Methodische Flexibilitaet: Ein Ansatz zur Lösung des Software-Dilemmas in der Statistik

abstract: standard-software for statistics and related fields has been developed for more than 20 years. currently a broad segment of the market is covered by huge and monolithic program systems having their roots in the sixties. these programs are difficult to maintain and there are no facilities for changing and improving them by the user. one possible way to fight this software crisis and to remove the discrepancy between the wishes of the users and the current state of statistical software is the development and implementation of concepts that introduce a new level of flexibility. this paper mainly deals with the aspects of methodological flexibility and the tools that must be supplied to enable the user to extend existing and develop new methods. the final chapter describes and compares implementations of methodological flexibility in five statistical packages (tsp , rats, sas/iml, s, gauss).;