User evaluations of information systems are frequently used as measures of
MIS success, since it is extremely difficult to get objective measures of s
ystem performance. However, user evaluations have been appropriately critic
ized as lacking a clearly articulated theoretical basis for linking them to
systems effectiveness, and almost no research has been found that explicit
ly tests the link between user evaluations of systems and objectively measu
red performance. In this paper, we focus on user evaluations of task-techno
logy fit for mandatory use systems and develop theoretical arguments for th
e link to individual performance. This is then empirically tested in a cont
rolled experiment with objective performance measures and carefully validat
ed user evaluations. Statistically significant support for the link is foun
d for one measure of performance but not for a second. These findings are c
onsistent with others which found that users are not necessarily accurate r
eporters of key constructs related to use of IS, specifically that self rep
orting is a poor measure of actual utilization. The possibility that user e
valuations have a stronger link to performance when users receive feedback
on their performance is proposed. Implications are discussed. (C) 2000 Else
vier Science B.V. All rights reserved.