Forecasting international relations: A proposed investigation of three-mode factor analysis

Abstract

Patterns, indicators, and forecasts of international environments and behavior are of interest to policymakers as they try to link national goals and policy alternatives and to scholars as they try to develop a “meteorology” of international relations. A problem exists, however, in delineating trend patterns so that precise and reliable forecasts can be made. This is a methodological problem arising from the nature of our data on nations and the many variables that need be analyzed.

Research is proposed on three-mode factor analysis as a possible solution to this problem. Suggested is 1) an analysis of 1962–1968 dyadic conflict (U.S.S.R. -U.S., China—U.S.S.R., etc.) by month and comparison with alternative methods, 2) an analysis of 1962–1965 dyadic conflict behavior and a test of our ability to forecast to 1968 conflict by month from the results, and 3) a series of artificial experiments on contrived data with known patterns and trends. This research should better enable us to judge how well three-mode factor analysis contributes to clarifying patterns, indicators, and forecasts of international environments and behavior.