Selected Empirical Research on Heuristics and Decision-Making Performance
| Study and Context | Research Goal | Main Findings |
|---|---|---|
| Survey of 93 MBA students and 185 marketing managers in the U.S. (Hoch, 1988). | Ability of experts to predict the opinions of consumers compared with novices, with neither group having access to data. | Experts were no better than novices at using their intuition. Wrong decisions involving intuition cannot be subject to feedback. |
| Survey of 16 academics, 12 practitioners, and 43 high school students in the U.S. (Armstrong, 1991). | Differences between predictions of experts and novices. | No significant differences between any groups. |
| Two experiments with 114 and 59 MBA students in the U.S. (Cripps and Meyer, 1994). | Investigation of how consumers plan for the replacement of durable goods compared with optimal machine-replacement theory. | Subjects made persistently suboptimal decisions, based upon a conservative heuristic, favoring obsolescence-motivated replacement of durables than better performing alternatives. |
| Two experiments with 220 students and volunteers in Germany (Bröder, 2003). | Is the use of heuristics in decision making adaptive, and does information acquisition correspond to decision strategies? | Choosing the appropriate heuristic requires a meta-heuristic that integrates and evaluates cues from the environment that convey information about its payoff structure. Saving costs in the long run were traded off against not always choosing the best option, for example. |
| Three experiments with 52 students and faculty in the U.K. (Newell and Shanks, 2003). | Assessment of the parameters of the take-the-best (TTB) heuristic by offering the purchase of more information before the choice is finalized. | Complex decisions processes can be performed in simple ways. Although TTB is powerful, it is not the universal decision-making tool. |
| Survey of 561 projects containing 499 failures and 62 commercial successes in 1989–1994, including 1,143 entrepreneurs in Canada (Åstebro and Elhedhli, 2006). | Understanding why Canadian Invention Assistance Program analysts correctly forecast the likelihood that an invention would reach the market as often as or more often than linear additive statistical models. | Analysts use simple sums of counts, using significantly more cues than typically observed. The conjunctive model predicted 86 percent. Experts, however, correctly predicted 83 percent, significantly outperforming a log-linear additive statistical model (79 percent). |
| Three experiments with 497 students in the U.S. (Saini and Monga, 2008). | Investigation of whether consumer decision making is more heuristic when it comes to spending time rather than money. | Time and money might be seen as the same by economists, but in practice, consumers more likely will use heuristics to reduce time, whereas they more likely will use algorithms when it comes to spending money. |
| Analysis of three datasets including information on 2,330 apparel, 2,891 airline, and 2,357 CD customers in the U.S. (Wübben and Wangenheim, 2008). | Comparison of the outcomes of stochastic models versus the heuristics used by firms to predict future purchases. | Complex methods were only slightly superior to heuristics in terms of determining the (in)activity of customers, and there was no clear evidence that such models are superior to heuristics. |
| Three experiments with 530 undergraduates in the U.S. (Hutchinson, Alba, and Eisenstein, 2010). | Differences between optimal marketing-budget allocations and those predicted by heuristics. | Data-based inferences are subject to strong, heuristic-based biases that are not reduced by graphical presentations of the data, “real world” experience, or explicit training. |