To provide a theoretical base for investigating the influence of inferential error on group decision making, current literature on both inferential error and decision making is reviewed and applied to the Watergate incident. Although groups tend to make fewer inferential errors because members' inferences are generally not biased in the same direction, group interaction may promote collective inferential error by encouraging (1) the mass acceptance of atypical cases as representative, (2) the specialization of knowledge in complicated issues, and (3) the development of a shared outlook or schema. A recent model suggests possible sources for these group characteristics. As group participants have less time to individually consider a given concept, and consequently less time to activate a wide range of significant associations, their formulation of the target concept will tend to be simplistic and selective. A taped conversation among President Richard Nixon, John Dean, and John Mitchell shows how group inference errors develop. After Nixon denigrated its significance, Mitchell and Dean quickly dismissed the grand jury investigation of the cover up as unimportant, accepting Nixon's simplification of a complex and very serious issue as a "can of worms." (MM)