What's in the mix? Combining coding and conversation analysis to investigate chat-based problem solving

Coding interactional data for statistical analysis presents theoretical, methodological and practical challenges. Coding schemes rely on categories that are decided by their relevance to the analytical problem under investigation. We suggest that (1) endogenous and publicly displayed concerns of participants provide for the observable organisation and achievement of participation, and (2) these observable features of interaction constitute appropriate data for analysis. We identify coding categories using conversation analysis that are consistent with and are constitutive of the production of the local order of interaction. Coding is based on observed units of interaction. With the approach we are developing, we believe that statistical analysis will be more parsimonious and produce a more precise description of group process and interaction.