology, 9, 75-86. 52. Strong, Jr., E. K. (1938) Psychological Aspect of Business. New York: McGraw-Hill.

http://www.warc.com/print/5160p.asp

4/03/2004

A review and critique of the hierachy of effects in advertising

Page 12 of 13

53. Tsal, Y. (1985) On the relationship between cognitive and affective processes: a critique of Zajonc and Markus. Journal of Consumer Research, 12, 358-362. 54. Vaughn, R. (1980) How advertising works: a planning model. Journal of Advertising Research, 20, 27-33. 55. Vaughn, R. (1986) How advertising works: a planning model revisited. Journal of Advertising Research, 26, 5766. 56. Watts, F. N. (1983) Affective cognition: a sequel to Zajonc and Rachman. Behavioral Research and Theory, 21, 89-90. 57. West Coast Life Insurance Company (1920) The five steps of a sale. 58. Wolfe, H. D., Brown, I. K. & Thompson, G. C. (1962) Measuring Advertising Results. p. 7. New York: National Industrial Conference Board. 59. Zajonc, R. B. (1980a) Feeling and thinking: preferences need no inferences. American Psychologist, 35, 151175. 60. Zajonc, R. B. (1980b) Cognition and social cognition: a historical perspective. In Four Decades of Social Psychology, (Ed.) Festinger, L., pp. 1-42. Oxford: Oxford University Press. 61. Zajonc, R. B. & Markus, H. (1982) Affective and cognitive factors in preference. Journal of Consumer Research, 9, 123-131. 62. Zajonc, R. B. (1984) On the primacy of affect. Amencan Psychologist, 39, 117-123. 63. Zajonc, R. B. (1986) Basic mechanisms of preference formation. In The Role of Affect in Consumer Behavior: Emerging Theories and Applications. (Ed.) Peterson, R. A., Hoyer, W. D. & Wilson, W. R., pp. l-16. Lexington, Massachusetts: Lexington Books. 64. Zinkhan, G. M. & Fornell, C. (1989) A test of the learning hierarchy in high and low-involvement situations. Association for Consumer Research, 16, pp. 152-159.

Advertising Association

http://www.warc.com

NOTES & EXHIBITS

Thomas E Barry
Thomas E Barry is Professor of Marketing at the Edwin L Cox School of Business, Southern Methodist University, USA which he joined in 1970. Professor Barry has published widely in the advertising field in the Journal of Marketing Research, Journal of Advertising Research, Journal of Advertising, Current Issues and Research in Advertising, Interfaces, European Journal of Marketing, and other journals. He is a co-author of Advertising Management: Text and Cases and the author of Marketing: An Integrated Approach. He has served as associate dean for academic affairs in the Cox School as well as chairman of the marketing department and has received awards for excellence in teaching and research. He earned his doctorate in marketing at the University of North Texas in 1971.

Daniel J Howard
Daniel J Howard is Assistant Professor of Marketing in the Edwin L Cox School of Business, Southern Methodist University, which he joined in 1985. He earned his doctorate in marketing at the Ohio State University, USA in 1986. He is interested in advertising and consumer psychology and has published in Journal of Marketing Research, Psychometrica, Journal of Applied Social Psychology, Journal of Experimental Social Psychology, Journal of

http://www.warc.com/print/5160p.asp

4/03/2004

A review and critique of the hierachy of effects in advertising

Page 13 of 13

Advertising Research, Current Issues and Research in Advertising, and other journals. He has been sponsor of the student marketing club and has received awards for excellence in teaching from the Cox School and SMU.

FIGURE 1: THE LAVIDGE-STEINER TRADITIONAL ORDER HIERARCHY OF EFFECTS

a Indicates

a stage not presented in predecessor models

http://www.warc.com/print/5160p.asp

4/03/2004