Many social psychological phenomena exhibit convergent amplification, i.e., homeostatic or error-cancelling tendencies, the results of which can be studied by currently popular research and statistical methods. We argue, however, that at least as many social psychological phenomena exhibit divergent amplification, i.e., heteromorphic or catalytic tendencies, the results of which cannot be studied by currently popular methods. The latter phenomena are usually overlooked or are dismissed as spurious or scientifically intractable. But recent advances in the physical and biological sciences demonstrate both the possibility and utility of studying these phenomena within an expanded scientific framework. We review these advances, discuss their implications for research and statistical practice in social psychology, and explain the scientific advantages of analyzing social behavior in multifarious contexts.