Premium
Raw conditional probabilities are a flawed index of associative strength: evidence from a multi‐trait paradigm
Author(s) -
Skowronski John J.,
Betz Andrew L.,
Sedikides Constantine,
Crawford Matthew T.
Publication year - 1998
Publication title -
european journal of social psychology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.609
H-Index - 111
eISSN - 1099-0992
pISSN - 0046-2772
DOI - 10.1002/(sici)1099-0992(199805/06)28:3<437::aid-ejsp875>3.0.co;2-l
Subject(s) - associative property , psychology , trait , recall , raw score , index (typography) , context (archaeology) , conditional probability , cognitive psychology , association (psychology) , social psychology , raw data , representation (politics) , statistics , mathematics , computer science , paleontology , politics , world wide web , law , political science , pure mathematics , psychotherapist , biology , programming language
Skowronski and Welbourne (1997) argue that raw conditional probabilities may be a flawed index of associative strength in recall, and may need to be corrected for chance before they can be safely interpreted. Three experiments examined this idea in the context of an experimental paradigm used by Hamilton, Driscoll and Worth (1989). Participants in this paradigm were asked to read items describing a social target. The items each pertained to one of several different trait concepts, or were irrelevant to those concepts. Participants later recalled the items. The data supported Skowronski and Welbourne's conjecture. The raw conditional probabilities differed substantially from the chance‐adjusted probabilities. The data from a second dependent measure, inter‐item generation times, matched the pattern of adjusted conditional probabilities. In addition to their methodological implications, these results contradict the Complete Association Model of person representation proposed by Hamilton et al . Finally, these data raise the possibility that traditional associative models of person memory, which were based on raw conditional probabilities (e.g. Srull & Wyer, 1989), are flawed. © 1998 John Wiley & Sons, Ltd.