Thoughts and Comments

PREFERENCES AND PERCEPTIONS

One common theme in my research is the idea that when people make choices these are influenced by two elements: the perception of the choice options available to them and the preferences over these options, as they are perceived. Classical economic choice theory sees both of these as very well-defined and clear to the person who is making the decisions, and the predictions of what the choice will be are therefore very precise. However, the choice is never that clear.

Stochastic Choice and Rationality

Even if both the perceptions and the preferences are crystal clear to the person, choices still appear to have an element of randomness and we, as researchers, cannot predict the choice precisely. For example, if one choice option is to get $2 and the other is to get $3, and nothing else differs between the options, a stochastic choice process can still lead to a choice of the $2 option without implying irrational behavior. While the likelihood of selecting the obviously inferior option is bound to be low in such clear-cut cases, it does not require much complexity before the likelihood of an "inferior" choice becomes non-trivial. Thus, it is incorrect to treat choices of options that are clearly suboptimal as irrational since they are only irrational if we assume a deterministic, rather than stochastic, choice process.

Precision and Bias in Perceptions

Perceptions of choice objects can be both biased and stochastic. In the case of color preferences over specific objects, such as a bouncy balls offered from a machine that dispenses balls randomly, the perception of what color bouncy ball will be dispensed can be modeled using probabilities. If there are only a few balls in the machine, say 10 very visible balls, the perception can be fairly accurate so that the proportion of balls of different colors determines the probability. Thus, with 3 yellow and 7 blue balls the likelihoods would be 0.3 and 0.7, for the yellow and blue respectively. However, with a large number of balls such precision is not possible and likelihood may instead be best modeled with a probability distribution. A biased perception could lead to over- or under-estimation of the likelihood of a specific color. In the case of the likelihood 0.3 of a yellow ball, for instance, a positive bias would be perceiving the likelihood to be 0.4 and a negative bias would be perceiving the likelihood to be 0.2. A precise, and unbiased, probability distribution would assign zero mass to any probability outside 0.3, and similarly a precise, but positively biased probability distribution would assign zero mass to any probability outside 0.4. Imprecision, in the unbiased case, is captured by a probability distribution of the likelihood of a yellow ball that has positive mass outside 0.3. When a person has a precise and unbiased probability distribution over the likelihood of yellow, we can say that this person is certain and correct. If instead the person has an imprecise and unbiased distribution we can say that she is uncertain but correct. On the other hand with a precise but biased distribution, we can say that she is certain but wrong.

Perceptions Confounding Inferences about Preferences

Recognizing that both preferences and perceptions influence choices is essential when attempting to measure people's risk attitudes. In Experimental Economics the most common tasks offer respondents pairwise choices between risky options that differ in rewards and likelihoods of rewards. Both rewards and likelihoods are presented in very salient ways, which makes it likely that most respondents have fairly clear, and similar, perceptions of what they can get from the lotteries. Thus, variations in choices can then be interpreted as reflecting variations in risk preferences. In other literatures other tasks have been used to ascertain risk preferences of respondents, a popular one being BART (Balloon Analogue Risk Task, Lejuez et al. (2002)). In this task participants are presented with a simulation of a balloon that they can inflate. If they stop inflating the balloon before it pops they get an amount of money, and this amount is increasing the more they inflate the balloon, but if it pops they get nothing. The stage at which the balloon pops is not known to the participant. In this task participants know how much money is on the table, but they do not know the probabilities. Instead they form subjective perceptions about these probabilities, that confound the ability to infer risk preferences from their choices. When observing the variations in when participants stop inflating the balloon, the researcher cannot identify to what extent this choice reflects risk preferences and to what extent it reflects risk perceptions, i.e. subjective probabilities.

Lejuez CW, Read JP, Kahler CW, Richards JB, Ramsey SE, Stuart GL, Strong DR, Brown RA (2002). "Evaluation of a behavioral measure of risk taking: the Balloon Analogue Risk Task (BART)." Journal of Experimental Psychology: Applied, 8, 75-84.