Unionisation in a comparative neural network model: a trade union membership prediction in 12 states

The sociometric and theoretical models traditionally associated with the determinants of unionisation and based on international comparisons, generally use linear analysis of a restricted number of complex variables. The effectiveness of such models often have an unsatisfactory result. In this study a different theoretical and methodological approach has been used. It is based on a integrated scheme made by transposing the principles of the "Squashing Theory" (Buscema, 1994) into the social field, by the connectionist paradigm, and by the Artificial Neural Networks (ANN). This has permitted the carrying out of a larger data base (49 variables, 12 Nations, 12 Years maximum time interval), the adjusting of a model that has an elevated explanatory capacity (in terms of open variance) an ulterior capacity of generalization (in terms of a one-year span) and of metageneralization (in terms of generalization to nonprocessed nations). What emerges is surely the need for more and greater in-depth study of socializing facts which follow such aspects.