A complexity by involvement framework

Results from across the experiments in this study suggest a framework based on complexity and involvement (see Table 8-4). This framework builds on previous

66 In this meta analysis, the control condition was either a uniqueness or control set configuration, depending on the experiment, and the asymmetric dominance conditions were with two decoys. For experiment 1, both need for justification and no need for justification conditions were included. For experiments 2 and 3, the comparison was control versus asymmetric dominance. For experiment 4, individualized and aggregate alternatives conditions were included, as were both high involvement product categories. For experiment 5, medium and high raised competitor utilities conditions were included.

literature and findings from the present study. For example, work by Payne (1976) suggests that increased task complexity leads to the use of multiple choice strategies;

notably noncompensatory for screening alternatives followed by a compensatory strategy to make a choice. Work by Petty et al. (1983) also suggests that consumers shift to more elaborate processing when motivated by the task. Research regarding set configuration effects has found that the configuration of the set can directly influence choice in simple choice tasks (Hamilton, Jiewen, and Chernev 2007; Huber, Payne, and Puto 1982; Pratkanis and Farquhar 1992; Simonson 1989; Tversky 1972; Yoon and Simonson 2008). Findings in the current study, and from a previous study (Kivetz et al. 2004), suggest that set configurations can also influence choice in more complex choice tasks than typically researched. Furthermore, findings in the current study suggest that involvement may moderate these effects. More specifically, the occurrence of primary level set configuration effects in higher involving complex choice tasks, but not lower involving decisions suggests a qualification of the Bettman et al.’s (1998) argument that set configuration effects will be reduced by increased task complexity. It appears that set configuration effects may occur in complex tasks when the decision is more highly involving and not when less involving.

Alternatively, set configuration effects seem to be more likely to occur in simpler tasks when the decision is less involving and less likely when more involving. The framework below (Table 8-4) synthesizes the above findings into a complexity by involvement matrix highlighting likely set configurations influences and choice strategies in each section of the matrix.

TABLE 8-4:

A complexity by involvement set configuration effects framework

Involvement

As a task becomes more involving, consumers are likely to process more information and prefer compensatory to noncompensatory strategies as a means to choose the best, or optimal, alternative for them. However, increasing the number of alternatives to be evaluated reduces consumers’ ability to process all information. When these two forces interact, consumers are limited in the amount of information they can process, but are still motivated to make a good choice. As suggested by prior research (Bettman et al. 1998; Klemz and Gruca 2003; Olshavsky 1979; Payne 1976), consumers may begin with a noncompensatory strategy to create a set of considered alternatives, followed by another strategy to make their choice when faced with complex choice tasks. In contrast to the assertion by Bettman et al. (1998) that attraction effects are less likely to occur in such tasks, findings of an influence of asymmetric dominance and uniqueness on choices in the current study suggest that set configuration effects can occur in complex choice tasks. These set configuration influences seem to be based on the relationships between alternatives being salient.

Increased task complexity makes it more difficult to determine which alternative best

matches a consumer’s preferences. Consumers may then be motivated to seek out cues to help them with their decision. Asymmetric dominance and uniqueness may provide such cues. It seems that both uniqueness and asymmetric dominance can be salient in complex choice tasks, as suggested by results showing that respondents both observed these relationships between alternatives in the first experiment and were more likely to choose the target alternative than predicted by preferences (utilities).

For instance, as they are searching for information to aid their decision, consumers may detect the presence of asymmetric dominance and its associated good alternative (i.e. the dominating alternative): an attraction effect may occur. When prior preferences determine their set of considered alternatives, consumers may still be influenced by set configuration at a secondary level. They may have an idea of which alternatives to consider, but be uncertain of how to choose among them. When asymmetric dominance is present in the set, consumers may detect the presence of the good (i.e. target) alternative even if it is not one of their considered options. These consumers can then compare the good alternative with their considered alternatives.

This comparison can emphasize the potential for a loss and leads to selection of the considered alternative that is closest to the good alternative on an important attribute;

a secondary level effect.

When the task is less complex, on the other hand, consumers may have the cognitive resources available to employ compensatory processing and choose the alternative with the highest overall utility, as suggested by prior research (Bettman et al. 1998;

Payne 1976). In addition, research by Mishra et al. (1993) suggests that attraction effects are less likely to occur as involvement increases because consumers become more motivated to make a good choice. Increased motivation implies consumers will be more likely to use a compensatory strategy to make a good choice in more involving choice tasks. Combining these two areas of research suggests consumers in a highly involving, simple choice task are likely to use compensatory processing.

Support for compensatory processing in the current study was suggested by observed choice being in accordance with expected choice, an association between choice of alternatives and their utilities and no difference in choice of alternatives between asymmetric dominance and uniqueness conditions. Furthermore, with few exceptions, simple heuristics, such as choose the cheapest alternative, did not generally seem to

occur for high involvement categories in this study. This is an additional indication that respondents were expending some cognitive effort on the decision beyond resorting to some type of quick heuristic. Hence, set configuration effects may be unlikely to occur in highly involving, simple choice tasks and compensatory processing may be the norm.

Low involvement and complexity

In complex choice tasks that are less involving, consumers may be unwilling to expend the cognitive effort to process much information per alternative. Because they lack the motivation, they may not make the comparisons between alternatives necessary for uniqueness or asymmetric dominance to be salient. Instead, they may tend to use simple heuristics that require minimal effort to make a decision. For instance, the cheapest alternative was generally the most chosen alternative in low involving, complex choice tasks regardless of whether respondents prior preferences favored that alternative. Analyses of stated justifications for choice indicated that more respondents choosing the cheapest alternative were motivated primarily by price relative to respondents choosing other alternatives. This suggests respondents used a lexicographic strategy based on price to determine their choices. There also seemed to be several instances of respondents apparently using a choice the cheapest alternative strategy in the low involving simple tasks. This was indicated by higher than expected choice of the cheapest alternative and an analysis of respondents stated reasons for choice. In contrast to the more complex tasks, however, primary level set configuration effects should occur in low involving, simple choice tasks.⁶⁷ Regardless of not being observed in this experiment, set configuration effects have often been observed in simple tasks with few alternatives (Hamilton et al. 2007; Huber and Puto 1983; Mishra et al. 1993; Simonson 1989). Hence, it is reasonable to assume that primary level set configuration effects occur in low involving, simple choice tasks.

Regarding secondary level effects, preliminary support was found for a secondary level effect in low involving decisions from among few alternatives in Experiment 3.

67 Not finding set configuration effects in this instance in the current study is discussed earlier in this chapter.

However, no support for a secondary level effect in simple tasks was found in Experiments 4 and 5; calling the preliminary finding into question.