In two previously reported experiments, Loomes, Starmer and Sugden hav
e found that choices are systematically non-transitive, following a pa
ttern of 'cycling asymmetry' predicted by regret theory. However, ther
e are other potential explanations for these observations. This paper
reports four experiments designed to discriminate between alternative
explanations. There are three main findings. First, when the original
experiments are replicated, the same cycling asymmetry is found. Secon
d, this pattern is not the result of event-splitting effects. Third, b
oth the frequency and the asymmetry of cycles are greater when choice
problems are presented in act/event matrices than when the options are
described separately.