中文版《婴幼儿社会认知发展筛查量表》的信度和效度评价

目的对中文版《婴幼儿社会认知发展筛查量表》进行信度和效度评价。方法在广州市和佛山市选取无精神发育疾病、无重大生理疾病的正常婴幼儿843名,选取中山大学附属第三医院儿童发育行为中心确诊为孤独症的幼儿170例,对中文版《婴幼儿社会认知发展筛查量表》进行修订,并且评价量表的信度和效度。结果①信度分析：全量表及4个子量表的Cronbach′sα系数均〉0.7;全量表及4个子量表的分半信度系数为0.69-0.94;全量表及4个子量表的重测信度系数为0.79-0.95。②效度分析：各条目与全量表的相关系数为0.31-0.86;除认人子量表与所属条目n10的相关系数为0.40,各子量表与其他所属条目相关系数均〉0.5,但与其他子量表所属条目相关性不强;各子量表间、子量表与全量表间的相关系数均〉0.5。因子分析共提取6个公因子,解释总变异的60.63%,因子分析结果与原量表结构基本一致。2-3.5岁正常幼儿全量表和4个子量表得分都显著高于同年龄段孤独症幼儿。结论中文版《婴幼儿社会认知发展筛查量表》具有较好的信度和效度,且操作简便,可用于临床上评价0.5-3.5岁婴幼儿社会认知能力的发展情况。

Evaluation on reliability and validity of Chinese version-infant and early children social cognition development screening scale

Objective To test the social cognitive of normal children and autism spectrum disorder children,and to explore the reliability and validity of Chinese version infant and early children social cognition development screening scale.Methods After translating the screening scale and the instruction manual,843 normal children from children health care clinic and kindergarten without severe illness and 170 children who had diagnosed as ＂autism＂ were chosen.Their parents were willing to have face-to-face investigation.The simultaneously appearance of children was observed and the screening scale was filled.Lastly,SPSS 13.0 was used to analyse the reliability and validity of the Chinese version screening scale.Results ①Reliability analysis： the Chronbach＇s α coefficients： the coefficient of the whole checklist was 0.96,and those of partial checklists were all above 0.7;The split-half reliability： the coefficients of the whole scale was 0.94,and each scale was ranged from 0.69 to 0.94;The test-retest reliability： the coefficients of whole scale was 0.95,and that of 35 items exceed 0.6,and that of each scale was ranged from 0.79 to 0.95.②Validity analysis： content validity： the correlation coefficients between most items and whole checklist were ranged from 0.31 to 0.86,and the correlation coefficients between the partial checklists and whole checklist were above 0.5（except item e10）,but the relation was poor with each item in other partial checklists;Construct validity： 6 public factors were extracted by principal component analysis,which defined the 60.63% of the full variant.Empirical validity： the difference of different sex children in whole scale and four part scales had no statistically significance.Discriminant validity： The scores of children with autism in whole checklist and each partial checklist were lower than the normal children（P〈0.001）.Conclusions The test could be operated conveniencely,and the reliability and validity of this checklist is so good that it is useful f