Is the transitivity of choices a proper measure of rationality?

According to normative rationality, transitivity of choices is a benchmark for rational behavior. We argue that context should always be taken into account when discussing benchmarks. Using a series of simple mathematical tasks, we assessed participants’ responses against two criteria of rationality: transitivity and accuracy. In Study 1, the relation between the number of transitivity violations and the number of errors was best described by a reversed U-shaped function. This means that, as long as the participants use an adequate strategy and computational errors are merely accidental, transitivity violation positively relates to the number of errors. Otherwise, if the participants use an inadequate strategy that leads to systematic errors, transitivity violation negatively relates to the number of errors. We replicated these findings in Study 2. People can make decisions that comply with the transitivity criterion but are inaccurate; being erroneous, though transitive, does not make one’s behavior rational.