Structural modeling for dichotomous variables with LISREL is put into operation. Random samples from elementary and secondary school teachers take part in the survey on neuroticism and burnout. The correlation between neuroticism and burnout is estimated in the form of latent variables, with the hypothesized model getting adequate support from empirical data. Since data from EPQ are dichotomous, we have to compute polycorrelation matrix and asymptotical covariance matrix from raw data, which are introduced into the structural modeling with LISREL. Based on 856 available samples, the result shows that (1)neuroticism subscale acquires adequate internal consistency and constructive validity; (2)neuroticism acts as a significantly positive predictor for emotional exhaustion; (3)the model which emotional exhaustion partially mediated the relationships between neuroticism and depersonalization achieves moderate fit indices. It is suggested that any analogous research can follow the steps we propose.%文章探讨如何使用Lisrel进行二分数据的结构建模．随机调查中小学教师的神经质人格特质和工作倦怠状况，通过对预设模型的检验，从潜变量层面考察神经质对工作倦怠的影响．由于EPQ的数据形式为二分数据，使用Lisrel建模时，需要先从原始数据计算出多项相关矩阵和渐进协方差矩阵，然后根据这些矩阵估计相应的模型参数．856份有效问卷的数据分析显示：①神经质分量表具有较好的内部一致性信度和结构效度；②神经质对情绪衰竭有显著的正向预测作用；③情绪衰竭在神经质与去个性化之间发挥部分中介效应，中介效应占总效应的72．8％．Lisrel关于二分数据的结构方程建模方法将对社会科学研究起到一定的启示作用．
展开▼
