Teacher work is multidimensional, encompassing activity on a variety of instructional and non-instructional dimensions. An important concern in the debate over the use of student test score-based measures of teacher effectiveness in teacher evaluation, compensation, and accountability systems is that they emphasize only a limited range of contributions that teachers make to their schools, namely those related to increasing students’ basic skills in tested subjects. Value-added models (VAMs) or other student test-based measures are unlikely to identify teacher effectiveness in performance areas not strongly correlated with effectiveness at raising student standardized test score performance. For example, if teachers who are particularly adept at increasing student motivation or interpersonal skills but those outcomes are not closely related to student standardized test score growth, an evaluation system relying on VAM scores to evaluate teachers could label a teacher as low-performing despite important contributions to student development. This labeling is potentially a problem if the missed contributions are towards goals valued by schools (see for example Jackson, 2013, on the predictive value of non-cognitive skill development for long-term outcomes).

Motivated by these concerns and the growing empirical literature on the relationships among different measures of teacher performance, this paper asks three primary research questions:

1. Are teachers’ outcomes in different areas of teacher practice (instructional and non-instructional) positively correlated across dimensions, or is there evidence of tradeoffs among some kinds of outcomes?
2. How are the different dimensions of effectiveness predicted by teachers’ value-added in math and reading?
3. Which dimensions of teacher effectiveness do schools value, as measured by the degree to which they predict principals’ interest in retaining that teacher in the school?

Our analysis draws on a unique source of data: interviews with approximately 100 principals in Miami-Dade County Public Schools, the nation’s fourth-largest school district, in which they were asked to provide low-stakes assessments of their teachers’ performance on 8 different dimensions. We merge these assessments with administrative data on teachers and students to estimate value-added scores for each teacher. We then investigate the degree not only to which teacher performance dimensions correlate with one another but the degree to which each is correlated with measures of teacher value-added in math and reading.

Although we find that the 8 performance dimensions generally are positively correlated with one another, not all are significantly associated with teachers’ value-added scores, suggesting that test-based teacher performance measures fail to capture important contributions teachers’ make to their schools. Teachers’ contributions outside of classroom instruction (e.g., contributing to school leadership, building ties to the community) are less likely to be correlated with value-added. Yet analysis of principals’ responses about who they would like to keep in their schools suggest that they see both in- and out-of-classroom dimensions as valuable. Analysis (ongoing) of qualitative interview responses provided by principals helps illuminate these central results. Implications for the design of teacher evaluation systems are discussed.