| 英文摘要 |
This study examined the construct validity, and improved the test reliability and the estimation accuracy of service growth rates for the correlation between domains of outpatient, emergency and inpatient subscales by using multidimensional Rasch analysis. We took advantage of the Rasch model, using the software Conquest, to analyze 346 hospitals service growth rates data downloaded from BNHI (Bureau of National Health Insurance) website. The multidimensional form of the rating scale model was implemented to examine the fit of the 3 service subscales. For comparison, each subscale individually was also fitted to the unidimensional rating scale model. Standard item fit statistics was used to check model-data fit. After excluding 3 items for subscales of outpatient and inpatient respectively, and deleting 7 items for emergency subscales, the remaining items of each subscale in this study constituted a single construct. The test reliabilities and correlations between domains obtained from the multidimensional approach, (0.905-0.961) and (0.168-0.990), respectively, were much higher than those obtained from the unidimensional approach, (0.84-0.85) and (0.166-0.98), respectively. The results showed that every branches of BNHI were significantly negative growth on overall services besides the Eastern branch. All simultaneously shown growth rates negative on those three subscales were branches of Taipei, Northern, Central, and those three types of hospital, i.e., medical center, regional and local hospitals. |
本站僅提供期刊文獻檢索。
