Tuesday, February 2, 2010

Judgment Under Uncertainty

In their article “Judgment under Uncertainty: Heuristics and Biases”, Tversky and Kahneman discuss how subjective judgments are made to determine the likelihood of uncertain events prior to decision making. They discuss how decision makers try to assess the probability of an event, but how it is often difficult to do so accurately. Often, decision makers can estimate the general magnitude of an event (i.e. very likely or somewhat likely), but it is more difficult to make a more exact determination.
One of the problems that the authors address is the representative heuristic. In this situation, people might make judgments based on stereotypes without consideration to proportions. They use an example by suggesting that people might guess the likelihood of someone’s occupation by their personality traits without considering the actual number of people employed in that occupation. Assessing probabilities often does not include data on past probabilities of similar events. Experiments have shown that even when participants are told proportions ahead of time, they often ignore these proportions when guessing probabilities of events and base probabilities more on subjective and irrelevant data.
Another problem is that people often have erroneous expectations about probabilities. For example, if a coin is tossed six times, people have a tendency to believe that the sequence H-T-H-T-T-H is more common than the sequence H-H-H-H-T-H. This is because people have a tendency to think that probabilities of overall events also apply to probabilities of local sequences as well. This type of misconception has been shown by gamblers who will make decisions based on events that have taken place up to that point, rather than the probabilities of events occurring or recurring. Educated professionals have also been known to make similar mistakes in judgment.
Another problem is that people often try to make predictions when predictability is very low. For example, people often make decisions about the future profitability of companies based of favorable or unfavorable information. Unfortunately, this information often has little relevance to profitability. Therefore if there is no information that is directly linked to the profitability of the company, the predicted profitability should be considered to be entirely random. Similarly, decision makers often assume information to be more valid than it actually is. More astute individuals also make errors in judgment based on statistical misconceptions, such as normal distribution.
People also often assess likelihoods based on the availability of information. One way they do this is from their own experiences. Another problem is that people often use a means of finding information which is more convenient, even though it may be much less effective. Finally, they have a tendency to imagine or misconceive facts when there are not many that are readily available.
Decision makers often try to estimate an answer by starting at an initial value and then making some adjustments. Unfortunately, this process does not seem to work well because the adjustments are not appropriate. This is often because these adjustments are based on information that has been extrapolated, because the subjects believe that the same patterns will be observed throughout. Another problem witnessed is that subjects had a tendency to select impractical events when it came to disjunctive or conjunctive events as opposed to simple events. One finding claimed that when subjects had to choose between a simple event with a probability of 50% and a disjunctive event with a probability of 48% they would tend to select the less likely disjunctive event. Interestingly, when the subject had to choose between a simple event of 50% and a conjunctive event with a probability of 52% they would tend to choose the simple event. This study was consistent with a general finding that subjects tend to overestimate the likelihood of a conjunctive event and underestimate the likelihood of a disjunctive event.
Throughout the article, the authors discussed different biases that people have in making decisions. They also discussed how these biases were common among educated people as well as to laymen, although the mistakes made by professionals were not usually as elementary as those made by laymen. Educated decision makers tend to try to make decisions based on statistical data that is available to them. However, they often make improper assumptions and inferences from the data that is presented to them, which leads to many of the same misconceptions and errors in judgment.

No comments:

Post a Comment