HomeKAYÂ TAOvol. 13 no. 1 (1994)

Recent Advances in Causal Modeling: The Case of Latent Variable Path Analysis and Some Recommendations for Training in Social Science Statistical Analysis

Ben Teehankee

Discipline: Social Science, Sociology



Traditionally, those interested in testing complex social science models have resorted to multiple regression and its variants, especially path analysis. Advances in causal modeling techniques in the past two decades are making it possible to formulate and test increasingly complex theoretical formulations in the social sciences using latent variable path analysis (LVPA). The technique allows the causal analysis of multiple variables where some variables, called manifest variables, may be directly observable while others, called latent variables, are not. It is recommended that social science students locally be given a thorough grounding in basic multivariate analysis and, thereafter, to LVPA.