Grosskopf and Valdmanis (1987). This possibility is illustrated in Fig. 3. Suppose that Hospitals C, D, and E are all small hospitals and ysys represents its own separate frontier. Hospitals A and B are all large hospitals and the pooled frontier is represented by ypyp. Efficiency of Hospital C equals to OCV/OC and OCW/OC when measured relative to the separate and pooled frontiers respectively. The difference between separate and pooled frontiers is the distance between the two frontiers OC/OCV. This relative distance thus equals the ratio of the efficiency of C relative to the pooled frontier to the separate frontier (OC/OC)/(OCV/OC). The ratio captures the difference in the frontiers of large and small hospitals and it approaches unity when the difference between the two frontiers diminishes. The third step is to assess IT impacts on hospital efficiency using regression analysis in conjunction with DEA technique to explain the effect of IT investment as an influencing factor on the variation of the DEA efficiency scores. This approach, called a two-stage procedure, involves measuring the DEA scores in the first stage and regressing such scores against the influencing factor in the second stage. The influencing factor in the regression analysis is the factor that can affect the way hospitals transform traditional inputs to outputs but is neither completely controlled by hospital management nor a critical input needed for hospital production like physicians and nurses (Fried, Schmidt, & Yaisawang, 1995). In addition to IT investment, other factors can be reimbursement policy and ownership, for example.
Fig. 3. Separate and pooled frontiers.
B. Watcharasriroj, J.C.S. Tang / Journal of High Technology Management Research 15 (2004) 1–168
Assessing IT effects on performance of hospital production process that directly involves IT in its activities is considered to be more appropriate than measuring IT effects on performance measures at the enterprise level such as profitability and return on asset. This is because the link between IT investment and organizational performance measures tends to be diluted due to the effects of many other unidentified mediating and moderating variables. Using regression analysis to evaluate IT effects also has the advantage of not requiring the direction of IT impacts on the efficiency scores to be specified in advance. The sign of the coefficient will indicate the direction of the influence, and standard hypothesis testing can be used to assess the strength of the relationship. As the DEA efficiency scores have an upper bound of 1 and a lower bound of 0, the dependent variable of the