Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
This paper proposes a new estimator for least squares model averaging. We propose computing the model weights by minimizing a prediction model averaging (PMA) criterion. We prove that the PMA estimator is asymptotically optimal in the sense of achieving the lowest possible mean squared error. In simulation experiments the PMA estimator is shown to have good finite sample performance. As an empirical illustration, we demonstrate that using PMA to account for model uncertainty can lead to large gains in box office prediction accuracy.