Beginning teachers' experiences in university-based teacher preparation are rapidly replaced by the norms and expectations of the school setting and teachers with whom they work. To interrupt this pattern by which new teachers are pulled from the practices they have learned at the university, teachers must be encouraged to reflect about their teaching through action research, a process by which a teacher studies his or her own actions in the classroom as a means to improve practice. To explore the possibility of collaborative action research--action research done in teams of at least two--a study examined 35 apprentice English teachers and 22 mentor teachers. Data was collected from journals of both apprentices and mentors, and in the form of observation notes by university faculty, self-assessment, and interviews. Each of the studies began with a guiding question generated by the research group and directly related to instruction such as, "Are guided essays more effective than nonguided?" or "Will creative role playing increase comprehension?" Findings showed five recurrent themes concerning teacher attitudes toward data collected: (1) research questions based on attempts to improve teaching; (2) fear about conducting research; (3) adverse to using their students as potential pawns; (4) confusion over qualitative data collection and analysis; and (5) lack of time and commitment to formal reporting of results. Results revealed two major effects on participants substantiated by triangulation of the data: sustained interaction with peers increased, and the level and quantity of reflection on practice increased. (A table lists action research projects.) (TB)