John A. Sonquist

Problems in the Analysis of Survey Data, and a Proposal

Abstract Most of the problems of analyzing survey data have been reasonably well handled, except those revolving around the existence of interaction effects. Indeed, increased efficiency in handling multivariate analyses even with non-numerical variables, has been achieved largely by assuming additivity. An approach to survey data is proposed which imposes no restrictions on interaction effects, focuses on Importance in reducing predictive error, operates sequentially, and is independent of the extent of linearity in the classifications or the order in which the explanatory factors are introduced.

Networks of Power: Organizational Actors at the National, Corporate, and Community Levels.

Large organizations, particularly corporations, possess considerable resources, and with that comes considerable power, often extending beyond a single community or nation-state. Networks among large corporations enhance that power to the point that they exert a major impact on national and multinational economies and policies, influencing decision-making to achieve their own goals. Networks of Power applies interorganizational analysis to the study of power in three main areas: national policy domains, community influence structures, and national corporate structures. The main body of the text is comprised of original research by the leading authorities in the field and covers such areas as national policy decisions in health and energy, corporate structure, innovative theoretical and methodological approaches, and a critical review of network analysis of interorganizational relations and power. Also presented is an agenda for future research.

FINDING VARIABLES THAT WORK

This paper outlines a strategy for using two large-scale computer programs, the Automatic Interaction Detection and Multiple Classification Analysis algorithms. They supplement each other in the inductive task of formulating a model of the simultaneous effects of a set of independent variables on a criterion. MCA handles correlated predictors well, but assumes away interactions. Terms representing this type of joint effect are located by preliminary AID runs. John Sonquist is Research Associate and Head of the Computer Services Facility, Institute for Social Research, University of Michigan.