The effect of the time unit on software development effort estimates

Abstract

Estimates of software development effort are frequently inaccurate and over-optimistic. In this paper we describe how changes in the granularity of the unit of estimation, e.g., work-days instead of work-hours, affects the effort estimates. We describe four psychological mechanisms, how they interact and discuss the expected total effect of higher granularity units on effort estimates. We argue that the mechanisms in general imply that higher granularity effort units will result in higher effort estimates, e.g., that estimating software development work in work-days or weeks will lead to higher estimates than when estimating in work-hours. A possible implication of this predicted effect is that, in contexts where there is a tendency towards under-estimation, estimation in work-days or weeks instead of work-hours leads to more realistic estimates.