Special Session Summary Dual-Process Theories in Affective Persuasion

ABSTRACT - The main objective of this session is to look at some recent studies that use dual-process theories (see Chaiken & Trope, 1999) in the domain of persuasive communication. More specific, the dual-role of affect is taken into account. In the first study, affect-as-information (as a heuristic processing mode) is used to explain the influence of attitudinal valence on the degree of evaluative consistency among attitude related judgements. Consistent with dual-theories, support for an asymmetrical view of affect on persuasion effects is found (Forgas, 2000). In the second study, the effects of mood on the evaluation of negatively valenced mental categories are investigated. Consistent with dual-process theories, the affect-as-information hypothesis could only be supported when respondents’ processing motivation is low. The final study is more theoretical in nature, and discusses the strengths and weaknesses of existing dual-process models. On the basis of these weaknesses, an extended model, the CASCADES model, is presented. Following an introduction into the basic ideas of the model, some of its implications for understanding affective persuasion, especially with regard to the other two papers of the special topic session, will be discussed.

Citation:

Anick Bosmans (2001) ,"Special Session Summary Dual-Process Theories in Affective Persuasion", in E - European Advances in Consumer Research Volume 5, eds. Andrea Groeppel-Klien and Frank-Rudolf Esch, Provo, UT : Association for Consumer Research, Pages: 47-50.

European Advances in Consumer Research Volume 5, 2001      Pages 47-50

SPECIAL SESSION SUMMARY

DUAL-PROCESS THEORIES IN AFFECTIVE PERSUASION

Anick Bosmans, Ghent University, Belgium

ABSTRACT -

The main objective of this session is to look at some recent studies that use dual-process theories (see Chaiken & Trope, 1999) in the domain of persuasive communication. More specific, the dual-role of affect is taken into account. In the first study, affect-as-information (as a heuristic processing mode) is used to explain the influence of attitudinal valence on the degree of evaluative consistency among attitude related judgements. Consistent with dual-theories, support for an asymmetrical view of affect on persuasion effects is found (Forgas, 2000). In the second study, the effects of mood on the evaluation of negatively valenced mental categories are investigated. Consistent with dual-process theories, the affect-as-information hypothesis could only be supported when respondents’ processing motivation is low. The final study is more theoretical in nature, and discusses the strengths and weaknesses of existing dual-process models. On the basis of these weaknesses, an extended model, the CASCADES model, is presented. Following an introduction into the basic ideas of the model, some of its implications for understanding affective persuasion, especially with regard to the other two papers of the special topic session, will be discussed.

 

"PRIOR ATTITUDES AND CONSUMER JUDGEMENTS: THE EFFECT OF VALENCE, EXTREMITY AND ELABORATION"

Carlos Falces, Universidad Autonoma de Madrid, Spain

Benjamin Sierra, Universidad Autonoma de Madrid, Spain

Elena Alier, Universidad Autonoma de Madrid, Spain

Pablo Bri±ol, Ohio State University, U.S.A.

This research analyzes the influence of attitudinal valence on the degree of evaluative consistency among attitude related judgements. Think, for example, about consumers with very positive or negative attitudes toward a service provider, that are required to evaluate the quality they have experienced in a specific service encounter. According to the effect of attitudes on information processing, it could be expected that both groups of consumers show an attitude consistency bias in their quality judgements. In line with this hypothesis, people with an extreme attitude should show more similar scores among attributes than those with relatively moderate or neutral attitudes. However, it is possible that the influence of attitude extremity on attitude related judgements would be different as a function of attitudinal valence. Research on affect and information processing shows asymmetrical effects of affective valence on information processing tasks such as judgements. For example, the affect as information hypothesis proposes that negative affect signals that something is wrong in the environment and motivates a more detailed or analytic processing. On the other hand, positive affect informs the individual of a safe and/or pleasant environment, leading to a more holistic processing of the information at hand.

Applying this view to attitude valence, it could be expected that people with positive attitudes will be less motivated to consider the slightly differences among service dimensions and give more similar scores on those dimensions. On the other hand, a negative orientation toward an attitudinal object could lead to a more detailed consideration of different attributes and then, the probability of perceived differences among those attributes could be higher. As a consequence, the similarity of responses among judged attributes would be higher when people hold positive attitudes than when they hold negative ones.

Two field studies were conducted in order to test the possible effect of attitude valence in consumer judgements.

In the first study, a sample of customers of different restaurant services were classified according the extremity (high vs. moderate) and valence (positive vs. negative) of their previous global attitudes toward the service provider. Participants evaluated the quality of a specific service encounter using a perceived quality scale, which measures five different quality dimensions. A homogenity index was created using standard deviations of each respondents’ five quality dimension scores. Results showed a main effect of valence qualified by an interaction with the extremity of previous attitudes. Confirming the valence influence hypothesis, negative attitudes led to more dissimilar judgements among quality dimensions than positive ones. The difference between moderate and extreme attitudes was only significant in the case of negative valence and opposite to the extremity influence prediction. The more negative the attitude was, more heterogeneous judgements were made.

The second study examined if the valence effect remained the same when the elaboration likelihood was relatively high or low. The same design of the first study was used, adding the Need for Cognition Scale and a thought listing task as an index of the amount of processing. Results replicated the valence effect found in the first study and showed that compared to high need for cognition respondents, low need for cognition individuals made more similar judgements among quality dimensions when the valence was negative. Results of the thought listing task showed that people with negative attitudes gave more reasons for their evaluations than those holding a positive attitude. However no significant differences were found in the number of thoughts listed between positive nd negative high need for cognition individuals. This result suggests that valence effects can not be explained by differences in the amount of processing. As a conclusion, the results of both studies support the view of an asymmetry between negative and positive attitudes in consumer information processing.

 

"THE INFLUENCE OF MOOD ON THE EVALUATION OF NEGATIVELY VALENCED MENTAL CATEGORIES"

Anick Bosmans, Ghent University, Belgium

Patrick Van Kenhove, Ghent University, Belgium

Peter Vlerick, Ghent University, Belgium

Hendrik Hendrickx, Ghent University, Belgium

According to the affect-as-information hypothesis, people make mood-congruent judgements when their mood is seen as informative for the judgement to be made (Schwarz & Clore, 1983). In addition, the cognitive tuning extension of the affect-as-information hypothesis states that mood also provides procedural information by providing feedback about the nature of a situation (e.g. Clore, Schwarz, & Conway, 1994). Positive mood signals that a situation is benign and that a heuristic / non systematic processing strategy is adequate to meet processing goals. As a consequence, positive mood will result in a greater reliance on general knowledge structures. In contrast, moods that are more negative than usual signal that the situation is problematic, and that a more systematic processing strategy is adequate for dealing with the situation adaptively.

In the current two studies we tested this cognitive tuning effect using negatively valenced mental schema’s (blood donation and condom use). According to the cognitive tuning effect, it can be expected that positive mood will result in a greater reliance on general knowledge structures, hence in attitudes that are based on this negatively valenced mental schema and, as a result, in evaluations that are more negative. In contrast, following the evaluative component of the affect-as-information hypothesis, mood-congruent judgements can be expected.

In our first study, positive versus negative mood subjects were asked to evaluate a new (fictitious) blood donation initiative at our university (i.e. a service found to be associated with a negatively valenced mental knowledge structure). However, no effects of mood on blood donation evaluations were found. Hence, no support for the affect-as-information hypothesis, nor for its cognitive tuning extension were found.

A possible explanation for this null finding is the fact that subjects may have been too motivated to process the ad: The blood donation message was of high personal relevance for subjects, and subjects often reported negative felt affect. Both conditions have been found to increase processing motivation (for a review see Bagozzi, Gopinath, & Nyer, 1999). It has been suggested that (e.g. Forgas, 1995; Petty, Schumann, Richman, & Strathman, 1993) there are different roles for affect under high- and low- elaboration conditions. Petty et al. (1993) for example showed that, although in both high- and low- elaboration conditions positive mood had a positive influence on subjects’ product evaluations, positive mood had a direct effect on attitudes in low elaboration conditions and influenced attitudes indirectly by modifying the positivity of thoughts in high- elaboration conditions.

This would suggest that under conditions of low elaboration, mood serves as a simple heuristic cue in the formation and change of an attitude. With regard to our research, this suggests that, in line with studies investigating the effects of cognitive tuning under conditions of high versus low elaboration, our affect-as-information and cognitive tuning effects are likely to occur under conditions of low elaboration. Several cognitive tuning studies indeed did not find any cognitive tuning effects when subjects were highly motivated to process (i.e. under conditions of systematic processing) (e.g. Soldat, Sinclair, & Mark, 1997; Ottati, Terkildsen, & Hubbard, 1997). According to these studies, increased motivation can lead to sytematic processing even if affective cues suggest that the situation is benign (note that cognitive tuning is considered in the literature as a processing heuristic).

Study 2 was conducted to further test our post hoc explanation (i.e. that subjects were too motivated to use the processing heuristics). In this study, we measured attitudes towards a less negatively valenced (but still sufficiently negative) mental schema (that of condom use) and we manipulated involvement in order to obtain a high- and a low- elaboration condition. Given our above reasoning, we expect cognitive tuning of affect-as-information effects (as initially expected for the previous study) only when elaboration likelihood is low. When elaboration likelihood is high, no effects of mood on evaluation are expected.

In this second study we found that only when involvement was low, product evaluations were more positive under conditions of positive mood than under conditions of negative mood. This finding is consistent with the affect-as-information hypothesis. Moreover, no mood effects were found under conditions of high involvement. As discussed before, the latter null finding is consistent with dual theories (e.g. Petty et al., 1993; Forgas, 1995) concerning the effects of mood on persuasion. According to these theories, when motivation to process is high, mood influences the persuasion process through principles of affective priming, while under low motivation conditions, mood influences the persuasion process through affect-as-information heuristics.

In conclusion, under low elaboration conditions, we found, consistent with the affect-as-information heuristic, mood-congruent judgements. No evidence for cognitive tuning was found. This suggests that participants did not rely on their, with negative affect laden, general knowledge structures when making evaluations. Instead, people seemed to take their moods as informative for the judgement to be made.

It is clear that our study has straightforward implications for marketing and consumer behaviour practitioners. The present study shows that, even though a product or product class is associated with negative beliefs / feelings (e.g. life insurances, drugs, condoms,), people still take their moods as informative for the judgements to be made. It is important to notice however that this will only be the case under conditions of low involvement (as is the case in a lot of the persuasion attempts).

References

Bagozzi, R.P., M. Gopinath, P.U. Nyer, 1999. The role of emotions in marketing. Journal of the Academy of Marketing Science, 27(2), 184-206.

Clore, G.L., N. Schwarz and M. Conway, 1994. Cognitive causes and consequences of emotion. In R.S. Wyer and T.K. Srull (Eds.), Handbook of social cognition (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum. pp. 323-417.

Forgas, J.P., 1995. Mood and judgment: The affect infusion model (AIM). Psychological Bulletin 117, 39-66.

Ottati, V., N. Terkildsen, and C. Hubbard, 1997. Happy faces elicit heuristic processing in a televised impression formation task: A cognitive tuning account. Personality and Social Psychology Bulletin 23(11), 1144-1156.

Petty, R.E., D.W. Schumann, S.A. Richman, and A.J. Strathman, 1993. Positive mood and persuasion: Different roles for affect under high- and low-elaboration conditions. Journal of Personality and Social Psychology 64, 5-20.

Schwarz, N. and G.L. Clore, 1983. Mood, misattribution, and judgments of well-being: Informative and directive functions of affective states. Journal of Personality and Social Psychology 45, 513-523.

Soldat, A.S., R.C.Sinclair, and M.M. Mark, 1997. Color as an environmental processing cue: External affective cues can directly affect processing strategy without affecting mood. Social Cognition 15, 55-71.

 

"EXTENDING DUAL-PROCESS MODELS: THE CASCADES MODEL"

Andreas Strebinger, Vienna University of Business Administration and Economics, Australia

Originating from social psychology, dual-process modelsBin particular the elaboration likelihood model (ELM, Petty and Cacioppo 1986) and the heuristic-systematic model (HSM, Chaiken 1980)Bhave given rise to a boom-like output of academic studies in consumer research over the past two decades. The outstanding strength of these models, for one, is due to the highly parsimonious model formulation, which permits the characterisation of the impact of a persuasion variable based on no more than five functions. Certain characteristics of source, recipient, message and context (a) may, as peripheral (heuristic) cues, affect persuasion at a low level of elaboration likelihood, (b) act convincingly as central (systematic) cues at a high level of elaboration likelihood, or have a direct effect on elaboration likelihood due to a change in (c) motivation or (d) ability for central (systematic) information processing. The fifth function of a variable is based on the assumption that humans do not always process in a relatively objective manner, but (e) are biased in their persuasion by certain variables. A second strong point of the models is to be found in the enormous amount of empirical research which gives impressive evidence of their predictive ability and generalizability. In the field of consumer research alone, more than 100 publications substantiate the considerable utility of these models (Strebinger 2000, pp. 106ff).

Apart from these central strengths, a subjective inventory by the author highlights six problem areas, in which ELM and HSM still possess a potential for optimization. (1) Descriptive" character of cue classification: As Petty and Cacioppo (1986, p. 192) recognize, the question as to which variable adopts which of the five functions of the model under what circumstances is not solved by the ELM. In their critical assessment, Eagly and Chaiken (1993, p. 321) therefore refer to the ELM as primarily descriptive", thereby distinguishing it from explanatory". Despite their claim to the contrary (Eagly and Chaiken 1993, p. 342) it is hard to follow the authors of the HSM in their assumption that the HSM does solve the question why a heuristic cue should act as a heuristic cue. (2) Confounding of processing mode and content: A problem that is closely related to the descriptive" character of the models and that arises in particular in HSM is that the processing of a cue defined as heuristic in the above manner is equated with heuristic processing". (3) Lack of an intraprocedural structure": Despite outstanding contributions of their authors to the fields of automaticity research (e. g., Cacioppo et al. 1992; Chaiken and Bargh 1993), both models lack an in-depth structure of the procedure of a single, chronologically limited episode of processing (e.g., a single evaluation of a brand or a print ad), which in this context is defined as intraprocedural structure". (4) Lack of an interprocedural perspective": While some dual-process models such as the CEST come up with a long-term perspective by linking several temporally distinct processing episodes (here referred to as interprocedural perspective"), ELM and HSM do not, at least explicitly, allow for a change in the function of specific cues over several processing occasions. (5) Lack of a control function: A controlled mode of information processing implies not only literally the existence of a control authority". Even if ELM and HSM do not expressly use te term controlled", they nevertheless postulate a number of supervisory processing tasks without indicating precisely in what form these are performed. (6) Missing integration into theories of hierarchical expectancy-value formation: Obviously, both ELM and HSM assume the usage of hierarchically lower elements of expectancy-value formation (e. g., specific beliefs about a brand) to be probable when elaboration likelihood is high, whereas low ability or motivation will cause attempts to directly reach the next higher level (e. g., some expectancy-value component or a global attitude toward the brand). Apart from this implicit line of thought, both ELM and HSM remain quite vague, both in theoretical terms (cf. e.g., Eagly and Chaiken 1993, p. 241) and concerning the interpretation of their empirical findings (cf. e.g., Areni and Lutz 1988).

Based on these notions, the CASCADES model proposed by the author roots in ELM and HSM and tries to supplement them in regard to five aspects. (1) The above-mentioned problem of confounding the mode and the content of information processing is addressed by falling back on the distinction between Controlled and automatic information processing, the most important dividing line is regarded to be the level of consciousness (Spiegel 1958/1970, p. 17; Langer, Blank, and Chanowitz 1978, p. 48; Bargh 1989). On the automatic information processing side, a further distinction is made between a formerly conscious processing mode, which is "utomated by constant practice, and a Subconscious processing mode which is already based on subconscious learning (cf. Spiegel 1958/1970). (2) Automatic and controlled processes in varying combinations serve the fulfilment of two different functions: a Control and an "cting function. The tasks of the control function are to provide aims and strategies of processing as well as an ongoing supervision of the progress of the acting function. Resource-related problems are hypothesized to arise when both functions need to have access to controlled resources. (3) The problem of a descriptive cue classification and the absence of an interprocedural perspective is addressed by the CASCADES model in that it orients itself on a cost-benefit view of cue usage (e. g., Feldman and Lynch 1988). At the center of cognitive benefit, the Diagnosticity of a cue is defined according to Bayes’ theorem (Fischhoff and Beyth-Marom 1983). (4) This type of diagnosticity is interpreted as the meaningfulness of a cue in relation to a certain element of Expectancy-value formation (i. e., a specific belief, a certain expectancy-value component or the global attitude toward the object). (5) As the intraprocedural pattern is concerned, a better structure is achieved by the acceptance of a Sequence of two phases: One phase with exclusively automatic information processing merges into a phase of mixed automatic-controlled information processing which is optional subject to the principle of sufficiency, in which a review of the validity of conclusions is carried out based on cue diagnosticity. In storing the results of processing the circle is closed to processing episodes taking place at a later time. The underlying assumption is that (a) unconscious processing steps are stored in the implicit memory where they serve the creation of subconscious conclusion patterns or the simple recharging of automated processes; and (b) the conclusions of conscious processing are preserved in the explicit memory.

The paper demonstrates the usefulness of the proposed model in explaining some previously puzzling empirical findings and in deriving several new hypotheses of theoretical and practical value to consumer research.

References

Areni, Charles S. and Richard J. Lutz (1988), "The Role of Argument Quality in the Elaboration Likelihoo Model," in Advances in Consumer Research, 15, 197-203.

Bargh, John A. (1989), "Conditional Automaticity: Varieties of Automatic Influence in Social Perception and Cognition," in Unintended Thought, John A. Bargh and James S. Uleman, eds. New York: The Guilford Press.

Cacioppo, John T., Beverly S. Marshall-Goodell, Louis G. Tassinary, and Richard E. Petty (1992), "Rudimentary Determinants of Attitudes: Classical Conditioning Is More Effective When Prior Knowledge about the Attitude Stimulus Is Low than High," Journal of Experimental Social Psychology, 28 (May), 207-233.

Chaiken, Shelly (1980), "Heuristic Versus Systematic Information Processing and the Use of Source Versus Message Cues in Persuasion," Journal of Personality and Social Psychology, 39 (November), 752-766.

Chaiken, Shelly and John A. Bargh (1993), "Occurrence Versus Moderation of the Automatic Attitude Activation Effect: Reply to Fazio," Journal of Personality and Social Psychology, 64 (May), 759-764.

Eagly, Alice H. and Shelly Chaiken (1993), The Psychology of Attitudes. Fort Worth: TX: Harcourt Brace Jovanovich.

Feldman, Jack M. and John G. Lynch (1988), "Self-Generated Validity and Other Effects of Measurement on Belief, Attitude, Intention, and Behavior," Journal of Applied Psychology, 73 (August), 421-435.

Fischhoff, Baruch and Ruth Beyth-Marom (1983), "Hypothesis Evaluation From a Bayesian Perspective," Psychological Review, 90 (July), 239-260.

Langer, Ellen, Arthur Blank, and Benzion Chanowitz (1978), "The Mindlessness of Ostensibly Thoughtful Action: The Role of "Placebic" Information in Interpersonal Interaction," Journal of Personality and Social Psychology, 36, 635-642.

Petty, Richard E. and John T. Cacioppo (1986), Communication and Persuasion: Central and Peripheral Routes to Attitude Change. New York: Springer Verlag.

Spiegel, Bernt (1958/1970), Werbepsychologische Untersuchungsmethoden: Experimentelle Forschungs- und Prnfverfahren. Berlin: Duncker & Humblot.

Strebinger, Andreas (2000), "Der Marktfnhrer-Effekt in der Markenbeurteilung: Vereinfachte Kaufentscheidungen in AbhSngigkeit von Involvement, Markenbewu¯tsein und Produktwissen," Dissertation, Vienna University of Economics and Business Administration, Austria.

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