2.4. Data analysis plan and reliability
Audio files from group interviews were fully transcribed, even preserving ungrammatical discourse when it occurred.
Interviews were then analyzed for the following themes utilized in the present report: self-presentation (personal identity), self presentation
through relationships (social identity), and gendered self-presentation (gender identity). The first and the last author
separately read through the interviews and assigned a code (or no code if no theme was present) to each turn in the transcript. A
turn was counted each time a new person spoke. A single turn could incorporate multiple themes, and so the codes were not
mutually exclusive. Because of the intrinsic lack of independence of conversational turns and the possibility of a single turn
touching upon more than one theme, we limit ourselves to descriptive statistics in presenting quantitative results.
Coder reliability was calculated for 33% of the data, including one male and one female focus group. The kappa for personal
identity was .88, for social identity .90, and for gender identity .94. All of these kappas are in the excellent range, as defined by
Landis and Koch (1977).
Within each of these codes, important sub-themes were identified using the methodology of discourse analysis. In order to be
reported, these sub-themes had to appear in at least two focus groups and also had to appear in both gender groups (unless it was a
gender-related theme that only applied to one gender). In this way we were able to develop a qualitative typology of the ways in
which the peer interaction of emerging adults may express or influence their identity development. We used this qualitative
methodology because our primary goal was to investigate MySpace phenomenology and the meanings that users constructed for their MySpace experience. A secondary goal was to identify and describe important processes that could guide future quantitative
research.