Toy-model demonstration of a maximum-likelihood estimator (red), a maximum-likelihood estimator with first-order bias correction (green), and an estimator exploiting the full posterior (blue). Data points display the estimator average (y-axis) over 10⁶ data points at varying signal-to-noise levels (x-axis). The true value to be estimated is x = 1. The panels show different signal-to-noise regimes; − nppt denotes y = 1 − n/ 10³.