Stern et al. 2014 — NIH-funded retracted papers carry $392,582 mean attributable cost

Stern, Casadevall, Steen and Fang 2014 measured the mean attributable cost of retracted NIH-funded papers at approximately $392,582 per retraction, with a peer-reviewed median of $239,381, against an aggregate of ~$46.9 million in unadjusted NIH funding across retracted papers from 1992-2012 (S-0050).

The distribution is right-skewed (the median is below the mean), so the median is the conservative estimate for per-paper retraction cost. The methodology attributes funding to retraction by direct cost — not by total grant value — and is therefore conservative.

The figures are critical inputs to two parts of the paper:

1. C-0022 (reproducibility cost): retraction is one mechanism by which non-reproducible work is eventually withdrawn from the literature, and Stern's measurement is the strongest available estimate of what each such withdrawal costs.
2. C-0005 / M-0003 Term A: the sunk-grant value attributable to a single unverifiable paper is anchored on Stern's figures when grant-specific data is unavailable.

The cost is conservative because it captures direct NIH funding only, not downstream policy decisions, follow-on research, regulatory products, or the citation cascade the retracted paper had triggered before withdrawal.