Psychology 458/558
Judgment and Decision Making
Prof. Bertram Malle
Fall 1995

Lecture 11: Nov 7
Heuristics and biases

1. Anchoring and (insufficient) adjustment

A. Given any kind of starting value, people anchor their judgment on this value. They are not able to avoid the influence of starting values even if they know that the values are random, uninformative, etc. Examples: age estimation, African nations, real-estate study, response scale effects, expectations.

B. Insufficient computation/correction. Sometimes people do not adjust the starting value sufficiently because they run out of time (e.g., 1x2x3...vs. 8x7x6...). Compare this notion to the Spinozan belief model (Gilbert, 1991).

Parallel phenomena exist in the social domain: primacy effects (first impression never get fully corrected!), attitude attribution, belief perseveration, negotiations (a high demand first), persuasion (a daring request first).

It is a very time- and energy-saving heuristic to start at a particular value and adjust itif necessary. The problem is that these values are not always diagnostic, and people do not adjust them sufficiently. One compounding effect for difficult judgments is that people have a hard time saying "I don't know." They therefore use whatever information is available (even if it might be useless) to try to construct an answer.

Tips to avoid the disadvantages of this heuristic:

2. Availability heuristic

For many judgments of frequency or probability, people make use of what is spontaneously available in their mind, i.e. what seems easy to retrieve (or imagine). (For example, how many people in class have long hair? Most likely, you will try to count the people with long hair that your mind can picture.) The ease of retrieval is surely influenced by previous encounters with the real objects or events, but it is also influenced by vividness and priming effects; and even previous encounters only partially reflect real frequencies but are often based on biased sampling:

Example for when it works well: Predicting (in 7 sec) how many flowers you will be able to name in 2 min. In this case, the ease of retrieval during the 7 sec search is very diagnostic of the actual amount of items stored in memory. (In class, the correlation between estimate and actual retrieval was r = .85.)

Examples for when it doesn't work as well: Bus stop patterns (some patterns are easier to imagine but not necessarily more frequent); partners' or room mates' estimations of how often they themselves do the dishes, vacuum the carpet, or take out the trash (you remember your own cases much better than your partner's); estimation of death causes (some are easier to imagine, probably from numerous news reports, but they are not necessarily more frequent).

Tips to avoid the disadvantages of this heuristic:

3. Representativeness heuristic

People often judge how likely a given object (or person or event) belongs to a certain category by judging how representative the object is of their stereotype, schema, or script of that category. Whether an object fits your stereotype of a category is, however, not always a good indicator of how likely it is that the object belongs to that category.

Example: Linda the bank teller; Bill the accountant; invasion in Germany; engineers and lawyers; shy Steve; p(suicide) vs. p(suicide after depression)? In all these cases, how well the event or person fits our stereotype contradicts its actual probability. Note that the most unspecific descriptions (e.g., being human, being female) are the most "probable" because they hold for many people; more specific descriptions might fit a stereotypes well, but they are less probable because they hold only for a few people.

Tips to avoid the disadvantages of this heuristic: