Die Mehrebenen-Analyse. Ihre praktische Anwendung und theoretische Annahmen

Multilevel analysis has become more common during the last ten years. A reason might be that this method allows us to analyse micro- and macro relationships in one step. Loosely, one can characterize it as a combination of variance analysis and regression analysis. Nevertheless, questions occur frequently concerning its practical application, the statistical requirements, the differences to OLS-Regression, and so on. Answering these questions, this paper has two particular aims. The first aim is to show the practical use and to give an idea of the possibilities of this method. The second aim is to discuss statistical requirements and frequently asked questions. For this purpose the application of a two level regression is illustrated by using data from ISSP and analysing the perception of vertical conflicts in an international view. This paper starts with a practical application. The random intercept model is discussed, where the intercepts vary among macro units and therefore regression lines run at different levels. Afterwards a random slope model is shown. Here, the regression coefficients also vary among countries. Among these practical applications other topics are discussed, e. g. model fit, minimum sample size, explained variance, estimation-methods, and standardisation of coefficients. Finally there is given a brief overview of this method and different software programs.