causal modelling

causal modelling

a family of techniques of statistical modelling aimed at providing specification and testing of the causal relations underlying correlations between a number of variables. Based on the work of Herbert Simon (1957a), and pioneered in sociology especially by Hubert Blalock (1961), the approach requires the researcher to formulate and test successive theoretical models of the causal relations between variables, seeking a model which best fits the data. Included under the general category of causal modelling are PATH ANALYSIS and LOG LINEAR ANALYSIS. Causal modelling has been criticized as dependent on initial assumptions which cannot be regarded as fully tested by the data. Technical sophistication may disguise this. Nonetheless, causal modelling is important in making possible a more satisfactory exploration of causal relations than is achieved in simpler forms of correlation analysis.

Collins Dictionary of Sociology, 3rd ed. © HarperCollins Publishers 2000