Evan Carter

Who they are

Carter and McCullough did not run a new experiment. They reanalysed one that already existed, which is what makes the paper useful as an illustration of how a literature can be wrong without any individual study being wrong. The meta-analysis led by Martin Hagger in 2010 had pooled close to 200 experiments and reported a medium-to-large effect. Carter and McCullough applied bias-correction techniques to the same body of results and asked a different question: given the sample sizes of the studies in this literature, is the distribution of reported effects the distribution a real effect of that size would produce?

It was not. Small studies reporting large effects were far too numerous, which is the signature of a literature filtered by what gets submitted and published rather than by what was found. Once the correction was applied, the true effect could be zero.

That conclusion was contested when it was published, and it was ahead of the direct evidence. The direct evidence arrived afterwards and pointed the same way. Hagger's 23-laboratory preregistered replication returned d = 0.04 in 2016, and the 36-laboratory test led by Kathleen Vohs returned d = 0.06 in 2021. A 12-laboratory replication led by Dang and colleagues in the same year did find a small significant effect at d = 0.10, so the range is not flat zero. It is nowhere near 0.62, and Carter and McCullough said so first, from the existing data, before anyone had collected any new.

There is a citation problem attached to his name that is worth stating plainly, because it runs in the opposite direction to the usual one. Nir Eyal's Indistractable credits Carter with a 2010 meta-analysis of nearly 200 experiments. Carter did not run that meta-analysis. Hagger, Wood, Stiff and Chatzisarantis did, and it reported a medium-to-large effect. Carter and McCullough's contribution is the 2014 to 2015 bias-correction reanalysis of that dataset, and it reaches the opposite conclusion. Two papers, opposite findings, one credited to the wrong author. Anything published on this site uses the correct attribution.

Key papers

The site sources one paper of his directly. No claim is made here about the rest of his output, which the vault does not cover.

Carter, E. C., & McCullough, M. E. (2014), "Publication bias and the limited strength model of self-control: Has the evidence for ego depletion been overestimated?," Frontiers in Psychology, 5:823. Reanalysed the existing ego-depletion meta-analytic dataset for small-study effects, found very strong signals of publication bias, and concluded that the bias-corrected effect could be zero.

Key ideas

A meta-analysis can be arithmetically correct and still wrong, because it inherits whatever filter decided which studies got published.

Small-study effects, many underpowered studies reporting large results, are detectable in a literature and can be corrected for.

The corrected estimate for ego depletion could not be distinguished from zero, two years before any large preregistered replication had been run.

Bias correction and direct replication are independent instruments. On this question they agreed, which is the strongest form the argument can take.

The 2010 meta-analysis is Hagger and colleagues, not Carter. The misattribution is common and reverses what his paper actually claimed.

Concepts on this site

Ego Depletion

The Replication Crisis

Find them elsewhere

Only a Google Scholar profile could be confirmed, verified at mail.mil for the US Army Research Laboratory, with ego depletion and meta-analysis among its listed interests. No personal or institutional page and no social accounts were found

Checked 2026-09-10. Handles and affiliations go out of date, so treat this block as a snapshot rather than a standing fact.

---