On the descriptive value of loss aversion in decisions under risk

Five studies are presented that explore the assertion that losses loom larger than gains. The first two studies reveal equal sensitivity to gains and losses. For example, half of the participants preferred the gamble "1000 with probability 0.5; -1000 otherwise" over "0 with certainty." Studies 3, 4, and 5 address the apparent discrepancy between these results and the evidence for loss aversion documented in previous research. The results reveal that only under very specific conditions does the pattern predicted by the loss aversion assertion emerge. This pattern does not emerge in short experiments or in the first 10 trials of long experiments. Nor does it emerge in long experiments with two-outcome symmetric gambles, or in long experiments with asymmetric multi-outcome gambles. The observed behavior, in these settings, reflects risk neutrality in choice among low-magnitude mixed gambles.