Saturday, April 25, 2009

Cognitive Bias in Decision Making


Good article - we are all subject to these biases. Getting to know them and eliminate them is a form of cognitive shadow work (just like emotional shadow work). This article offers further examples of the ways in which we are not nearly as rational as we would like to believe.

Cognitive Bias in Decision Making

Published on September 11, 2008

13 cognitive biases in decision making. These biases are caused by heuristic method, which is commonly used by decision maker to make daily decision.

These biases are caused by heuristic method, which is commonly used by decision maker to make daily decision.

Bias can be defined as distortions in human mind that can lead to misperception and misjudgment. Decision makers usually rely on heuristic methods to aid them in making daily decision. Heuristics are simplified methods that are implemented without complete information and accurate calculation. Because of that, heuristic methods are very vulnerable to biases. In this article I will describe biases that are caused by heuristic methods.

Biases Emanating From the Availability Heuristic

Decision makers usually assess frequency or probabilities an event base on available data in their memory. Decision makers can be biased when they are unaware of unavailable important information. These biases make decision maker overlook many important information and inhibit their objectivity.

Ease of Recall

An event whose instance are more easily recalled will appear more numerous than an event with equal frequency whose instance are less easily recalled. Decision makers are prone to overestimate unlikely events because of their susceptibility to vividness and novelty of information.

Example:

  1. People believe that car accidents cause more deaths in United States than stomach cancer. This believes occur because car accidents get more exposure from media than stomach cancer.
  2. Bombardment of repeated uninformative advertising makes the product more easily recalled. It is often the best way to get us to buy a product.
  3. A manager is usually gives more attention to performance during the three months prior to the evaluation. Performance during this period is easier to be recalled than the previous nine months.

Irretrievability

Decision makers are biased in their assessment because, some information are categorized into groups which are more retrievable from their memory.

Example:

  1. Orion Capital is a small insurance company that insured white water rafting, small company workers’ compensation and bungee jumping business. They managed to very low loss occurrence while, their competitors categorize this kind of operation as extremely risky and shied away.
  2. Consumers associate some location with particular type of product. To maximize traffic, the retailer needs to be in location that consumer associate whit this type of product or store.

Presumed Association

Decision maker tend to overestimates correlation between two events based on the number of similar association.

Example:

  1. A lot of mutilation cases in Indonesia were done by homosexual murderers. Because of that, homosexual murderers are associated with the tendency to mutilate their victims.
  2. Because a lot of marijuana users who are delinquents, marijuana use is associated with delinquency

Biases Emanating From the Representativeness Heuristic

Decision makers usually determine probability of an event base on its similarity to other event. Unfortunately, sometime decision maker are unable to make appropriate association between two occurrences.

Insensitivity to Base Rates

Decision makers tend to ignore quantitative information when they get some qualitative information.

Example: In their experiment Kahneman and Tversky presented participants with descriptions of people who came from a fictitious group of 30 engineers and 70 lawyers. The participants then asked to rate the probability that the person described was an engineer. The probability of a person to be an engineer is 30%. But, the participants assess higher probability if the description of a person is match to their stereotype of an engineer. It means that their assessments were very affected by the person’s description.

Insensitivity to Sample Size

Decision makers tend to ignore whether the information they are getting is representative for the whole population.

Example: Eight of ten women in Jakarta suffer from calcium deficiency. Without information about the number of women involved in this survey, we cannot draw any conclusion from this statement.

Misconceptions of Chance

Sometimes decision maker tend to assume that the probability of current event is determined by previous event. They doesn’t aware whether both events are independent or not.

Example: In behavioral finance consumers usually rely on previous information to predict future outcome. Because of that, they become overly pessimistic about previous loser and optimistic about previous winner. As a result, previous loser tend to be undervalued and previous winner tend to be overvalued. This behavior causes price to deviate from their fundamental value.

Regression to the Mean

Decision makers typically assume that future outcome can totally be predicted from past outcome. In fact, above or below average result don’t necessarily continue forever.

Example:

  1. People sometimes buy a company assuming that good historical performance is sustainable. Intel stock peaked in the summer of 2000, three to four months after the demise of many of its customers.
  2. People use head to head statistic to predict the outcome of a football match. They doesn’t aware that both club already have totally different players, different coach and different strategy.

Conjunction Fallacy

Sometimes decision maker falsely judge that the probability a conjunction is higher than a more global set of occurrence.

Example: The occurrence of a forest fire depends on humidity level, a source of ignition, and a minimum wind speed within a specified area and time. If an expert estimates a probability of 0.1 for each contributing factor, then his estimation about fire probability should be 0.001. However, if the expert were affected by conjunction fallacy, he might judge a fire to be highly probable.

Biases Emanating From the Anchoring and Adjustment

Decision makers usually make assessment by adjusting an initial value to yield final decision. Because of this bias, decision makers are prone to make un-objective and inaccurate decision.

Insufficient Anchor Adjustment

Decision maker typically make insufficient adjustments when establishing final value. This bias usually occurs when a decision maker sees a situation very similar to past event. This similarity make them thinks that similar outcome will be obtained.

Example: In 1995 Dell and Gateway were very similar company. Some investors thought that similar investment outcome will be obtained from both company. In fact, $1 invested in Dell in 1995 would be worth about $19 in 2005. While, $1 invested in Gateway would worth about 33 cents.

Conjunctive and Disjunctive Events Bias

Conjunctive means that several events must occur together to obtain desired outcome. Disjunctive means that only one of many events needs to occur to obtain desired outcome. Decision maker tend to overestimate the probability of conjunctive events and underestimate the probability of disjunctive events.

Example: Suppose a contractor has five divisions. In order to finish a building on time all division should finish their job on time. Suppose each division has 90% probabilities to finish their job on time. Then he may claim that he is 90% confident whether a project will be finished on time. Unfortunately the real probability that all division will finish their job on time is only 59 % or 0.9*0.9*0.9*0.9*0.9.

Overconfidence

Decision makers tend to overestimate their abilities and their knowledge when solving complex problem. They also tend to be overconfident when facing uncertainty. A survey of German stock market forecaster demonstrated that they were overconfident in their prediction. Further, great market experience that measured by correct prediction will increase their overconfidence level.

Two More General Biases:

The Confirmation Trap

People unconsciously search for supporting evidence for their decision and ignoring any refuting evidences.

Example: In investment, once investors purchase a stock they seek evidence that confirms their decision. They are also ignoring information that disconfirms their decision.

Hindsight and the Cures of Knowledge

Sometimes decision makers claim that what was happened was predictable before. For example, in 1991 Martin Bolt and Jon Brink invited Calvin College students to predict the U.S. Senate vote on Supreme Court nominee Clarence Thomas. 58 % of the participant predicted his approval. A week after his confirmation, Martin Bolt asked other students to recall what they would have predicted. “I thought he would be approved,” said 78 % of the participant.


No comments: