This study investigated the construct validity of task inventory ratings fr
om job incumbents and supervisors in 2 jobs on the 4 task dimensions of imp
ortance, criticality of error, difficulty to learn, and time spent. Job inc
umbents completed task ratings on either the typical relative scales or abs
olute scales developed for each dimension; supervisors completed relative r
atings only. Relative ratings from incumbents and supervisors were each com
pared in multitrait-multimethod analyses with the absolute ratings. Ratings
on all task dimensions demonstrated high convergent validity, and time spe
nt and difficulty to learn ratings met criteria for discriminant validity i
n at least 1 job. Practical implications regarding redundancy among ratings
on different task dimensions, the use of both incumbents and supervisors a
s job analysis respondents, and the use of absolute scales in task analysis
are discussed.