Standardized Regression Coefficients

Abstract

In clinical efficacy studies the outcome is often influenced by multiple causal factors like drug-noncompliance, frequency of counseling, and many more factors. Path analysis, structural equation modeling, and Bayesian networks are able to account not only Bayesian Networks direct but also indirect relationships across a network of variables. Unfortunately, p-values are not provided, but Akaike information indexes can be used to assess and compare better and worse networks. In this chapter the results of traditional regression analyses were compared with those of the novel networks. It is shown, that worthwhile additional information is given by the network methodologies.