5.2. LimitationsFour limitations of the study should be considered.First, the study used IT unit informants to assess their business application domain knowledge and their client’s technical knowledge. Since there is a realistic possibility that they might overestimate the former and underestimate the latter, the results should be interpreted with caution.Second, the DCR effectiveness measure assumed that client departments had the greater responsibility for exercising DCRs. (No such assumption wasmade in the DMR exercise effectiveness measures.)This assumption fails to fully account for self control through which IT departments also share some of this responsibility. This assumption does not severely threaten the findings because self control has been found to be used only to a limited degree, and its relationship with performance remains empirically untested. However, caution is warranted in asserting that the client department exercised DCRs because the average value of DCR decentralization is 3.76, which suggests that they are close to the midpoint of the scale but skewed slightly towards IT. It is plausible that this skewing arises from the use of IS directors as the primary informants.Third, the model explained 13.8% of the variance in ISD effectiveness but only 6.4% in ISD efficiency (i.e., <10%) beyond the controls, although these &R2 values were significant. The smaller &R2 for ISD efficiency therefore should be viewed as a limitation of the study.Fourth, projects in the study were relatively small (40 person-months, on average). This might partially account for why both classes of decision rights in the sample leaned slightly towards IT. Although the most visible IT project failures studies have focused on large, mega projects, the majority of the routine development work in organizations encompasses smaller projects.
正在翻譯中..