ABSTRACT
This chapter introduces quantitative methods useful for identifying patterns in online talk. The first step is often the transformation of message text into coded values through content analysis. This has traditionally been done manually, but new computational methods incorporating natural languages processing techniques now also offer semi- and fully automated approaches. Once quantitative representation of the data is generated, various statistical and computational analyses are possible. Statistical analyses involve either group comparisons or an examination of relationships between variables generally done through group-level analyses or multilevel modelling. Temporal analysis is an area of current methodological innovation focusing on the flow of events over time or their relative arrangement. In addition to these methods, computational approaches for structure discovery provide entry points into large online corpora that can identify areas where in-depth examination of the talk is warranted. Social network analysis provides a set of tools for modeling online talk as a collection of interactions or relationships. Topic modeling allows a researcher to identify a set of underlying themes that are present in varying degrees in different segments of online talk. This chapter provides an overview of these different methods and guidance as to which is most appropriate given the intended outcomes of the study and the research design.
