System and user attributes associated with successful searching

A study was performed to assess system and user attributes associated with successful searching. A group of highly experienced retrieval system users were given eight queries-half to be completed with a Boolean system and the other half to be completed with a natural language system-and asked to identify "instances" of information related to a topic. A variety of demographic, experience, cognitive, personality, search mechanics, and user satisfaction attributes were measured and evaluated for association with "instance recall." The system oriented analysis showed user preference for the Boolean system but equivalent instance recall with the two systems. The user oriented analysis showed improved performance with occupation type, user satisfaction, and number of search terms per cycle. A variety of cognitive and personality attributes did not show any association with better searching.