Saving the Best for Last: Rating Elevation in Low-Performing or Novice Teachers’ Final Formal Observations

Abstract

Evaluators “save the best for last” by systematically elevating the final year-end observation scores for novices or low-performing teachers. Examining 5,251 K–12 teachers, we document a striking reversal: evaluator scores typically fell below teacher self-assessments across initial and midyear observations but exceeded them by approximately 0.25 SD in final year-end observations. This pattern was robust to unobserved differences across annual teacher-evaluator pairings, the instructional domains observed, and observation timing. The data suggest that evaluation systems are not implemented uniformly across performance tiers, revealing tensions between accountability and developmental support that merit explicit acknowledgment in evaluation framework design.