Detection and solving of regression modeling problems in SPSS

Regression modeling has been frequently used in numerous qualitative and quantitative economic analyses. During such type of analyses it is investigated how one or more independent variables explain the dependent one. After selection of relevant variables in accordance with economic theory and practice and estimation of regression model parameters with basic regression diagnostic, it is necessary to examine if the basic assumptions are fulfilled. Namely, even in a situation when the model calculation provides numeric results, there are regression problems causing incorrect regression image. Examples of such problems are autocorrelation of residuals, multicollinearity, heteroscedasticity of residuals variance and normality assumption. In those models the estimated parameters and the representative indicators could be false. The parameters estimation and all relevant testing and solving of the appropriate regression problems are performed and presented in SPSS on the concrete model analysis of economic variables. This approach can be useful for the students of the faculties of economics as part of their scientific and specialized analyses.