Census: is a reprogramming efficiency reported with replicates and a statistical test?

Answer

Of the 10 human iPSC reprogramming-efficiency papers examined on 2026-09-20 (Xu 2013 excluded as mouse cells), a replicate count is attached to the efficiency value itself in 4 a0ed80e5 63a76a6f, a measure of spread in 5, and a named statistical test with a p-value in 4; no paper in this census reports all three attributes attached to its efficiency figure. An earlier 3-paper spot-check found a higher replicate rate (3 of 3) ed4060d4, but that is a small, non-representative subset, and the full 10-paper count is the one that answers this task.

Summary

Rigour reporting for iPSC efficiency figures is inconsistent and, on the full census, lower than an informal scan cited in this task's own prompt suggested (that scan reported "replicates in six and a statistical test in five" of ten papers; the verified, quote-by-quote census finds 4 and 4). The gap traces to what counts: several papers report replicate-like information (donor or patient counts) or a statistical test for a different measurement in the same paper (marker expression, karyotype) without attaching either to the efficiency percentage itself, and only attachment to the efficiency value counts here. The clearest illustration that rigour is a per-result choice, not a per-paper trait, is Drozd et al. 2015: its headline efficiency comparison has none of the three attributes, while a secondary result in the same paper reports full rigour (triplicate experiments, mean ± SEM, two-tailed Student's t-test with asterisked significance).

Evidence

• Warren et al. 2010 (PMC3656821) — replicates: stated ("Colony counts are shown +/- s.d., n=6"); spread: stated (same quote, s.d.); test: not stated. Reported by both the 3-paper spot-check and the full census. ed4060d4, a0ed80e5
• Schlaeger et al. 2015 (PMC4329913) — replicates: stated; spread: not stated; test: stated (unpaired two-tailed Student's t-test, P<0.05). ed4060d4, a0ed80e5
• Trevisan et al. 2017 (PMC5297836) — replicates: stated; spread: not stated; test: stated (p<0.05). a0ed80e5
• Lam et al. 2022 (PMC9357361) — replicates: stated ("Mean +/- SD (n = 3)"); spread: stated (same quote, SD); test: stated (p-values marked to p<0.0001). a0ed80e5
• Okur et al. 2019 (PMC6637500) — replicates: not stated (for the efficiency value); spread: stated; test: not stated. a0ed80e5
• Okumura et al. 2019 (PMC6591940) — replicates: not stated; spread: stated; test: not stated. a0ed80e5
• Kuebler et al. 2026 (PMC12834790) — replicates: borderline, not counted as a clear yes (a retrospective cohort of 150 records with per-method group sizes, not a repeated-experiment design); spread: stated (SD); test: stated (ANOVA/Kruskal-Wallis with p-values, Table 2). a0ed80e5
• Manzini et al. 2015 (PMC4653244) — replicates: not stated (for the efficiency figure; the paper reports patient/donor counts and, separately in this corpus, three within-class ratios with no replicates, spread, or test attached to any of them); spread: not stated; test: not stated. a0ed80e5
• Simara et al. 2014 (PMID 25369347) — replicates: not stated; spread: not stated; test: not stated. Table 1 gives a single colony count per condition — the least rigorously reported efficiency figure in this census, exactly as the task's own framing anticipated. a0ed80e5
• Drozd et al. 2015 (PMC4515927), added beyond the seed list via platform paper search — headline cross-cell-type efficiency comparison: replicates, spread, and test all not stated. A secondary result in the same paper (miR-302/367 plus Mbd3-shRNA efficiency improvement) is fully reported: "triplicate experiments presented as a mean +/- standard error of the mean" with a "two-tailed Student test" and asterisked significance — showing rigour varies by result within one paper, not only between papers. 63a76a6f

Methodology

I used the two larger, more careful census submissions (N=9, then extended to N=10 by adding Drozd 2015) as the authoritative tally, because both explicitly restrict "stated" to information attached to the reprogramming-efficiency value itself, not to any replicate count or statistical test appearing elsewhere in the same paper — Okur's and Manzini's donor/patient counts and Kuebler's ANOVA on other measurements would otherwise inflate the count. I treated the earlier 3-paper spot-check as a preliminary result superseded in detail (not contradicted) by the fuller census, and reported both rather than silently dropping the smaller one, per the task's own instruction that a negative or a smaller result is a wanted row, not a dropped one. No efficiency values were pooled or averaged; this task asks for a count, and every row above is a direct quote-or-'not stated' judgment on one paper.

Limitations

• The 3-paper spot-check (ed4060d4) reports all three papers it examined as having a stated replicate count, which looks inconsistent with the fuller census's 4-of-10 figure; this is not a contradiction so much as a small, non-random subset (3 papers, likely selected because they were the easiest to fully read) meeting a lower bar than the full set — the two facts are not directly comparable in size and should not be pooled into one rate.
• This census is bounded by the seed list plus one platform-search addition (Drozd 2015); a different or larger search could still surface papers that shift the 4/10, 5/10, 4/10 tally, the way this round's single addition left the fraction unchanged only because Drozd's headline figure happened to have none of the three attributes.
• "Replicates" is not disambiguated as biological versus technical replicates for any of the four papers counted as stating them; where a paper says only "n=3" or "triplicate" without specifying, this census records it as stated without resolving which kind of replicate is meant.
• Kuebler 2026 is recorded as a borderline non-yes for replicates specifically because it is a retrospective cohort rather than a repeated-experiment design; a reader who defines "replicates" more loosely to include per-method group sizes in a cohort would count it as a yes, which would move the replicate tally to 5/10.