Markov Analysis

Markov analysis provides a means of analyzing sequences, often called “chains,” to determine whether some sequences occur more frequently than expected due to random chance. Communication applications of this technique usually involve an analysis of the sequence of moves or issues in a conversation. The focus becomes whether certain sequences take place more often than other sequences. Markov analysis can be used to also examine whether certain sequences occur less frequently than would be expected due to random chance. The Markov analysis examines the probability of particular sequences to determine whether or not the sequence occurs more frequently than would be expected due to random chance.

Consider the simplest set of binary possible combinations in a conversation, a question and answer. If one wished to establish that a question elicits an answer, the use of Markov analysis would provide a means for establishing such a claim. What would happen is that if a question is asked, the next move in the conversation should be an answer. The conversation becomes broken up into units, each of which is coded. The question of the probability of an event coded as A is based on the probability of events B, C, or D taking place. Markov analysis argues that A is followed by B more than expected (e.g., compared to a random set of responses possible to the first event A).

[Page 906]This entry examines using and coding transcripts, establishing and evaluating probabilities, the kinds of claims and outcomes one can establish based on probabilities and the application of Markov analysis in the context of communication research. To explore these topics, the entry will rely on a core example; a series of recorded and transcribed conversations from a divorce mediation between a husband and wife (it is assumed here that the couple is an opposite gender couple to reduce confusion but the same analysis could be done with husband1 and husband2 and wife1 and wife2) involving a single mediator who is mediating a discussion on child custody and placement issues. In this example, the unit of analysis is the conversational turn, regardless of length. Interjections and interruptions that fail to qualify as a codable element are not considered. Each person takes a turn in the conversation and that turn is then coded. In this example, it is also assumed that the coding involved three possible moves that a spouse can make: (a) attack (e.g., say something derogatory about the other spouse); (b) bolster (e.g., defend the self against a perceived attack); and (c) integrate (e.g., accept or work toward an agreement of a proposal or statement). It is further assumed that a mediator’s move can be coded as one of three possibilities: (a) a regulative move (a move that regulates or controls the interaction), (b) a development move (a turn that tries to develop a resource or option or propose something for consideration to the spouses), and (c) a solidification move (a move that positively acknowledges some element of the situation). The research question is whether or not some moves are more productive for a mediator to make in response to an attack sequence. Essentially, when a husband and wife make an attack on each other, should the mediator make a particular conversational move to reduce the level of attacking?

...