Summary

The lexicographic decision rule is one of the simplest methods of choosing among decision alternatives. It is based on a simple priority ranking of the attributes available. According to the lexicographic decision rule, a decision alternative is better than another alternative if and only if it is better than the other alternative in the most important attribute on which the two alternatives differ. In other words, the lexicographic decision rule does not allow trade-offs among the various attributes. For example, if quality is considered to be more important than cost, no difference in price can compensate for a difference in quality: The lexicographic decision rule chooses the item with the best quality regardless of the cost.

Over the years, the lexicographic decision rule has been compared to various statistical learning methods, including multiple linear regression, support vector machines, decision trees, and random forests. The results show that the lexicographic decision rule can sometimes compete remarkably well with more complex statistical methods and even outperform them, despite its naively simple structure. These results have stimulated a rich scientific literature on why, and under what conditions, lexicographic decision rules yield accurate decisions. Due to the simplicity of its decision process, its fast execution time, and the robustness of its performance in various decision environments, the lexicographic decision rule is considered to be a plausible model of human decision making. In particular, the lexicographic decision rule is put forward as a model of how the human mind implements bounded rationality to make accurate decisions when information is scarce, time is short, and computational capacity is limited.