BIS-11信效度的荟萃分析和再验证研究

发布时间：2018-07-12 14:37

本文选题：巴瑞特冲动性量表 + 荟萃分析　；参考：《皖南医学院》2017年硕士论文

【摘要】：目的:1.系统评价巴瑞特冲动性量表(BIS-11)信效度,探讨不同地域文化背景下信效度的差异,寻找影响BIS-11信效度的预测来源。2.在不同中国青少年样本中检验BIS-11中文版的信效度,检验量表跨文化的适应性及跨人群的差异性。方法:1.计算机检索Pubmed、Science Direct、wiley-Blackwell、万方、中国知网、维普中文科技期刊全文数据库及百度学术检索国内外BIS-11信效度相关文献,检索时限为1995-2016年1月1日,按纳入与排除标准筛选文献并提取资料。对Cronbach’s a采用Rodriguez、Maeda使用的a系数效应量转换方法,重测信度利用Fisher’s Z转换,然后利用R语言的metacor package对信度系数效应量进行meta分析,分析不同地域的信度差异;利用SPSS 18.0对信度系数进行加权回归分析,以找出信度系数的预测来源。采用资料分析总结概括BIS-11的因子结构分析。2.在正常青少年、心理障碍青少年、违法青少年及网络成瘾青少年这四组青少年样本中进行BIS-11中文版的信效度再验证。采用SPSS 18.0在不同青少年样本及总样本中分析BIS-11量表总分及分量表的内部一致性信度、分半信度,计算正常、异常青少年的重测信度,不同样本量表总分及各分量表得分差异,在不同青少年样本及总样本中验证一阶三因子模型、一阶六因子模型、二阶三因子模型、二阶二因子模型及两个双因子模型的效度。结果:1.荟萃分析共纳入35个研究(包含45个样本),Cronbach’s a、pearson’s r、ICC三类信度系数效应量的漏斗图结果支持Cronbach’s a继续meta分析。Meta分析结果显示所有样本的Cronbach’s a系数为0.81,95%CI(0.79,0.83);欧美地区为0.79,95%CI(0.77,0.82);亚洲地区为0.84,95%CI(0.81,0.86);其他地区为0.77,95%CI(0.74,0.80)。异质性检验:所有研究值为74.8%(P0.001);亚组分析,欧美、亚洲、其他地区的I2值依次为71.2%(P0.001),68.2%(P0.001),41.5%(P=0.14)。Meta回归分析显示样本量、年龄、人群、测验版本、地域与a系数效应量多元回归相关系数R=0.56,5个变量共解释a系数效应量30.9%,测验版本和人群两变量回归系数达到显著[-0.34(P0.05)、-0.61(P0.001)]。即人群、修订版本对a系数存在显著的正向影响。因子结构分析的结果显示BIS-11所测的冲动性维度不适应单维模型,多数研究支持Barratt原始三因素模型,验证性因素分析结果显示对嵌套Patton六因子模型的高阶因子模型也有较好的模型拟合结果。2.在总样本及四组青少年样本中,三个分量表及总量表有着较好的内部一致性信度,除正常青少年及总样本的认知分量表内部一致性信度低于0.80(分别为0.67、0.79),其他均在0.80以上。总样本BIS-11的分半信度为0.88,重测信度正常青少年、网络成瘾青少年分别为0.79、0.72。四组青少年样本的总量表及各分量表得分差异有统计学意义(P0.001)。结构效度分析显示Barratt的三因素模型[总样本:RMSEA=0.047,95%CI(0.044,0.050),AGFI=0.91]、Patton的六因素模型拟合较佳[总样本:RMSEA=0.055,95%CI(0.052,0.058),AGFI=0.91],双因子模型也有不错的拟合效果[总样本G-3模型:RMSEA=0.041,95%CI(0.039,0.044),CFI=0.91;总样本G-6模型:RMSEA=0.055,95%CI(0.052,0.058);CFI=0.85],高阶因子模型拟合不佳。结论:1.BIS-11在不同地域文化使用均有良好的信度,该量表具有较好的跨文化适应性。人群、测验版本对a系数起正向预测作用,临床人群内施测得出内部一致性高于非临床人群,BIS-11修订版本较其他版本信度要高。BIS-11中文修订版在正常、异常青少年中再验证分析时显示了较好的心理测量学特性。2.三因素结构显示较好的稳定性,在中国青少年样本中三因子结构模型也有较好的效度,在解释BIS-11分数时可首选三因素理论模型,双因子模型可作为多维结构的新视角对冲动性维度进行解释。
[Abstract]:Objective: 1. to evaluate the reliability and validity of Barrett Impulsivity Scale (BIS-11), to explore the difference in the reliability and validity of different regional cultural backgrounds, and to search for the reliability and validity of the Chinese version of the prediction source.2. in different Chinese adolescent samples, the adaptability of the test scale and the difference of cross population in different Chinese adolescents. Methods: 1. methods: The computer retrieves Pubmed, Science Direct, wiley-Blackwell, Wanfang, China knowledge network, VP Chinese sci-tech journal full text database and Baidu academic retrieval of BIS-11 reliability and validity related literature at home and abroad. The retrieval time limit is 1995-2016 years in January 1st, according to the inclusion and exclusion criteria, select the literature and extract the data. Cronbach 's a is used Rodriguez, Maeda use. The method of a coefficient effect conversion, retest reliability using Fisher 's Z conversion, and then using METACOR package in R language to analyze the reliability coefficient effect by meta analysis, analyze the difference of reliability in different regions, and use SPSS 18 to carry out weighted regression analysis on reliability coefficient to find out the source of reliability coefficient prediction. The factor structure analysis of BIS-11.2. was used to reverify the reliability and validity of the Chinese version of BIS-11 in four groups of adolescents, mental disorders, juvenile delinquents and Internet addicts. SPSS 18 was used to analyze the internal consistency letter of the total scores and subscales of the BIS-11 scale in different adolescents and total samples. Degree, split half reliability, normal calculation, abnormal teenager retest reliability, difference of total score of different sample scale and the score of each subscale, test the first order three factor model, first order six factor model, two order three factor model, Nikai Ni factor model and two double factor model. The result: 1. meta score. A total of 35 studies (including 45 samples), Cronbach 's a, Pearson' s R, and ICC three reliability coefficient effects are supported by Cronbach 's a continued meta analysis of.Meta analysis results. CI (0.81,0.86); other areas for 0.77,95%CI (0.74,0.80). Heterogeneity test: all research values were 74.8% (P0.001); subgroup analysis, Europe, America, Asia, and other regions, I2 values in turn were 71.2% (P0.001), 68.2% (P0.001), 41.5% (P=0.14).Meta regression analysis showed the sample size, age, population, test version, region and a coefficient effect multivariate regression correlation The coefficient of coefficient R=0.56,5 was 30.9%, the coefficient of a coefficient effect was 30.9%, the regression coefficient of test version and population two variable reached significant [-0.34 (P0.05), -0.61 (P0.001)]. That is, the revised version had a significant positive effect on the a coefficient. The result of factor structure analysis showed that the impulse dimension measured by BIS-11 was not suitable for the single dimension model, most of the research branch was used. With the Barratt three factor model, the results of confirmatory factor analysis showed that the high order factor model of the nested Patton six factor model also had better model fitting results..2. had good internal consistency reliability in the total sample and four groups of youth samples, and there was a better internal consistency reliability in the three subscales and the total scale, except for the cognitive scores of the normal adolescents and the total samples. The internal consistency reliability of the scale was less than 0.80 (0.67,0.79), and the others were above 0.80. The split half reliability of the total sample BIS-11 was 0.88, the retest reliability was normal young people. The total table and the difference of the scores of the 0.79,0.72. four adolescents were statistically significant (P0.001). The structural validity analysis showed B The three factor model of arratt [total sample: RMSEA=0.047,95%CI (0.044,0.050), AGFI=0.91], Patton's six factor model is better fitting [total sample: RMSEA=0.055,95%CI (0.052,0.058), AGFI=0.91], and double factor model also has good fitting effect [total sample G-3 model: RMSEA =0.041,95%CI (0.039,0.044)). 95%CI (0.052,0.058); CFI=0.85], the high order factor model is not fit. Conclusion: 1.BIS-11 has good reliability in different regional cultural use. The scale has better cross-cultural adaptability. Population, test version has a positive predictive effect on a coefficient, and the internal consistency of the clinical population is higher than that of non clinical population, BIS-11 revision. Compared with other versions, the Chinese revised version is.BIS-11 in normal, and the analysis of abnormal adolescents shows better psychometric characteristics when the.2. three factor structure shows good stability. The three factor structural model in Chinese teenagers also has better validity, and the three factor theory can be preferred when interpreting BIS-11 scores. The two factor model can be used as a new perspective of multidimensional structure to explain the impulsivity dimension.
【学位授予单位】：皖南医学院
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：B841.7

【参考文献】