Problems of endogeneity in social science research

Many examples can be cited from social sciences where a unidirectional causality between two variables is not meaningful. There is a simultaneous two way relationships between these two. The endogeneity problem arises when we estimate on the basis of such incorrectly presumed unidirectional causality. This problem has received attention not only in economics but also in disciplines like demography and population studies, financial accounting studies, development studies, sociology and physical education. This chapter discusses the nature, sources, and consequences of such endogeneity bias, and necessary solutions. The chapter also discusses two related issues: problem of Identification which concerns whether all the parameters of the system can successfully be estimated; and the endogeneity problem arising out of omitted variables and measurement errors.