首都医学发展科研基金函评立项指标体系的分析与评价

目的根据首都医学发展科研基金的实际函评结果，对函评立项指标体系进行全面的分析与评价，为进一步完善指标体系提供依据的方式和方法。方法应用克朗巴赫系数、分半信度法、结构方程模型的验证性因子分析、专家评定法的内容效度指数、高低分组法鉴别指数，结合实际函评结果，对指标体系的信度、效度、区分度、合理可行性等分析评价。结果①信度：一级指标及指标体系总体的克朗巴赫系数、分半信度均在0．8以上。②校度：指标体系的条目层次分析，标化因子载荷范围在0．57～0．86均大于0．5，路径分析具有统计学意义（t值大于1．96），标准误范围在0．01～0．02，二级指标的路径性能表现良好；模型整体拟合效果的主要拟合指数均达到标准（RMSEA=0．065、NNFI=0．99、IFI=0．99、CFI=0．99、GFI=0．95、AGFI=0．93、x^2=1028．21），指标体系结构效度良好。③内容效度：指标体系内容效度指数为0．97（均值S-CVI），一致性好；每项二级指标条目内容与研究目的关联认可的一致性，除“学术思想”的内容效度指数0．57（I-CVI）较低外，其他二级指标的I—CVI均为1．00。④区分度：各二级指标，一级指标、总分的鉴别指数范围在0．51～0．76，均大于0．40，区分度良好。⑤合理可行性：对2007年立项评审项目的指标体系得分分析，中位数为57．67，均值与中位数相近，分布趋近正态分布，符合函审的要求，合理可行。结论首发基金函评立项指标体系的信度、效度、区分度良好，能较好的反应实际情况，合理可行，所应用的方法对分析、评价及完善科研基金函评立项指标体系，具有实际的应用和借鉴价值。

Evaluation of the written project approval indexes system of Capital Medical Development Research Foundation

Objective The purpose of this study was to evaluate the written project approval Indexes System of Capital Medical Ddevelopment Research Foundation. We analyzed the approval results of application projects and proposed measures to improve the index system. Methods We evaluated its reliability with Cronbachls alpha coefficient, split-half reliability; validity was evaluated with contract validity and content validity, discrimination was studies according to discretion grouping law, and feasibility was studied with the percentage-of-completion method. Results ① Reliability： Cronbach＇s alpha coefficient of factor and total scoresland split-half reliability were all over 0.80. ② Contract validity： the standard factor loading coefficients of the index system were over 0.70 （range： 0. 57-0. 86）, indicating path performed well. All factor loading coefficients were statistical significant （T-values were 30.49-54.25, all were over 1.96）. Fit Indexes （RMSEA = 0. 065.NNFI = 0.99. IFI ： 0.99.CFI = 0.99.GFI ： 0.95.AGFI = 0.93.？（2 = 1028.21.x^2/DF = 11.82） indicated the systemts goodness of fit was good. ③ Content validity： Scale-level CVI, S-CVI was 0.97 and Inter- rater agreement was 0.93, over 0.90, which indicated that indexes system＇s whole Content validity performed well. Except for the item Academic thought item-level CVI being 0.57 all items＇ item-level CVI were 1.00. ④Discrimination： Discrimination index of every primary index, secondary index and total score ranged from 0. 51 to 0.76 （all were over 0.40）, meaning that discrimination performed extremely well. Conclusion The indexes system＇s reliability, contract validity, content validity and discrimination conformed to the actual application projects appraisal results.