The Unit of Analysis
A critical issue for studies of interaction research concerns the unit of analysis (Sillars & Overall, this volume). Units of analysis represent oneās conceptual understanding regarding where one can find the best information regarding communication (Heyman, 2001). Accordingly, units of analysis reflect oneās perspective regarding the nature, function, and scope of conflict interaction. Oneās selection of the unit of analysis constitutes a critical first step in conducting interpersonal conflict processes.
Researchers tend to select one of three types of units. Time is one type of unit of analysis. Interpersonal conflict researchers tend to select very brief periods of time for observation (e.g., 15 seconds to 1 minute) (e.g., Sillars, Roberts, Leonard, & Dun, 2000). Although small units of time allow the application of codes in precise ways, time as a unit of analysis requires a researcher to punctuate conflict interaction according to predetermined durations, regardless of whether more or less time would help gain insights about conflict communication.
Other researchers rely on the thought unit for coding. The thought unit is the āsmallest semantic unitā (Notarius, Markman, & Gottman, 1983, p. 119), for example, one-word responses, independent clauses, and even short nonverbal behaviors that convey an idea (e.g., āughā can mean ānoā). Thought turns clearly portray shifts in ideas and arguments, which can assist researchers who examine how partners develop ideas with each other (Canary, Brossmann, Brossmann, & Weger, 1995). Yet thought turns vary in their inclusiveness across studies; for example, extensive development of one idea might entail one thought or a set of different subordinate thoughts. One way to constrain how much data one partner provides is to code for both time and thoughts. However, Bakeman and Gottman (1986) argued that duration of time is implied when coding behaviors.
Third, researchers can rely on speaker turns as their unit of analysis. That is, switches in who holds the communication floor determine oneās unit of observation. Although speaker turns appear clear and reliable, they remain vulnerable to within- and between-partner variability. For example, it is not uncommon for one person to talk more than does his or her partner. In such cases, the loquacious person could be elaborating quite a bit on one idea in long speaker turns, whereas the less verbal partner might offer only a couple of sentences in short speaker turns, yet both personsā speaker turns would be seen as equal.
One important decision concerns whether microanalytic units of observation or macroscopic units offer the better choice to observe interpersonal conflict. Naturally, data analyses depend on and are constrained by oneās unit of analysis (e.g., coding every 30 seconds vs. paragraphs of speaker turns; Floyd, OāFarrell, & Goldberg, 1987). Analysis of microanalytic conflict communication yields very precise results regarding specific behaviors. Researchers often build macro-level measures from micro-level units; that is, they decide to observe communication as it emerges in small units (e.g., at the tactical level, such as problem description, solution proposal, positive metacommunication, among many others) and then compile these small tactical units into relevant and more inclusive strategies (integrative behaviors, withdrawal, etc.). Naturally, one cannot deconstruct macro-level codes into micro-level behaviors, because the more specific information simply is not there.
Additionally, researchers have constructed rating systems. One clear advantage of rating systems versus coding systems is that rating systems require much less time for rater training (Floyd et al., 1987). Moreover, rating systems tend to provide mezzo-level data (i.e., ratings of data are more abstract than are codes of the same data).
Three Popular Conflict Coding Systems
As indicated, specific microanalytic codes researchers use to categorize conflict communication systematically operationalize conflict behaviors (see Sillars & Overall, this volume). Over 20 conflict coding systems exist (Woodin, 2011). To provide the reader a concrete grasp of conflict coding systems, in this subsection I summarize three popular coding systems. For other schemes and in-depth analyses of codes, see Kerig and Baucom (2004), Filsinger (1983), and Sillars and Overall (this volume).
First, Rausch et al. (1974) pioneered observational conflict research and derived a coding scheme that contained cognitive acts, affiliative acts, and coercive acts. Cognitive acts include the following behaviors: opening the issue or probe, seeking information, suggesting a course of action, agreeing with the otherās statement, and others. Affiliative acts involve using humor; accepting the other personās plans, actions, ideas, motives, or feelings; seeking reassurance; introducing a compromise; and so on. Finally, coercive acts include using power to force the other person to agree, rejecting the other, demanding compensation, disparaging the other, and the like.
Second, the most widely used conflict coding system is the Marital Interaction Coding System (MICS) (Weiss, 1993). Heyman, Eddy, Weiss, and Vivian (1995) conducted a factor analysis of the MICS-IV codes, using archived interactions from 995 couples. Four factors defined the specific codes from the MICS: Hostility, Humor, Constructive Problem Discussion, and Responsibility Discussion. Hostility involves a priori blame (with a hostile voice), put-down, criticize, negati...
