Experimental Versus Nonexperimental Research

A between-participant experiment requires that at least one independent variable is manipulated, and participants must be ran­domly assigned to the levels of the experimen­tal independent variable.

Experiments in which conflict is the inde­pendent variable typically involve one of the following procedures used to manipulate the incompatibility of goals, the incompatibility of actions, or the interdependence of the parties:

• In a group, creating some level or type of conflict and observing the resulting com­munication

• Providing individuals with a hypothetical scenario involving some level or type of conflict and then having them respond to a questionnaire

• Having individuals play particular roles (e.g., employer vs. employee; partners working on a school project) and observing the resulting communication

Note that conflict is generally not used as a dependent variable; instead it is represented by variables that indicate how the conflict is man­aged (the process at work). Variables that may be independent or dependent include conflict styles (e.g., Cai & Fink, 2002; Oetzel & Ting- Toomey, 2003), and variables that indicate the outcome of a conflict include satisfaction with the other party, negotiated outcomes such as profits, and the degree to which there has developed an impasse in the relationship (e.g., Cai et al., 2000; see the discussion of outcome and process variables below).

In their book on processes of social con­flict, Pruitt and Carnevale (1993) noted the difficulties in studying conflict. Experimental studies provide control of extraneous vari­ables but in a simulated setting. On the other hand, manipulating conflict in laboratory or field experiments raises other concerns: If the manipulation is effective, the ethics of the investigation may be problematic; if the manipulation is ineffective, the lack of internal validity means that the investigation cannot inform us about conflict processes. Naturalistic studies are hard to come by and, when avail­able, have their own problems of internal and external validity. Furthermore, in every­day settings, individuals and organizations are often hesitant to allow researchers to observe sensitive, confidential, or private interactions associated with conflict.

Laboratory experiments and field exper­iments differ in the location of the data collection, and this factor has implications for external validity. In addition to random assignment, other issues that are relevant to experiments include the following: (a) whether the study is double blind, meaning that nei­ther the experimenter nor the participant is aware of the conditions to which the par­ticipant is assigned; this approach can counter any expectancy effects; and (b) whether the manipulations are effective in bringing about the states or conditions associated with the manipulations. For example, we should deter­mine whether our attempt to create a conflict condition and no-conflict condition in small groups actually creates groups that differ in level of conflict. Two methods for determining the effectiveness of the manipulations are the use of pilot studies prior to the main research and the use of manipulation checks during the main research (but see O’Keefe, 2003 ). Conflict research often either assumes or requires that the parties to the conflict experience or express some level of emotion, such as anger, hostility, or jealousy. Manipulating these emotions can be quite challenging, both because they are difficult to elicit in a laboratory setting and because ethically the researcher must be care­ful regarding the potential risks to the partici­pants in eliciting such emotions.

And of course culture can play a role in the validity of the research. Attributions, behav­iors, and expectations about conflict may not be relevant to certain cultures. Assuming a shared understanding across cultures about what a conflict is, how it should be managed, and what emotions are likely to be involved can lead to incorrect conclusions about the culture and about how people communicate about conflict.