Unlike in elementary school, high-school teacher effects may be confounded with both selection to tracks and unobserved track-level treatments. I document sizable confounding track effects, and show that traditional tests for the existence of teacher effects are likely biased. After accounting...