The Effect of Three Contingency Factors on Consumer Choice Strategies: a Test of Awareness of Costs and Benefit

Elizabeth Cooper-Martin, Georgetown University
ABSTRACT - The choice strategy a decision-maker uses is contingent on many factors. This paper explores the effect of three contingency factors: product class involvement, similarity among alternatives and information structure. One explanation for contingency effects is that decision-make s select among choice strategies according to costs (time and effort) and benefit (choosing the best alternative). This assumes some awareness of these costs and benefit. To test this assumption, this paper studies choice strategies with a self report and an objective measure, for both costs and benefit. Each contingency factor has the same effect on both measures, for costs or benefit. These results support the assumption of awareness.
[ to cite ]:
Elizabeth Cooper-Martin (1989) ,"The Effect of Three Contingency Factors on Consumer Choice Strategies: a Test of Awareness of Costs and Benefit", in NA - Advances in Consumer Research Volume 16, eds. Thomas K. Srull, Provo, UT : Association for Consumer Research, Pages: 130-136.

Advances in Consumer Research Volume 16, 1989      Pages 130-136

THE EFFECT OF THREE CONTINGENCY FACTORS ON CONSUMER CHOICE STRATEGIES: A TEST OF AWARENESS OF COSTS AND BENEFIT

Elizabeth Cooper-Martin, Georgetown University

ABSTRACT -

The choice strategy a decision-maker uses is contingent on many factors. This paper explores the effect of three contingency factors: product class involvement, similarity among alternatives and information structure. One explanation for contingency effects is that decision-make s select among choice strategies according to costs (time and effort) and benefit (choosing the best alternative). This assumes some awareness of these costs and benefit. To test this assumption, this paper studies choice strategies with a self report and an objective measure, for both costs and benefit. Each contingency factor has the same effect on both measures, for costs or benefit. These results support the assumption of awareness.

INTRODUCTION

Choice strategies are the processes used to evaluate alternatives and select one. One of the major findings from research on decision making is that such strategies appear to be a contingent behavior; i.e., what choice strategy the decision-maker uses depends on characteristics of the decision task (Einhorn and Hogarth 1981). Variables that may affect choice strategies are referred to as contingency factors. Many of these contingency effects can be explained by a cost/benefit framework (Payne 1982). It proposes that selection of a choice strategy is a function of a strategy's costs and benefit, compared to alternative strategies (Beach and Mitchell 1978; Johnson and Payne 1985; Payne, Bettman and Johnson 1988; Russo and Dosher 1983; Wright 1975). Typically, the costs are time and cognitive effort to make the choice and the benefit is choosing the best alternative from the choice set.

As noted by Bettman (1988), a cost/benefit framework assumes that the decision-maker has some awareness of both the effort and the accuracy of his/her choice strategies. A choice strategy's accuracy is the likelihood that it will lead to selecting the best alternative. But these processes have not been studied. Thus Bettman called for research on self reports of accuracy and on how characteristics of the choice task affect these self reports. This paper addresses this need for further research on awareness of accuracy, as well as examining awareness of effort.

Specifically, this study collected a self report on the extent to which consumers chose the best alterative and one on the cognitive effort expended by consumers in their choice strategies. A more objective measure for costs and one for benefit were also collected. To explore the assumption of awareness, the effects of contingency factors on both self reports and objective measures were tested. This study included three contingency factors: product class involvement, similarity among alternatives and information structure. They are described more fully below.

CONTINGENCY FACTORS

There are three types of contingency factors which can affect choice strategies. task and context variables (Payne 1982) and individual characteristics (Beach and Mitchell 1973). The latter (e.g., expertise) describe the decision-maker. Task variables (e,, number of attributes) are general, structural features of the choice set. Context variables (e.g., quality of options) are factors associated with the particular alternatives in a set. One of each we was included in this study.

The individual characteristic studied was product class involvement. Et characterizes the relationship between a consumer and a product class and reflects the product's importance, perceived risk and symbolic value (Laurent and Kapferer 1985). Product class involvement is enduring, in contrast to situational involvement which is concern for a specific situation or purchase occasion (Richins and Bloch 1986). Higher product class involvement increases decision times (e.g., Tyebjee 1979).

The task variable examined was information structure. It refers to the organization of alternatives in a choice set (Van Raaij 1977). This contingency factor affects decision time (e.g., Bettman and Zins 1979) and attribute versus brand processing (e.g., Biehal and Chakravarti 1982). (Attribute processing is examining information for one attribute across several alternatives. Brand processing is examining information for one alternative across several attributes).

This study tests a new type of information structure, hierarchical. It is sometimes used by retailers. An example is a display of sweaters with all the navy ones on the left, all turquoise ones in the middle and all green ones on the right. Within a color group, all long-sleeved sweaters are on the left and all sleeveless ones are on the right. Thus, in a hierarchical structure, all alternatives with the same level on one attribute, i.e. color, are grouped next to each other. Within these groups, all alternatives with the same level on a second attribute, i.e. sleeve style, are contiguous. This can be continued with other attributes, but which is first, second, etc. is arbitrary.

The context variable, similarity among alternatives, is also tested. Similarity between two alternatives increases with the number of attributes common to both and decreases with the number of attributes unique to each (Tversky 1969). Several studies have shown that this variable affects which alternative in a set is chosen (e.g., Huber, Payne, and Puto 1982). But relatively little research has explored the choice strategies behind this effect (Biggs et al 1985; Russo and Dosher 1983).

HYPOTHESES

The purpose of this study is to explore whether decision-makers are aware of the costs and benefit of their choice strategies. It is assumed that if self awareness of costs (or benefit) changes in response to a contingency factor in the same way as actual CostS (or benefit), then the decision-maker is aware of the characteristics of his/her choice strategy. Thus, two general hypotheses are proposed.

FIGURE

EXAMPLE OF A HIERARCHICAL STRUCTURE

H1 The effect of a contingency factor on self awareness of choosing the best will be the same as that factor's effect on choosing the best.

H2 The effect of a contingency factor on self awareness of choice strategy costs will be the same as that factor's effect on choice strategy costs.

METHOD

To test these hypotheses, an experimental study of choice was conducted. It focuses on choice strategies during actual choices from sets of real, physically present products. Such products were used to simulate a normal shopping environment. The data reported are from a larger study of the effect of contingency factors on consumer choice strategies. For a more complete description of the method, see Cooper-Martin (1988).

Design

Two levels of each contingency factor were tested: product class involvement (high and low), information structure (hierarchical and random) and similarity among alternatives (similar and dissimilar). Operationalizations follow. (Scales for all measures are available from the author).

Product class involvement was measured with four scales based on those of Laurent and Kapferer (1985). The scales were product class importance, importance of consequences of purchase, symbolic value and subjective probability of a bad purchase. Based on a pretest with the sample population, sweaters were used to elicit high involvement and coffee mugs to elicit low involvement. All subjects were more involved with sweaters than mugs on all aspects of involvement (t tests, df = 35, p < 0.001 for all tests).

Similarity among alternatives was measured by the number of attributes shared by alternatives. In the similar sets, any pair of alternatives varied on a maximum of three attributes and a minimum of one. In the dissimilar sets, any pair of alternatives varied on a maximum of seven attributes and a minimum of four. Each set contained 12 alternative . There were three similar sets and one dissimilar set for each product class. Subjects rated the dissimilar set less similar than each similar set for both sweaters and coffee mugs (t tests, df = 35, p < 0.001 for all tests). (Attribute values 'or all sets are available from the author).

Information structure was determined by how the products were arranged. For the hierarchical structure, products were arranged so that all those with the same level on one attribute, e.g., feature, were contiguous (see Figure). Within a group with the same level on this attribute, products were arranged left to right by level on a second attribute, e.g., color, and so on for a third attribute, e.g., handle. To generalize, each subject saw a different hierarchical structure. In the random structure, order was determined by random numbers. To generalize, six different random structures were used.

Six choice sets were tested: random dissimilar, random similar and hierarchical similar for both low product class involvement and high product class involvement.

Subjects

The subjects were a convenience sample of 36 female consumers from metropolitan New York City. All subjects were the same sex because clothing was one of the products used. Ages ranged from 18-70; mean age was 34. Household income ranged from $7500-$14000 to over $100,000; median level was $25,000-$39,000. Level of education ranged from some high school to doctorate degree; the median was some graduate study. As an incentive, subjects received $25.

Procedure

Three to four weeks before the experiment. each subject answered questions on involvement with five product classes, to conceal the experimental ones. Each subject was run individually. The design was completely within-subject; each subject chose from all six choice sets but saw a different set of alternatives for each one. For each choice, she selected a product for herself, but did not keep her selection and provided a concurrent verbal protocol. After every choice, she gave two self reports: one on her choice strategy's costs and one on its benefit.

Dependent Measures

The dependent variables were the benefit of choosing the best and choice strategy costs. There were two measures for each. To measure the decision-maker's awareness, self reports were used. Each self report was a multiple-item scale, developed specifically for this study. To measure actual benefit and costs, a more objective measure was also used for each.

Choosing the best. The self report on choosing the best contained questions that reflected the consumer's use of a choice criterion of choose the best alternative (see the Appendix, panel A).

Choosing the best requires considering all "subjectively nontrivial dimensions" (Wright 1975, p. 61). Thus the objective measure of choosing the best was the number of important attributes examined. They were operationalized as attributes that could affect choice (Myers and Alpert 1977). Before the experiment, each subject listed attributes that were important when choosing sweaters (or mugs) and rated whether each attribute was different across different sweaters (or mugs). Attributes rated as having a big difference were labeled important. For each choice, the measure equalled the number of attributes labeled important and mentioned in that choice's protocol.

Choice strategy cost,. The self report on choice strategy costs consisted of questions that concerned cognitive effort (see the Appendix, panel B).

The objective measure of choice strategy costs was decision time. Other researchers have used (Christensen-Szalanski 1978, 1980) or suggested decision time as a measure of choice strategy costs (Beach and Mitchell 1978; Russo and Dosher 1983). The actual measure was the number of seconds between the moment the subject saw all the alternatives and the moment when she indicated her chosen alternative.

RESULTS

The contingency factor effects were tested separately for each dependent measure with a 2 (product class involvement) X 3 (similarity among alternatives combined with information structure) repeated measures ANOVA. Both factors were within subject. The main effect for involvement tested the impact of involvement. Separate Helmert contrasts were used to test the effect of similarity and the effect of information structure (Bock 1975). The effect of each contingency factor was tested with two levels of a within-subject factor. Hence, the homogeneity assumption was satisfied and only univariate results are presented (LaTour and Miniard 1983). All results are shown in the Table.

Choosing the Best

Product class involvement. As expected, the effect of product class involvement on choosing the best is the same for both measures. Both self report on choose the best (F(1,35) = 7.46, p < 0.01) and important attributes examined (F(1,333 = 5.15, p < 0.05) are greater for high versus low involvement.

Similarity among alternatives. As expected, similarity has the same effect on the two measures of choosing the best. Both self report on choose the best (F(1,35) = 5.04, p < 0.05) and important attributes examined (F(1,33) = 6.95, p < 0.01) are lower for similar than for dissimilar sets. These results indicate subjects choose the best less when selecting from a set of similar alternatives.

However, the interaction between product class involvement and similarity is significant only for self report on choose the best (F(1,35) = 4.76, p < 0.05). This scale is lower for similar than dissimilar sets for both levels of involvement; the difference is greater for low than for high involvement. (For choosing the best measures, no other interactions are significant, p > 0.10.).

Information structure. As predicted, information structure has the same impact on both measures of choosing the best. Neither self report on choose the best (F(1,35) = 0.17, p > 0.10) nor important attributes examined (F(1,35) = 3.44, p > 0.10) varies between random and hierarchical structures.

Choice Strategy Costs

Product class involvement. As expected, both measures of choice strategy costs show the same response to product class involvement. Self report on costs (F(1,35) = 4.09, p < 0.07) and decision time (F(1 35) = 13.04, p < 0.01) are each greater for high than for low product class involvement. (For the cost measures, no interactions are significant, p > 0.10.).

Similarity among alternatives. As hypothesized, similarity among alternatives in a set affects the two choice strategy costs in the same way. Self report on costs (F(1,35) = 8.52, p < 0.01) and decision time (F(1,35) = 18.29, p < 0.01) are both lower when choosing from a set of similar alternatives than from a set of dissimilar ones.

Information structure. As predicted, information structure has the same effect on both measures of choice strategy costs. Neither cost varies between a hierarchical structure and a random one: self report on costs (F(1,35) = 0.14, p > 0.10) and decision time (F (1,35) = 2.56, p > 0.10).

Discussion

Before summarizing, it is useful to note the study's limitations. The sample was unrepresentative; it was female, highly educated, high income and from one geographic region. Although differences between choices should be unaffected, absolute levels of behavior may differ from other samples. Also, subjects did not choose in a natural setting. But they reported relatively high consistency between their normal choice strategies and those in the study.

To summarize, each of the contingency factors has the same effect on the self reports as on the more objective measures, for both choosing the best and choice strategy costs. These results support the cost/benefit framework assumption that decision-makers do have some awareness of the costs and benefit aspects of a choice strategy. For each aspect, the self report and the more objective measure both changed in the same direction, or both did not change. These results suggest that subjects' awareness of their behavior coincided with their actual behavior. Further, she self reports increased or decreased between different levels of two contingency factors (i.e. product class involvement and similarity among alternatives). These changes support Bettman's (1988) suggestion that decision-makers adapt their choice strategies to different choice conditions via tradeoffs between accuracy and effort.

TABLE

MEAN VALUE OF BENEFIT AND COST MEASURES BY CHOICE CONDITION

The specific effects of the contingency factors can be summarized as follows. Higher levels of product class involvement and lower levels of similarity among alternatives increase choosing the best and choice strategy costs. The third contingency factor, information structure, does not affect choosing the best or choice strategy costs.

The results on product class involvement confirm past work which found that decision time rises with involvement (Jacoby, Chestnut and Fisher 1978; Tyebjee 1979).

The results on similarity among alternatives are the first to include choice strategy costs. Previous work studied similarity's effect on brand versus attribute processing (Russo and Dosher 1983) or on compensatory versus noncompensatory strategies (Biggs et al 1985). Compensatory strategies were defined as ones in which the decision-maker examined all attributes for each alterative. Such a strategy should have higher costs than a noncompensatory one, in which the decision-maker eliminated some alternatives. Biggs et al found compensatory strategies to be more frequent in similar than dissimilar sets, which suggests that choice strategy costs are higher for similar sets. In contrast, this study found choice strategy costs are lower for similar sets.

One possible explanation is that the effect of similarity depends on the number of alternatives in the set. The assumption of an interaction between these two variables would explain the contradiction. Biggs et al (1985) used three alternatives per set and this study used 12.

The third contingency factor studied was information structure. Its lack of effect on choice strategy costs seems surprising given previous work. Several studies have found that information structure affects attribute versus brand processing (e.g.. Biehal and Chakravarti 1982). These measures reflect the order or structure of information processing, whereas measures of costs and choosing the best do not. Perhaps information structure only affects choice strategy measures that reflect task or structure.

Past work has also found that information structure affects decision time (e.g., Bettman and Zins 1979). These studies used written, verbal descriptions as stimuli and tested two structures: attribute and brand. The latter presents all the attributes on one alternative together, e.g., car by car. The former presents information on each attribute for all alternatives together. With written descriptions, decision times are longer with an attribute than with a brand structure (Bettman and Kakkar 1977; Bettman and Zins l 979).

The current study used real, physically present products as stimuli. With them, a hierarchical structure is the information structure closest to an attribute structure and a random structure is closest to a brand structure. But decision times do not differ between hierarchical and random structures. Nor do these two structures vary that much, due to the stimuli. In both, information is still mainly organized by brand. Thus the lack of differences in choice strategy costs between these two structures may reflect the fact that different arrangements, i.e., information structures, of real, physically present products are not distinct enough to affect choice strategies.

Thus a study that directly compares choice strategies from real, physically present products to those from verbal descriptions would give more specific information about possible biases from either stimuli. Future work could also determine whether brand and attribute structures using verbal descriptions affect choosing the best or costs other than time.

IMPLICATIONS AND CONCLUSIONS

The findings on the effects of the contingency factors should aid both policy-makers and marketing managers. The former want consumers to make good choices and should be interested in helping them choose the best. Manufacturers and retailers want consumers to buy, which is likely to mean choose the best. To increase choosing the best for a product category, dissimilar sets of alternatives in a random structure appear most useful. A random structure seems just as effective as a hierarchical one. (Random and hierarchical seem to be the only, feasible structures with real, physically present products).

The results on choice strategy costs can also be useful. For example, similar sets could make decisions easier for those with less processing ability, e.g., children. A concession stand trying to maximize service during intermission should offer similar alternatives, to encourage faster decisions. But if managers want consumers to linger, e.g., in a shopping mail, then displays should contain dissimilar alternatives.

In addition to the research ideas already mentioned, there is a need for more research to understand decision-maker's awareness of the costs and benefit generated by their choice strategies. The present study did not examine h(-w well-calibrated the self reports were; this would be useful to confirm the current findings. Clearly, the effects of additional contingency factors could be tested. However, the purpose of studying self awareness of choice strategy costs and benefit is to better understand how subjects adapt to contingency factors. This study used self reports taken after the choice strategies; self reports generated during the choice strategy might tell more about adaptation processes.

In conclusion, this study supports the assumption that decision-makers have some awareness of the costs and benefit of their choice strategies. This finding supports a cost/benefit framework which suggests that decision-makers select a choice strategy based on the costs and benefit of various choice strategies. The results also extend our knowledge about the effects on choice strategy costs and benefit of three contingency factors: product class involvement, similarity among alternatives and information structure.

APPENDIX

SELF-REPORTS ON CHOICE STRATEGIES

REFERENCES

Beach, Lee Roy and Terence R. Mitchell (1978), "A Contingency Model for the Selection of Decision Strategies," Academy of Management Review, 3 (July), 439-44.

Bettman, James R. (1988), "Processes of Adaptivity in Decision Making." in Advances in Consumer Research, Vol. 15, ed. Michael J. Houston, Provo, UT: Association for Consumer Research, 1-4.

Bettman, James R. and Pradeep Kakkar (1977), "Effects of Information Presentation Format on Consumer Information Acquisition Strategies," Journal of Consumer Research, 3 (March), 233-240.

Bettman, James R. and Michel A. Zins (1979), "Information Format and Choice Task Effects in Decision-Making," Journal of Consumer Research, 6 (September), 141 - 153.

Biehal, Gabriel and Dipankar Chakravarti (1982), "Information Presentation Format and Learning Goals as Determinants of Consumers' Memory Retrieval and Choice Processes," Journal of Consumer Research. 8 (March). 431441.

Biggs Stanley F., Jean C. Bedear, Brian G. Gaber and Thomas J. Linsmeier (1985), 'The Effects of Task Size and Similarity on the Decision Behavior of Bank Loan Officers," Management Science, 31 (August), 970-987.

Bock, R. Darrell (1975), Multivariate Statistical Methods in Behavioral Research, New York: McGraw-Hill.

Christensen-Szalanski, Jay J. J. (1978), "Problem Solving Strategies: A Selection Mechanism, Some Implications, and Some Data," Organizational Behavior and Human Performance, 22(October), 307323.

Christensen-Szalanski, Jay J. J. (1980),"A Further Examination of the Selection of Problem Solving Strategies: The Effects of Deadlines and Analytic Aptitudes," Organizational Behavior and Human Performance, 25 (February), 107- 122.

Cooper-Martin, Elizabeth (1988), "The Effect of Product Class Involvement, Similarity Among Alternatives, Information Structure, and Time on Consumer Choice Processes," working paper no. 88-07, School of Business Administration, Georgetown University, Washington, D.C.

Einhorn, Hillel J. and Robin M. Hogarth (1981), "Behavioral Decision Theory: Processes of Judgment and Choice," Annual Review of Psychology, 32, 52-88.

Huber, Joel, John W. Payne and Christopher Puto (1982), "Adding Asymmetrically Dominated Alternatives: Violations of Regularity and the Similarity Hypothesis", Journal of Consumer Research, 9 (June), 90-98.

Jacoby, Jacob, Robert W. Chestnut and William A. Fisher (1978), "A Behavioral Process Approach to Information Acquisition in Nondurable Purchasing", Journal of Marketing Research, 15 (November), 53244.

Johnson, Eric J and John W. Payne (1985), "Effort and Accuracy in Choice," Management Science, 31 (April), 395-414.

Latour, Stephen A. and Paul W. Miniard (1983), 'The Misuse of Repeated Measures Analysis in Marketing Research," Journal of Marketing Research, 20 (February), 45-47.

Laurent, Gilles and Jean-Noel Kapferer (1985), "Measuring Consumer Involvement Profiles," Journal of Marketing Research, 22 (February), 4153.

Myers, James H. and Mark I. Alpert (1977), "Semantic Confusion in Attitude Research: Salience vs. Importance vs. Determinance," in Advances in Consumer Research, Vol. 4, ed. William D. Perreault, Jr., Atlanta, GA: Association for Consumer Research, 106-110.

Payne, John W. (1982), "Contingent Decision Behavior," Psychological Bulletin, 92 (2), 382-402.

Payne, John W., James R. Bettman, and Eric J. Johnson (1988), "Adaptive Strategy Selection in Decision Making," Journal of Experimental Psychology: Learning, Memory and Cognition, 14 (July), 534552.

Richins, Marsha L. and Peter H. Bloch (1986), "After the New Wears Off: The Temporal Context of Product Involvement," Journal of Consumer Research, 13 (September), 280-285.

Russo, J. Edward and Barbara Dosher (1983), "Strategies for Multiattribute Binary Choice," Journal of Experimental Psychology: Learning, Memory, and Cognition, 9 (October), 676-696.

Tversky, Amos (1969), "Features of Similarity," Psychological Review, 84, 4 (July), 327-352.

Tyebjee, Tyzoon T. (1979), "Response Time, Conflict, and Involvement in Brand Choice," Journal of Consumer Research, 6 (December), 295-304.

Van Raaij, W. Fred (1977), "Consumer Information Processing for Different Information Structures and Formats," Advances in Consumer Research, 4, ed. William D. Perreault, Jr., Atlanta: Association for Consumer Research, 176-184.

Wright, Peter (1975), "Consumer Choice Strategies: Simplifying vs. Optimizing," Journal of Marketing Research, 11 (February), 60-67.

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