To determine the impact of causal relationships and shared features in the use of analogy, subjects in two separate studies were informed of the behavior of six members of a little known primitive tribe and asked to make predictions on the behavior of additional tribe members based on their perceived similarities with the original group. Results indicated that people are more likely to use analogies when elements in the base (familiar item) and the target (unfamiliar item) appear to reflect a similar causal relationship. To discover whether or not people use analogy in the absence of any known causal relationship, subjects in a third study were asked to make predictions on behavior that was identified with only a nonsense word. Being unable to use any personal background knowledge in their predictions, subjects were forced to rely on global similarities between target and base items. Results demonstrated that in the absence of a known causal relationship, people will use a single shared characteristic between base and target as the basis for an analogy. (MM)