This working paper illustrates—theoretically and through simulations—that the Rothstein falsification test is not definitive in indicating bias in value-added model estimates of current teacher contributions to student learning.