RESULTS
The descriptive statistics (means and standard deviations) for all study variables are shown in Table 1. Path analysis based on least squares regression analysis (Blalock 1964; Asher 1976) was used to test the four hypotheses in order to examine the direct and indirect effects of the variables in the model. The results for QJ implementation are shown in Table 2. As indicated, both hospital culture and implementation approach are significantly and positively associated with a greater degree of QI implementation. Specifically, there is a significant association between hospitals with group/developmental cultures emphasizing teamwork, support, development of everyone's potential, and a willingness to undertake some degree of risk and the degree of reported QJ implementation. Similarly, hospitals using a more prospector-like approach to implementation emphasizing decentralized control, empowerment, and "just-in-time" training of physicians report a greater degree of implementation. Thus, both hypothesis 2 pertaining to culture and hypothesis 3 pertaining to implementation approach are supported by the data. Neither bed size nor whether or not the hospital met all the criteria for being a CQl/TQM site are significantly associated with QI implementation. The overall fit of the model is quite good with 54 percent of the variation in QI implementation being explained by hospital culture and implementation approach. Additional analysis examining the effect of region (west versus all other), location (urban versus rural), and local market competition (number of HMOs, number of competing hospitals, percent of revenue from capitation) revealed these variables to be non-significant and did not change the culture/implementation results.