Models and Methods
Structural equation models have achieved increasing popularity in the social sciences. Much of the credit for this popularity can be attributed to the flexibility and power of the methods themselves. Equally important has been the availability of computer software for performing the modeling process.
An enormous amount of material has been written on structural modeling. There are now numerous textbooks and monographs for the beginner. Users in need of such materials are referred to books by Long (1983a, b), James, Mulaik and Brett (1982), Kenny (1979), Everitt (1984), Loehlin (1987), Bollen (1989), and Hayduk (1987) among others. All of these books have significant virtues. The reader with a serious interest in the subject should probably at least browse through all of these books.
For a very interesting debate on the value of structural models in the social sciences, the Summer 1987 issue of the Journal of Educational Statistics is strongly recommended. This issue contains a critique of path analysis by D.A. Freedman, and responses to that critique by a number of writers.
Discussion of the deeper aspects of the theoretical connections between causal inference and statistical modeling is beyond the scope of this chapter. For important and illuminating accounts of this topic, see the books and papers by Scheines (1994), Spirtes, Glymour, and Scheines (1993), and Steyer (1992, 1994).