亚健康自评量表的考评与评判模型的初建

Scientific evaluation and preliminary establishment of the judge model of sub-health self-rating scale

目的初步建立亚健康评判模型并考评其科学性。方法整群抽样获得有效个体6 205例,分析量表的可行性、信度及效度。量表各分值通过原始粗分、转化分和T分数计算,亚健康自评量表采用百分位法,以现患率为分界,建立亚健康临床参考评判分值表;随机抽出1 000份问卷分别进行专家定性评判和标准化评判,将这两种方法的评判结果进行Kappa吻合度考核;采用logistic回归方法建立"3+1"评判的模型,采用拟合指数2、总判对率及Nagelkerke R2考核该模型的预测性能。结果 (1)量表可行性好,接受率89.03%,应答率99.67%,10min之内能完成量表。信度分析提示,内部一致性信度测得总量表及躯体、心理、社会领域的Cronbach'sα系数分别为0.942、0.915、0.856和0.850,各因子的Cronbach'sα系数为0.72～0.88;分半信度测得总表条目及躯体、心理、社会领域的分半信度分别为0.938、0.933、0.890和0.881,各因子的分半信度为0.73～0.90。效度分析:一阶探索性因子分析抽取10个因子(方差累积贡献率为56.63%);二阶探索性因子分析(Promax斜交旋转法)获得2个公因子(方差累积贡献率为56.55%),FⅠ包含6个躯体因子,FⅡ包含心理和社会等4个因子,提示心理和社会因子间的相关程度较大。验证性因子分析结果显示GFI=0.858,CFI=0.963,AGFI=0.846,RMSEA=0.055 1,NFI=0.961,IFI=0.963,TLI=0.961。内容效度:各条目与其所属因子间呈正相关(相关系数r为0.51～0.88);各因子与其所属领域之间呈相关性(相关系数r为0.52～0.89),与其他领域未见明显相关;各领域得分与量表总分之间呈相关(相关系数r为0.77～0.93)。(2)大学生亚健康临床参考评判各领域分值、总分评判与专家定性评判的Kappa值均达0.7以上。"3+1"评判模型预测模型的拟合度较好。结论本量表有较好的可行性、信度与效度,有一定的实用价值。本研究初步建立了"3+1"评判的模型,该模型有较好的预测率与解释能力,可作为亚健康的评判标准。

Objective To establish and evaluate the judge model of Sub-health Self-rating Scale（SSS）. Methods Through cluster sampling,6205 effective samples were obtained.Feasibility,reliability and validity analysis were conducted.The clinical sub-health reference parameters,including original score,transformed score and T score,were worked out by percentile method,according to the prevalence rate of sub-health.One thousand valid questionnaires by random sampling were collected to evaluate the consistency between expert judge model and score judge model.The triadic judge model of sub-health by logistic regression was conducted for analysis with fit index χ2 and Nagelkerke R2. Results（1） The feasibility was sufficient with acceptance rate and finish rate of 89.03% and 99.67%,respectively.According to internal consistency analysis,the Cronbach′s alpha indexes of the total scale and the dimensions of body,mood,and society were 0.942,0.915,0.856,and 0.850,respectively,ranging between 0.88-0.72.Split-half reliability of the total scale and the dimensions of body,mood,and society were 0.938,0.933,0.890,and 0.881,respectively,ranging from 0.73 to 0.90.According to validity analysis,there were 10 factors extracted by exploratory factor analysis as 1-order factor（Varimax,FI） with cumulative proportion of 56.63%;and 2 factors as of 2-order factor（Promax,FII） with cumulative proportion of 56.55%.FI included 6 factors of physiological dimension,while FII included 4 factors of psychological and social relationship dimensions,which suggested obviously negative relationship between psychological and social factors.According to confirmatory factor analysis,GFI,CFI,AGFI,NFI,IFI and TLI were 0.86,0.96,0.85,0.055,0.961,0.963 and 0.961,respectively.In content validity,there was significant correlation between each item and factor（r ranged 0.51-0.88）.There was also significant correlation between each factor and its own domain（r ranged 0.52-0.89）.There was significant correlation between subscale and total scale（r ranged 0.77-0.93）.（2） There was significant consistency between expert judge model and score judge model（K0.70）. Conclusion The scale is of better feasibility,reliability and validity and can be used in practice.This study initially established the judge model,which had good prediction rate and interpreting ability.To some extent,the model can be a judging standard of sub-health.