中文摘要：

目的分析自测健康评定量表（SRHMSV1．0）评价城镇老年人健康状况的信效度。方法2015年11月-2016年7月，采用分层随机抽样法选取广东省广州市、茂名市、汕头市、韶关市、深圳市的老年人（≥60岁）700名为调查对象。其中，广州市和深圳市各200名，茂名市、汕头市和韶关市各100名。采用SRHMSV1．0对其进行调查，分析SRHMSV1．0信效度。结果共发放问卷700份，回收有效问卷599份，有效回收率为85．6％。SRHMSV1．O的Cronbac＇s α系数为0．946，生理、心理、社会自测健康子量表的Cronbach’s α系数分别为0．901、0．889和0．899。SRHMSV1．0的分半信度系数为0．875。除条目4、5、6得分与身体症状与器官功能维度（B1）得分，条目27得分与心理症状与负向情绪维度（M2）得分的相关系数均〈0．600外，其余各条目得分与相应维度得分的相关系数均〉0．600（P〈0．01）。各维度得分与相应子量表得分的相关系数为O．511-0．926（P〈0．01）。初始验证性因子分析（CFA）模型的卡方自由度比值（）（2／af）=3．850，增值拟合指数（IFI）=0．854，Tucker-Lewis指数（ru）：0．840，比较拟合指数（CH）=0．854，近似误差均方根（RMSEA）=0．069，根据各指标的模型良好适配范围可知，初始CFA模型拟合效果未达到理想预期。初始CFA模型的标准化路径系数显示，“条目4←B1”“条目5←B1”“条目7←B1”“条目27←M2”的标准化路径系数分别为0．30、0．28、0．38、0．32，小于预设的标准0．50，考虑删除。将条目4、5、7、27删除后再一次进行拟合，调整后CFA模型的X2／df=3．891，IFI=0．875，TLI=0．861，CF／=0．875，RMSEA=0．070，可以看出模型的增值拟合情况略有改善，但绝对拟合指标和简约拟合指标仍不理想。因此，需要对模型进行修正，根据修正指数（g1）的大小变化依次建立以下联系：“e8-e

英文摘要：

Objective To analyze the reliability and validity of Self - rated Health Measurement Scale Version 1.0 （ SRHMS V1.0） for health assessment of elderly peop!e. Methods From November 2015 to JUly 2016, we selected 700 elderly people aged 60 or over from 5 cities （Guangzhou, Ma0ming, Shantou, Shaoguan and Shenzhen） of Guangdong Province by stratified random sampling with. a sample size of 200, 100, 100, 100 and 200, respectively. We surveyed them with SRHMS V1.0 and analyzed the reliability and validity of the scale. Results A total of 700 questionnaires were distributed and 599 valid questionnaires were collected, the effective recovery rate was 85.6%. The Cronbach＇s a of SRHMS V1.0 and its three subscales （physical health, mental health, and social functioning） were respectively 0. 946, 0. 901, 0. 889 and 0. 899. The spilt - half coefficient of SRHMS V1.0 was 0. 875. The item - domain correlation coefficients were all greater than 0. 600 except that the correlation coefficients of SR4, SRS, SR6 with physical symptom and organ function domain （ B1 ） and the correlation coefficient of SR27 with mental symptom and negative mood domain （M2） were less than O. 600 （ P 〈 O. 01 ） . The domain - subscale correlation coefficients ranged from 0. 511 to 0. 926 （P 〈0. 01 ） . The X^2/df, IFI, TLI, CF1, RMSEA of the initial model of confirmatory factor analysis （CFA） were respectively 3. 850, 0. 854, 0. 840, 0. 854, 0. 069, which demonstrated that the fitting effect did not achieve the desired effect according to good adaptation scope of each index. The standardized path coefficients of ＂ SR4←B1 SRS←B1 SR7←B1 SR27←M2＂ were respectively 0. 30, 0. 28, 0. 38, 0. 32 in initial CFA model. And these four items were deleted because their standardized path coefficients were all less than the preset criteria of 0.50. We adjusted the CFA model by deleting SR4, SR5, SR7 and SR27, and the fitting results were x2/df = 3. 891, IFI =0. 875, TLI=O. 861, CFI =0. 875, RMSEA =0. 070, which indicate