Exploring the Nature of Consumer Information Processing Heuristics

Christian Pinson, INSEAD, (The European Institute of Business Administration, Fontainebleau, France)
ABSTRACT - The present paper attempts to place four recent investigations of consumer cognitive processes within the broader context of consumer information processing heuristics. A few suggestions are made to advance our understanding of consumer judgmental performance.
[ to cite ]:
Christian Pinson (1983) ,"Exploring the Nature of Consumer Information Processing Heuristics", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 476-478.

Advances in Consumer Research Volume 10, 1983      Pages 476-478

EXPLORING THE NATURE OF CONSUMER INFORMATION PROCESSING HEURISTICS

Christian Pinson, INSEAD, (The European Institute of Business Administration, Fontainebleau, France)

ABSTRACT -

The present paper attempts to place four recent investigations of consumer cognitive processes within the broader context of consumer information processing heuristics. A few suggestions are made to advance our understanding of consumer judgmental performance.

INTRODUCTION

There is well documented evidence that people oversimplify their experience by selectively attending to certain features of the information available in their environment and by using a variety of heuristics to process incoming stimuli. The impact of preconceptions on information processing is one of the better demonstrated findings in twentieth century psychology. Evidence for this phenomenon comes from examinations of the effects of stereotyping schemas, and lay theories and biases on judgments of individuals and objects (e.g., Kahneman, Slovic, Tversky 1982; Nisbett and Ross 1980).

The four papers presented in this session are good illustrations of consumer behavior researchers' attempts to understand the types of schematic knowledge structures and processes that provide consumers with a simplistic and useful, though often erroneous, interpretation of their environment.

THE MAINTENANCE OF STEREOTYPIC BELIEFS

Vivian Peeters convincingly documents the biasing impact of stereotypic beliefs on the encoding and retrieval of information and highly stresses that such beliefs are extremely resistant to most attempts to change them even when the evidence on which they are based has been totally discredited. The main reason given for this persistence is that people tend to seek confirming evidence and to ignore disconfirming information.

The paper by Jennifer Crocker and Renee Weber takes a broader and apparently more optimistic view of stereotypes. These authors describe three possible models of stereotype change. The bookkeeping model assumes willingness to gradually incorporate disconfirming information into existing stereotypes. As disconfirming evidence accumulates, the individual may eventually shift the default attribute(s) of the stereotype. The conversion model portrays stereotype change as occurring suddenly in response to new dramatic, disconfirming instances. In the third model, called the subtyping model, change is viewed as occurring through a process of categorical differentiation. As additional information is acquired a sub-type is formed whenever the incoming information is inconsistent with the existing stereotype. This process of "gradual restructuring serves the purpose of accommodating' incongruent instances - when the accumulated evidence becomes too incongruent established stereotypes may be modified. If the information does not deviate too much from the existing stereotypes, it may be retained as exceptions, i.e., unrepresentative instances of the overall stereotypic category.

Which of these three models best describe stereotype change? The empirical studies carried by Crocker and Weber are interpreted by these authors as generally supporting the use of the subtyping model. Subjects tended to subtype the inconsistent instances into a subcategory of "exceptions". What conclusions can be drawn from these two papers? The overall impressions that one gets is that people are remarkably able in resisting attempts to change their stereotypic beliefs. The literature review done by Peeters casts doubt on the willingness of subjects to gradually accumulate disconfirming information. Her conclusion should be reinforced after reading the studies reported by Crocker and Weber. If subjects predominantly follow the subtyping motel, this should be interpreted as indicating a strong tendency to disregard incongruent instances rather than to integrate them.

In the following discussion, several general issues implicitly raised by the two papers on stereotypes will be briefly addressed.

A first issue has to do with the distinction made between the bookkeeping, conversion and subtyping models. Are they alternative models of stereotype change or rather do they represent the various stages of what wight be called the stereotype change process? The bookkeeping model seems to correspond to a situation where the individual is still able/willing to assimilate inconsistent information into the stereotype. When the incoming evidence becomes too inconsistent, the subject may either totally ignore it or retain it as a sub-type, i.e., as an exception or deviation which does not alter the stereotype. However, if the incongruence becomes too drastic the individual can no longer reject it and may be forced to change his stereotype. The three models may then be conceived of as three stages of a "hierarchy of cognitive/affective effects". The first stage is characterized by an assimilation process, the second stage by an accommodation process, and the third stage by a conversion process. To change an existing stereotypic belief one needs to move the individual up the above cognitive/affective ladder. What are the implications of this proposed re-conceptualization of stereotype change?

If stereotype change occurs through a sequence of processes (assimilation + conversion) future research should not address the issue of which model best describe stereotype change but rather under what conditions which model will operate. More specifically it is incumbent on theorists in this area to account for the following factors:

a. the extent to which existent stereotypes are distorted views of "reality"

b. the strength of existing stereotypes

c. the process which led to their formation. Should the same process be reconstructed to change them?

d. the personal and social functions served by these stereotypes

e. the similarity or dissimilarity between individual and (reference) group stereotypes.

Knowing more about these factors should allow researchers to better understand the nature, amount, direction, and process of change required for each individual/situation dyad.

Another important issue which does seem to be appropriately dealt with in the literature concerns what happens after a stereotype has been successfully changed. If the function that the previous stereotype used to serve has not changed, will this not lead to the emergence of new forms of stereotypes? These new stereotypes may be more difficult to detect and change than the original ones.

In short one should avoid the trap of developing a stereotypic view of stereotypes as being nothing but useless or even destructive distortions of reality. If stereotypes exist, it is probably because they correspond to a personal and/or social function. This suggests that the study of stereotype change should be broadened to incorporate the "veridicality" required by the task environment. Therefore the personal and social usefulness of stereotypes should be raised and researchers should not "a priori" exclude from their paradigm the possibility of finding that certain stereotypes prove "useful" or heuristically acceptable - under certain circumstances. Should these circumstances change, the individual may then need to undergo a process of stereotype change.

HEURISTIC CUE UTILIZATION

The various traditions in cue utilization research can be conveniently represented by Figure 1.

FIGURE 1

A TYPOLOGY OF CUE UTILIZATION BEHAVIORS

"Rational cue utilization" (cell I) refers to situations where individuals are willing and able to use cues with high predictive value and high confidence value. The cue utilization behavior is usually found in low involvement/low risk judgmental tasks. When it occurs in high involvement/high risk judgmental situations its disfunctionality makes it very costly for the individual.

The paper by Tony Schellinck deals with the "conservative" and "risk-taking" heuristics (cells II and III) that people resort to when they do not have access to cues with high predictive/high confidence values. The key and fascinating question raised by this research is formulated as follows: why do consumers often rely on cues which they admit themselves are poor predictors of the characteristics under consideration?

The study reported by Schellinck investigates two situational factors, time and perceived risk. It is hypothesized that (1) as time pressure increases there is a greater tendency to use high confidence value/low predictive value cues (cell III in Fig. 1) and as perceived risk increases there is a tendency to depend on high predictive value/low confidence cells (cell II in Fig. 1). The main finding was that time pressure had a significant effect, that the perceived risk effect was not significant and that the time X perceived risk interaction was not really significant.

Several conceptual and methodological problems exist in this study, some rather minor. Perhaps the most serious of these is the total elimination of personal factors from the study's design and analysis. It is difficult to conceive that a subject's choice of certain cues could be done in isolation of cognitive and affective characteristics. One can very well understand and agree with the investigator's decision to focus his study on two situational factors (i.e., time and perceived risk) at the exclusion of other possible variables. What seems less defendable is the lack of attention given to these 'third factors" in the interpretation of the findings.

The failure to find a significant effect for the perceived risk variable and for the perceived risk X time pressure interaction may be explained by the operationalization of perceived risk. Risk was manipulated by, a) using two different prices ($20 vs. $450) and b) using two different reports from a highly credible source indicating a high vs. small likelihood of satisfactory product performance. Although the manipulation checks indicated a strong difference in perceived risk (p < .0001), this does not guarantee that subjects affectively internalized the cognitive meaning of the perceived risk difference. One can doubt that students enrolled in an introductory organization behavior class can identify with the task of selecting papers - the product retained for the investigation. Another serious limitation has to do with the fact that subjects were instructed to select only four out of the eight cues available for each brand. Such a rigid instruction eliminates any possible individual differences in the number of cues that subjects would normally select in an unconstrained situation. Another related issue is that the 8 cues presented did not include any high confidence value/high predictive value cues nor any low confidence value/low predictive value cues. The data available to subjects were again artificially restricted.

As to cue utilization (as opposed to cue selection) behavior the findings reported by Schellinck are quite surprising. Figure 3 of his paper indicates that the number of high predictive value cues is highest . . . in the low perceived risk/low time pressure situation. Such a finding does not seem to have face validity and one can only speculate about its possibly artifactual origin. Subjects were asked to report the cues that they actually used in arriving at their decision. This suggests the possible existence of a social desirability bias. Another tentative explanation is that because of some of the problems mentioned earlier the whole experiment did not have sufficient ecological and psychological representativeness.

Finally in the last part of his study, Schellinck shows that subjects using high confidence value/low predictive value cues tend to more often pick up a lower quality product when asked to select the brand with the highest quality. Although such a finding seems congruent with what one would normally expect, it is intriguing to note that this lower quality brand (brand 3) had by are the largest market share. This seems to indicate that real life subjects may not behave according to the research paradigm used by Schellinck.

CONSUMER SUBJECTIVE COGNITIVE CLASSIFICATIONS

Banwari Mittal presents a conceptual framework which describes the cognitive and affective processes underlying consumers' structuring of their product environment. This structuring or ordering is guided by a principle of parsimony and is directed at improving the individual's efficiency in dealing with his information processing environment. Within this framework the consumer choice process is described as involving two stages.

In the first stage (the elimination stage) the consumer eliminates those alternatives which do not satisfy certain criteria. In the second stage (the selection stage) the consumer attempts to select the "best" alternative. Mittal proposes that the process of "cognitive classification" is a means of implementing the elimination stage. Re suggests that consumers sort products in a decision-tree-like manner. These decision-tree-like classifications are described as "pathways" leading to a category including one's final choice (to be the "own category"). The second stage of the consumer choice process involves either a "minimum effort" strategy or an "information integration" strategy depending on whether the produce to be judged is trivial or important.

Before going further in the analysis of Mittal's paper, a few critical comments seem necessary.

Firstly, one may question the assumption made by Mittal (as well as others) that consumers necessarily use a two-stage information process. One could easily imagine that when simple products are involved and/or when experience accumulates, subjects - and particularly those who are cognitively complex - may employ a one stage process. Secondly the notion of 'own category' proposed by Mittal is not sufficiently well defined. Many readers will find it difficult to distinguish it from the familiar concept of 'evoked set'. Mittal explains that "own category' is specified by criterial dimensions whereas, as he sees i, 'evoked set- focuses more on the category member alternatives. The specific nature of these criterial dimensions is not clearly spelled out and one can hardly imagine that they would not contain some form of evaluation.

Another set of comments has to do with the nature and determinants of the 'pathways-. Future research should suggest measures to characterize these pathways and attempt to identify the situational (e.g., product involvement, product complexity, time pressure) and personal (cognitive/affective traits) factors that have an impact on the existence and use of these cognitive structures.

The second part of Mittal's paper focuses on the affective processes underlying product selection. A distinction between 'expressive" vs. 'functional' products is introduced. Expressive products are processed in a holistic or global manner whereas "functional" products are assumed to lend themselves to more elaborate, more conscious processing. The distinction between affective vs. cognitive processing is not new but yet interesting. However, it is still conceptually and methodologically extremely tenuous. More research is called for, before it can prove really useful as products are never purely' expressive or functional - a point which is acknowledged by Mittal himself.

GENERAL CONCLUSION

The four papers appropriately stress the importance of doing more research on consumer judgmental heuristics. They also implicitly point out to a need to better measure their information processing performance in order to eventually correct it.

In that respect the distinction introduced by Norman and Bobrow (1975) may prove useful. These authors show that active processes compete for limited processing resources. Consumers may be constrained in their performances because of limits in the amount of available processing resources (such as motivational, affective and cognitive) or by limits in the quantity/quality of data available to them.

A variety of measures of information processing performance can also be used. A few are proposed below:

a. veridicality: the extent to which the judgment is accurate and based on an isomorphic representation of the phenomenon under study

b. optimality: the extent to which the judgment is close to the judgment that an expert (expert's rule) or a majority of people (majority's rule) would arrive at

c. consistency the extent to which the final judgment is consistent with prior steps in the judgmental sequence

d. stability: the extent to which the judgment is stable over time under similar task conditions

REFERENCES

Kahneman, D., Slovic P. and Tversky, A., eds. (1982), Judgment Under Uncertainty: Heuristics and Biases. Cambridge: Cambridge University Press.

Nisbett, R. and Ross, L. (1980), Human Inference: Strategies and Shortcomings of Social Judgment, Englewood Cliffs, N.J.: Prentice Hall.

Norman, D.A. and Bobrow, D.G. (1975), 'On Data-Limited and Resource-Limited Processes , Cognitive Psychology, 7, 44-64.

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