Illusory Correlation

An illusory correlation occurs when a person perceives a relationship between two variables that are not in fact correlated. In the first study to demonstrate this phenomenon, participants were presented with pairs of words from two stimulus lists. Each word from the first list was paired an equal number of times with each word from the second list. Later, when participants were asked to estimate the number of times words from each of the two lists had been paired together, they consistently overestimated the number of pairings that had occurred between (a) pairs of words that differed visually from the others (i.e., unusually long words) and (b) pairs of words that shared some semantic association (e.g., lion and tiger). Thus, although all pairs occurred equally often, people gave higher frequency estimates for certain types of word pairs.

The importance of this bias for social psychology concerns its role in stereotyping. Historically, stereotypes were believed to result from defective personality types or were based on overgeneralization of some kernel of truth that existed in the world. Illusory correlation studies provided another basis of stereotyping by suggesting that people might form a stereotype about a group simply as a by-product of the way their minds normally process information about the world.

In a study to test this hypothesis, researchers presented participants with a series of statements about members of two groups, Group A and Group B. The statements described members of the groups performing desirable (e.g., “John, a member of Group A, visited a sick friend in the hospital”) or undesirable (e.g., “John, a member of Group A, always talks about himself and his problems”) behaviors. Each participant read 18 desirable statements and 8 undesirable statements about members of Group A and 9 desirable statements and 4 undesirable statements about members of Group B. The total number of statements about Group A was double that of Group B (i.e., 26 statements vs. 13 statements), but the ratio of desirable to undesirable statements was identical for both groups (i.e., 18 desirable and 8 undesirable vs. 9 desirable and 4 undesirable).

Because membership in Group B and the undesirable statements were the two less frequent occurrences (like the pairs of longer words in the original research), they were more noticeable. Later, participants overestimated the number of times they had read about a member of Group B doing something undesirable. Moreover, participants also rated Group B less favorably than they did Group A. Thus, people perceived a relationship that didn't exist in what they read.

In another experiment, desirable behaviors were used as the novel social occurrence. Participants again read statements about Group A and Group B. However, this time both groups performed more undesirable than desirable behaviors (i.e., Group A: 16 undesirable and 8 desirable; Group B: 8 undesirable and 4 desirable). Again, the ratio of desirable to undesirable statements was the same for both groups. When asked to estimate how many undesirable versus desirable behaviors members of both groups had performed, participants consistently overestimated the frequency of members of the smaller group (i.e., Group B) performing the less frequent behavior (i.e., desirable behavior). Consequently, in this study, Group B was rated more favorably than Group A.

...