Article #19
2000
 
 
 
Newsletter list 

Newsletter 

Previous Article 

Next Article 


 


Interattribute Correlations Influence Whether Decision Strategies are Option-based or Attribute-based

Barbara Fasolo, Gary H. McClelland and Katharine Lange
Boulder, CO

This year our research has focused on how people make decisions amongst different options in the presence of "unfriendly" environments. In unfriendly environments, choice attributes are negatively related, and decision makers need to make trade-offs to form an overall evaluation of each option. But, trade-offs are hard to make. Our goal was to understand if decision makers ignore these trade-offs and use the same simple strategies that they would adopt in the presence of friendlier positive inter-attribute correlations, or if they adjust to the "unfriendliness" of the decision environment by using different strategies.

In two experiments, we presented decision makers with choices on Web-based Information Display Boards that were either friendly (characterized by positive inter-attribute correlations) or unfriendly (characterized by negative inter-attribute correlations). The change in correlation was manipulated between subjects in the first experiment, and within subjects in the second experiment. In both experiments, the change in correlation was implemented by simply changing the attribute values, while keeping the other choice variables constant.

We found that decision makers perceive attribute correlations in real choice situations and adjust decision strategies accordingly. In particular, when inter-attribute correlations are positive and the choice environment is friendly, decision strategies are simpler and attribute-based. The decision is judged easier, the attribute values are considered more predictable, the information search is not thorough, but is systematic. When correlations are, instead, negative and the choice environment is unfriendly, decision strategies are more effortful and option-based. The decision is judged difficult, the attribute values are more unpredictable, hence the information search is more erratic and more thorough.

This adaptivity to changes in correlation seemed to happen instantly upon "feeling" the friendliness of the choice problem. Participants who first encountered unfriendly choices were right away more option-focused than participants who first encountered friendly choices. This suggests that decision strategies are affected "online" by the correlation structure of the attribute set at hand, and not as much by pre-existing beliefs about how these attributes are usually interrelated in the real world (e.g., larger size goes with more weight).

Overall, decision makers were "fast" (decisions among five digital cameras described on eight attributes were made in less than 30s, too little time for making complex computations), "frugal" (not all information was explored, especially in friendly environments), but were not unilaterally attracted to shortcuts based on attributes. In unfriendly environments decision makers rather converged toward more effortful and option-based strategies.

In sum, we believe these two experiments provide evidence that attribute correlation powerfully affects information search strategy and ultimate choice in ways that are not always "non-normative."

This research reinforces an optimistic view of decision makers able to flexibly adapt search and decision strategies according to the structure of their decision environment.

Our next step is to see whether the same results hold in a real decision environment - the Internet - where an increasing number of users naturally encounter "friendly" and "unfriendly" information display boards and make consequential decisions from them.

Contact Barbara Fasolo

Previous Article                Next Article

Newsletter list