In Chapter 8, we noted that studies employing different statistical techniques had often produced inconsistent results, even within the same data set. Such findings raise the specter of conclusions about the presence and form of interaction adaptation patterns being tied to method. Faced with the dilemma in our own research of choosing the most appropriate and informative design and statistical analysis options, we decided to compare several methods empirically to see concretely just how much results might differ when applied to real interaction data. Our conceptual comparisons led us to anticipate that rather than producing contradictory results, different methods applied to the same data set would produce complementary results but might vary in the unique insights they offered. In this chapter and the next, we report several such comparisons conducted on data from experiments employing actual dyadic interaction. We chose these data sets because of our substantive interest in the patterns that might be uncovered as well as our methodological interest in what the different statistical analyses might yield.

For our first analysis, we selected a data set from an experiment expressly featuring changes in conversational involvement and conducted within an interview context.