潜在类别分析原理及实例分析

Principle of Latent Class Analysis and Case Analysis

目的介绍潜在类别分析的原理、方法和技术,探讨潜在类别分析在多个二分类反应变量中聚类的应用。方法采用Mplus软件,对太原市青少年吸烟知识的调查问卷进行潜在类别分析。结果 3804名学生总的被分为5个潜在类别,每一类别的人数分别为2879人、367人、248人、234人和76人,各类别的概率分别为0.757、0.096、0.065、0.062和0.020。结论潜在类别分析用于多个二分类反应变量的聚类有很好的效果。

Objective To Introduce the principles, methods and techniques of the latent class model to explore the application of latent class analysis for classification of multiple binary response variables. Methods Using latent class model and applying Mplus software, we gave 3804 adolescents, who came from Taiyuan a classification according to their knowledge about smoking. Results The population including 3804 individuals is divided into 5 latent classes, and the five classes＇ number of adolescents was 2876,367,248,234 and 76 respectively, and the probability of the five classes is 0. 757,0. 096,0. 065,0. 020 and 0. 062 respectively. Conclusion The latent class analysis has a good effect on classification of multiple binary response variables.