| Objective: 1.To systematically evaluate the reliability and validity of the Barratt Impulsiveness scale(BIS-11),investigate the difference in reliability and validity between different regional cultural backgrounds,and look for the predictors influencing the reliability and validity of BIS-11.2.To test the reliability and validity of Chinese BIS-11 among different adolescent samples,and test the cross-cultural adaptation of this scale,as well as its cross-crowd difference.Methods: 1.The studies related to the reliability and validity of BIS-11 that were generated from 1995 to 1 January 2016 were searched by computer in the Pubmed,Science Direct,wiley-Blackwell,Wanfang Data,CNKI,Chinese Sci-Tech Periodical Full-text Database,and xueshu.baidu.com.Studies were screened and data was extracted in accordance with the inclusion and exclusion criteria.The a coefficient effect size conversion method used by Rodriguez and Maeda was adopted for Cronbach's alpha,Fisher's Z transformation was adopted for test-retest reliability.Then,the metacor package in R language was used for meta-analysis of reliability coefficient effect size,to analyze the difference in reliability between different regions;SPSS 18.0 was used for weighted regression analysis of reliability coefficient,to identify the predictors of reliability coefficient.Data analysis was conducted to sum up the factor structure of BIS-11.2.The reliability and validity of Chinese version of BIS-11 were retested among four different groups of adolescent samples including normal adolescents,adolescentswith mental disorder,delinquent juvenile and adolescents with Internet addiction.SPSS 18.0 was used to test the total BIS-11 score of different adolescent samples and master samples,as well as the internal consistency reliability and split-half reliability of sub-scales,to calculate the test-retest reliability of normal and abnormal adolescents,and the difference between the scale score of different samples and the score of various sub-scales,and verify the three-factor model,six-factor model,two higer order models(including two second-order factors and three second-order factors)and two bi-factor models among different adolescent samples and the master samples.Results: 1.35 studies(including 45 samples)were included in meta-analysis;the funnel plot result of three reliability coefficient effect sizes including Cronbach's alpha,pearson's r and ICC supported continual meta-analysis of Cronbach's alpha.The result of meta-analysis showed that the Cronbach's alpha coefficient of all samples was 0.81,95%CI(0.79,0.83);0.79,95%CI(0.77,0.82)in Europe and America;0.84,95%CI(0.81,0.86)in Asia;0.77,95%CI(0.74,0.80)in other regions.Heterogeneity inspection: the 2 value of all studies was 74.8%(P< 0.001);the 2 value of subgroups analysis of Europe,America,Asia and other regions of were followed by 71.2%(P< 0.001),68.2%(P< 0.001)and 41.5%(P=0.14).Meta regression analysis showed the correlation coefficient R of multiple regression with sample size,age,populations,test versions,regionswas 0.56,these five variables explained 30.9% of the total variance,the test versions two significant variable regression coefficient and the crowd [0.34(P< 0.05),0.05(P< 0.001).The test versions and populations had positive effect on the alpha coefficient[-0.34(P<0.05)?-0.61(P<0.001)].The result of factor structure analysis showed that the dimensions of impulsivity measured by BIS-11 did not adapt to unidimensional model.Most studies supported the original three-factor model of Barratt,and the result of confirmatory factor analysis showed that it also had a good model fitting result for the high-order factor model nested with Patton six-factor model.2.In the master samples and the four groups of adolescent samples,the three sub-scales and total scale had a good internal consistency reliability,and the internal consistency reliability in the cognition sub-scale of the normal adolescent samples and master samples was lower than 0.80(0.67 and 0.79 respectively),while it was higher than 0.80 in other samples.The split-half reliability coefficient of the total sample was 0.88,test-retest reliabilities of normal adolescent and adolescents with Internet addiction were 0.79,0.72.The difference in score between the total scale and sub-scales of the four groups of adolescent samples showed statistical significance(P<0.001).The structure validity analysis :Barratt's three-factor model and Patton's six-factor model showed good model fit [G-3 models of the total sample: RMSEA = 0.041,95% CI(0.039,0.044);CFI = 0.91.G-6 model of the total sample: RMSEA = 0.055,95% CI(0.052,0.058);CFI = 0.85].While higher-order factor model showed bad model fit.Conclusions:1.The BIS-11 used in different regional culture all had good reliability,therefore this instrument has good cross-cultural adaptation.Populations,test versions have positive effect on the alpha coefficient of BIS-11.There is higher internal consisistency among clinical population than non-clinical population,and the revised version of the BIS-11 has higher reliability than other versions.Chinese version of BIS-11 has good psychometric properties among normal and abnormal adolescents.2.The meta-analysis indicatesthe three factor structure display greater stability,and this study in four samples of adolescents also verified the three-factor model.For BIS-11 score interpretation,the three-factor model should be first selected for explaining the dimensions of impulsivity,while the bifactor model can be used as a new perspective of multidimensional structure to explain the dimensions of impulsivity. |
PDF Full Text Request