Integrative Multi-Method Research

Abstract

The social sciences are in the middle of a boom in multi-method research. An increasing number of books and articles combine techniques from different methodological families within a single study. In conjunction with this trend in application, there has been a marked increase in methodological debate regarding the merits of and best practices for multi-method designs. This book advances the proposition that well-designed and well-executed multi- method research has inferential advantages over research relying on a single method. I argue that multi-method research can test assumptions that are generally untested in single-method research, thereby transforming key issues of descriptive and causal inference from matters of speculative assertion into points of empirical debate. Yet in order to realize these advantages, multi-method research must be designed from the start with a clear focus on testing assumptions – a priority that informs decisions about case selection, statistical analysis, and the substantive targeting of qualitative inquiry. While multi-method research has potential advantages for diverse goals, including concept formation and refinement (e.g. Pearce et al. 2003), description (Campbell and Fiske 1959; Eid and Diener 2006), and applied policy evaluation (Smith and Lewis 1982; Greene, Caracelli, and Graham 1989), this book focuses on designing multi-method research for causal inference. This emphasis is not intended as a slight against the other families of goals just listed. After all, these goals are deeply interrelated. Good conceptualization and description in particular are essential components of successful causal inference; causal claims, in turn, are routinely central to work in normative theory and policy evaluation. Instead, causal inference is emphasized because it has been the primary focus of sustained debate regarding multi-method social science, with some scholars arguing that multi-method work has no advantages for this goal vis-a-vis single-method designs (Beck 2006; Ahmed and Sil 2009; Kuehn and Rohlfing 2009; Beck 2010). Hence, showing how multi-method research can improve causal inference is more urgent than demonstrating its less-contested role in other domains. Multi-method designs are not a panacea for the challenges of causal inference. Even the best designs will leave some issues unaddressed, and some common designs arguably have little advantage in comparison with single-method designs. Nonetheless, when carefully constructed and executed, multi-method research can make a major contribution to the social sciences. This book is intended to help scholars design, execute, and evaluate such research.