Three approaches to in-service development for teachers (action research, skills development, and materials dissemination) were compared using a multimethod evaluation design with innovation-specific and general outcome measures for students and teachers. Thirty-three teachers and their students participated in a project to teach students how to evaluate their work. There were no treatment differences on a self-reported use of evaluation procedures, personal teaching efficacy, or in general student outcomes (goal orientations, attributions for success and failure, and self-efficacy). There were two small but statistically significant differences favoring action research: (1) teachers in the action research condition scored higher on outcome expectancy because they had greater access to teachers who had successfully used student self-evaluation to increase student achievement and motivation; and (2) students in the action research condition scored higher on attitudes toward evaluation because their teachers had a better understanding of how to share control of evaluation, a core teacher function. The modest differences were attributable to the short duration of the treatments and to the neglect of student cognitions about self-evaluation. (Contains 9 tables and 72 references.) (Author/SLD)