ANTTILA, STEN TORBJORN; FILDR

                        UPPSALA UNIVERSITET (SWEDEN), 1994

                        SOCIOLOGY, THEORY AND METHODS (0344)
 

                         The consequences of four fundamental methodological problems in macro-sociology are examined. The
                         first problem concerns the use of nation-states as units of analysis in comparative research and is
                         investigated in relation to a correlational study of the causes of democracy. It is argued that when
                         nation-states are the units of analysis, the statistical assumption of independent observations is violated,
                         a difficulty known as Galton's problem. If a holistic approach is adopted where the independence of units
                         is not assumed but investigated, Galton's problem is avoided. This implies, however, that many additional
                         assumptions must be introduced concerning the interrelations among units thus making it practically
                         impossible to test the models under investigation. The second problem regards the choice of empirical
                         criteria used in the periodization of macro-historical processes and is related to studies of capitalist
                         evolution based on analyses of long waves and capitalist organization. It is claimed that a variety of
                         indicators can be specified for the identification of periods when the concepts used in the theory are
                         empirically vague. Consequently, different periodizations can be supported depending on the specific
                         indicators chosen. The third problem refers to the empirical support of macro-sociological theories and is
                         analyzed in relation to the claim decreasing nationalization of politics indicates the end of organized
                         capitalism. It is argued that the linkage between the theory and the empirical illustration is dubious. Owing
                         to implicit assumptions and vague concepts, observable implications cannot be deduced from the
                         theory. It is concluded that when a strictly deductive approach is adopted only a few empirical cases can
                         be studied simultaneously. Finally, the fourth problem concerns the theoretical consequences of social
                         science images of large social systems. This problem is examined in relation to the image of liberal
                         democracy as a political thermostat. It is claimed that verbal images of macro-social processes conceal
                         complex relations. When these images are formally specified as models, the implied pattern of behaviour
                         of large social systems may counter the expectations based on the original image. A computer simulation
                         of the interaction between the electorate and the government is carried out showing that whether or not
                         a model of liberal democracy behaves as a political thermostat depends on the exact way in which the
                         model is formally specified. Assumptions that may seem unimportant or minor modifications of the model
                         may completely alter the behaviour of the system. (Abstract shortened by UMI.)

Social Systems Simulation Group P.O. Box 6904 San Diego, CA  92166-0904 Roland Werner, Principal Phone/FAX  (619) 660-1603