ABSTRACTS.

The article presents abstracts of two articles published in the September 1, 1974, issue of the periodical "Quality and Quantity." One article focuses on structural models of voluntaristics social action. Using a broad characterization of recent liberal, pluralistic, social science research interests defined in terms of Parsonian social theory, this article reviews recent developments in social science statistical practices. These developments are seen as converging toward a cross-disciplinary synthesis of rigorous "structural" or "causal" modeling approaches. The major negative thesis of the paper is the frequent inadequacy in empirical practice of simply specified, statistically manipulable structural models. The other article focuses on change and unreliability. The analysis is concentrated on the dichotomous case, two observations, and two response classes. Using a latent structural approach, two true latent classes are analytically separated from two manifest response categories.