We adapted our analytic strategies based on guidelines for coding open-ended data reliably, as outlined in Hruschka and colleagues (2004). First, by identifying key words and phrases in the responses, an initial master list of theme codes was generated for each domain by going through the entire set of focus group transcripts. Transcripts were used along with written notes to further extrapolate meaning from the responses. Second, a training session was held in which two raters independently coded a small part of each transcript (e.g., one group in each topic area), using the theme code master list, and then compared ratings, discussing any areas of disagreement, and coming to consensus on the specific meanings of various responses. The codes were then modified and refined to reflect newer guidelines established from the training session. Using the newly established codes, raters independently coded the full set of transcripts. An initial computation of inter-rater agreement warranted the necessity for a second session of code modification and refinement.
