Unless the decentralized system of governance over American education changes, we assume that there will always be different evaluation methods that rely on different data with results intended for different audiences. No single method or mechanism is likely to completely satisfy multiple, legitimate, and potentially incompatible demands for valid, reliable, and usable information.

The word “validity” is shorthand for the extent to which evaluation data support specific inferences about individual or organizational performance. In this report we define validity broadly to include (1) the quality of evidence and theory that supports the interpretation of evaluation results; and (2) the consequences of the use of evaluations for individuals, organizations, or the general public.

The fact that there are imperfections in evaluation systems is not a sufficient reason to stop evaluating, but rather an argument for investing in the development of improved methods. But trying to find the “perfect” set of measures is a fool’s errand; a more rational approach is to explore the relative benefits and costs of alternative approaches and to consider whether, on balance, there is evidence that the benefits outweigh the costs—not just the costs in dollars and cents but the costs defined more generally in terms of unintended negative consequences.

It is especially important to consider potential inequities that may arise from the interpretation of evaluation results and their application. Special attention should be paid to unintended impacts on the morale, capacity, and reputation of TPPs that cater to different pools of potential teacher candidates or that are intended to serve communities struggling to overcome socioeconomic disadvantage. We are not suggesting differential standards for the evaluation of program quality, but rather we are flagging the importance of studying how those standards may, for example, lead to the reduction in the supply of prospective future teachers and/or an interruption in the flow of potentially excellent and dedicated teachers into poor neighborhoods.

As we prepare this report, expectations for teaching and standards of student learning (and other valued outcomes) are again changing. The implementation of the Common Core State Standards, for example, and associated new assessment technologies will necessarily shape the context for evaluating TPPs. The need for flexibility in adapting to these types of changes must be balanced against the legitimate desire for evaluations designed to provide reliable trend information. Achieving a workable balance requires an appreciation of tradeoffs and an acceptance of compromises in designing systems with diverse purposes.