抑郁在青少年核心自我评价与自杀意念间的中介作用：基于Bootstrap法和MCMC法的实证研究

摘要： 考察抑郁在青少年核心自我评价与自杀意念间的中介作用。对502名高中生进行量表测评。通过偏差校正的Bootstrap法和有先验信息的MCMC法求出中介效应值的95%置信区间分别为[-.030,-.011]和[-.024,-.014]，提示抑郁的中介效应显著。效应量k2、R2med分别为.124、.104，偏差校正的Bootstrap法抽样5000次后，构建的效应量的95%置信区间分别为[.070,.178]、[.063,.156]，两种指标共同验证效应量为中等。研究结果说明抑郁在核心自我评价与自杀意念间起部分中介作用，效应量中等。
Abstract: The adolescent period is considered to be difficult and critical. Adolescents have to face a host of pressures, from changes of puberty to questions about who they are. Seriously depressed teenagers often think about death or suicide, which was associated with higher negative self-evaluation. Although suicidal ideation are relatively common among adolescents, many questions on their causes remain unsolved. Many studies have found that there was a close relationship between core self-evaluation, depression and suicidal ideation. For the majority of suicidal adolescents, depression plays a primary role. However, few studies focus on the mediating role of depression between core self-evaluation and suicidal ideation. Findings from this study are intended to provide empirical support for the strategies for the adolescents’ suicide prevention. To determine the specific effect of depression on relationship between core self-evaluation and suicidal ideation in adolescences, a total of 502 senior high school students were enrolled and measured on a series of questionnaires. Self-evaluation was assessed with Core Self-evaluation Scale, depression was measured by Self-rating Depression Scale, and suicidal thought was examined by Self-rating Idea of Suicide Scale. Two computer-intensive methods, Bootstrapping and Markov chain Monte Carlo methods were used to construct confidence intervals for the mediating effect. This was done using Mplus version 7.0, which involved repeated sampling from data set and estimating indirect effect in each resampled data set. Firstly, 5000 bootstrap samples were drawn with replacement from the original data, and the mediation model was re-estimated and the mediating effect was calculated within each bootstrapping sample. Secondly, the informative Markov chain Monte Carlo method was conducted, and two Markov chains and Gibbs sampler were used to sample from data set for 20,000 times. The 95% confidence intervals for the indirect effect were constructed through both Bootstrapping and MCMC methods. k2 and R2med were used to calculate the effect size of the mediation model. There were significant correlations between core self-evaluation, depression and suicidal ideation among adolescents. In addition, the bias-corrected Bootstrapping and Markov chain