Illusory correlation and stereotype formation:
making sense of group differences
and cognitive biases
Marïette Berndsen, Russell Spears, Joop van der Pligt
and Craig McGarty
Women are more romantic than men. Scientists are duller than artists. We often make such judgements about groups. Some of these judgements are based on folklore, others are based on observation or experience. When we do rely on observed data how good are we at detecting relationships between group membership and behaviour? Do we find it easy to detect differences between groups? Are our judgements biased? This chapter deals with these issues and focuses on the paradigm that has dominated research on the formation of stereotypic differences between groups over the last three decades: the illusory correlation paradigm. In this paradigm respondents are exposed to a series of behavioural instances each linked to an individual belonging to a specific group. The term illusory correlation refers to perceived associations between attributes and instances other than those contained in the data. In the present case it generally refers to the perception of a stereotypic association of certain features with a given group, typically when the available data is presumed to give little evidence for this (hence 'illusory').
Detecting relationships between events in the environment, between group membership and behaviour, is an essential ingredient of adaptive behaviour. The information derived from these relationships or covariations allow us to make sense of the world by explaining the past, controlling the present and predicting the future (Alloy & Tabachnik, 1984; Crocker, 1981). In these terms, detecting contingency is clearly important for our well-being and even our survival. Although it is well-known that people are able to detect relations between stimuli, they are certainly not perfect in this regard (e.g., Jennings, Amabile & Ross, 1990). For example, it is known that people find it difficult to detect non-contingency (Peterson, 1980) and see relationships where these do not exist. Part of our argument below is that they also see them where researchers think that they do not exist, but we are jumping ahead of ourselves. The central theme in this chapter concerns the perception of socially relevant stimuli,