Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
In the literature, it is a common empirical finding that survey based expectations are biased at the individual level. This has sparked a large debate if forecasters have asymmetric loss functions or the rationality assumption is violated. In this paper, I will show that the bias can in large part be explained by the pattern of missing observations in the survey. Thus the assumption of asymmetric loss functions is not required to satisfy the rationality assumption.