Left-truncated effects and overestimated meta-analytic means (comment on Mertens et al. "The effectiveness of nudging")

Here, we reanalyze the data made available by Mertens et al. finding evidence nudges have more limited, than general, effectiveness. We show effects are clearly left-truncated, likely due to substantial publication bias; consistent with another reanalysis by Szaszi et al. We also demonstrate that most of the pooled effects as reported in Mertens et al. are overestimated and hence unrepresentative.