she spent engaging in law reform and service. These estimates were cuffed easily from interview transcripts and aggregated by program. To measure the mobilization process, lawyers were asked how they informed themselves regarding poverty community problems; their responses were extracted from interview transcripts, and the information aggregated by program. The validity of these data was checked by examining summaries of law reform activity throughout the country reported in the journal, Clearinghouse Review, and interview responses of individuals outside the program.
The three sets of explanatory variables were measured with a mix of quantitative and qualitative data. Some of the quantitative data, such as the age of lawyers and the number of years of legal experience, were obtained from the questionnaire. Other data, such as career ambitions of staff, were extracted from interview transcripts. All of the individual-level information was coded on a single sheet for each lawyer and then aggregated by program.
Devising a methodology to analyze the qualitative data and compare across the five programs proved to be quite challenging. Qualitative interview data were sorted by topic, taped to notecards, and filed by topic. This was accomplished by physically reducing pages of the transcribed interviews so that one page could be taped to a 5" × 8" notecard. The reduced interviews were read, cut, taped to notecards, and filed by topic. The filing topics corresponded to the three sets of explanatory variables (e.g., relations with external groups, political attitudes, organizational policies) and to areas discussed frequently by respondents (e.g., working conditions, complaints about management personnel).
A variety of intensive techniques were employed to examine interview responses. First, I read all of the interviews from a particular community in one sitting. Then, I examined all of the interviews across programs for a particular role, and finally, I read all of the notecards in each substantive category within and then among programs at one sitting. These techniques served to uncover patterns of activity within and among the five programs and generated hypotheses that may be tested more rigorously in the future.