Although temporality is a key characteristic of the core concepts of CSCL--interaction, communication, learning, knowledge building, technology use--and although CSCL researchers have privileged access to process data, the theoretical constructs and methods employed in research practice frequently neglect to make full use of information relating to time and order. This is particularly problematic when collaboration and learning processes are studied in groups that work together over weeks, and months, as is often the case. The quantitative method dominant in the social and learning sciences--variable-centred variance theory--is of limited value for studying change on longer time scales. We introduce the event-centred view of process as a more generally applicable approach, not only for quantitative analysis, but also for providing closer links between qualitative and quantitative research methods. A number of methods for variable- and event-centred analysis of process data are described and compared, using examples from CSCL research. I conclude with suggestions on how experimental, descriptive, and design-oriented research orientations can become better integrated.