Search

This article describes how laboratory experiments are used by social psychologists and those in related fields to study conflict, negotiation, and mediation. In a laboratory experiment, the researcher experimentally controls one or more variables in controlled, artificial settings that induce processes likely to occur naturally. Laboratory experiments are the primary method used to establish cause-and-effect relationships among variables and to reduce alternative explanations; thus they are primarily concerned with validity of explanation. In this article, we describe several basic design features including conceptual replication, precise manipulation, and the use of a moderator variable, which all help assess the processes underlying a research finding. These design features foster conceptual internal validity, which describes the impact of one variable on another and the quality of such an explanation. Conceptual internal validity provides a basis for generalization of findings and thus new research. It also fosters strong inference, which helps build cumulative knowledge. Laboratory experiments are not well suited to answer all questions and problems – for example, they may not produce deep understanding of a particular historical event and must be supplemented by other methods, such as surveys and case studies.

In this article, we compare the relative popularity of a wide variety of methods and techniques used in the study of conflict and negotiation across five domains of inquiry: political science, communication sciences, social and personality psychology, economics, and organizational behavior. An analysis of articles on conflict and negotiation published between 1997 and 2001 suggests that laboratory experiments that entail coding of behavior and self-reported data using surveys are especially popular in psychology, organizational behavior, and communication sciences. Mathematical modeling, the use of experimental games, and the use of archival data are especially popular in economics and political science. Diverse methods can provide convergent insights, and this is observed clearly in work on gain-loss framing and on reciprocity in negotiation. We suggest that researchers adopt, or continue to employ, triangulation as an approach to validity: When two or more methods or data sources converge on a construct, we develop greater assurance that our conclusions are not driven by an error or artifact of any one procedure. Each method exhibits strengths and weaknesses, and to the extent they do not overlap but show common effects, we stand on more solid ground with our theoretical conclusions.