Development and Psychopathology, 14 (2002), 653–666 Copyright  2002 Cambridge University Press Printed in the United States of America DOI: 10.1017.S0954579402003139 Levels of analysis in etiological research on schizophrenia TYRONE D. CANNON AND ISABELLE M. ROSSO University of California, Los Angeles Abstract Etiological research spanning domains of inquiry as diverse as social psychology and molecular genetics has identified a number of potential factors that likely contribute to the development and clinical manifestation of schizophrenia. In this article, we first highlight the challenges inherent in developing cogent etiological models that represent both the diversity of suspected causal influences and their mechanisms of action. Then, using our own research program as a heuristic context, we present a general analytical framework for identifying and integrating multiple types of etiologic factors across different levels of analysis in the prediction of schizophrenia. In recognition of the myriad complexities of multifactorial causation, we argue that a multilevel causal perspective is required for the development and advancement of a fully nuanced theory of schizophrenia etiology and pathophysiology. In schizophrenia, as in other domains of psy- havioral outcome but do not necessarily oper- ate via precisely the same mechanisms. The chopathology, research has revealed a diver- sity of factors thought to represent causative challenge for psychopathology researchers is thus to specify etiologic models that integrate influences (e.g., Cannon, 1997; Carlsson, 1995; Cohen & Servan–Schreiber, 1992; Karayior- putative causal influences across multiple lev- els of analysis (Carpenter, 1987; Engel, gou & Gogos, 1997; Selemon & Goldman– Rakic, 1999). In some cases these factors de- 1977). The primary goal of this article is to describe an analytical framework for such in- rive from the same general domain of inquiry (e.g., anatomical brain lesion and altered neu- tegration. In so doing, we provide a concep- tual overview of work conducted in our labo- rotransmitter function), and in some cases they derive from quite disparate domains (e.g., so- ratory and elsewhere as it relates to a multilevel perspective on the etiology of schizophrenia. cial cognitive and biological influences). In either case, such findings imply multifactorial causation in which the various causal influ- Typology of Causal Influences ences ultimately coalesce in producing a be- It is important to recognize at the outset that most forms of psychopathology are complexly The research described in this article was supported by determined phenomena, whereby multiple caus- grants from the National Institute of Mental Health al influences that may be separated in time (MH52857 and MH48207), Theodore and Veda Stanley Research Foundation, and March of Dimes Birth Defects and place and distinct in terms of mechanism Foundation, and by gifts from the Rutherford Charitable of effect, integrate in the determination of a Foundation and the Staglin Family. (not necessarily static) behavioral endpoint (Cic- Address correspondence and reprint requests to: Ty- chetti & Tucker, 1994). In considering the rone D. Cannon, Department of Psychology, University ways in which two or more causal influences of California, Los Angeles, 1285 Franz Hall, Los Angeles, CA 90095; E-mail: cannon@psych.ucla.edu. may be integrated, it will be heuristic first to 653 NOTICE: This material may be protected By copyright law (Title 17 U.S. Code)