Log-linear analysis
Log - linear models are among the multivariate methods. With log - linear models nominally scaled data are analyzed. Through a logarithmic transformation can be represented, the problem most vividly in the analysis of multi-dimensional frequency tables, for example in the sense that the main effects and interactions of a multi-dimensional frequency table can be composed linearly.
There are different log - linear methods:
- As a general loglinear models are called methods, examine the non-directional relationships between nominally scaled data.
- Logit models to investigate the directional relationship between a dependent variable and other nominally scaled independent variables.
Log - linear models offer the possibility of so-called saturated and nichtsaturierten data analysis.