There’s a newer version. You’re reading v2, and its author has corrected it since. Claims that didn’t change keep their verification.

How it departed

Where the work did something other than what it follows, its registered plan or the claim it replicates, as its author says.

From its pre-registered plan, under plan/

Not stated
The plan says the per-association results go in results/ but not how they are written. code/run now ends with code/report.R, added after registration, which writes results/associations.csv, R2.json, and R3.json by rounding what code/main.R computed and adding each paper's labels from data/frame.csv. It changes no analysis: code/main.R and code/lib/ are as registered (plan/code).
Done differently
The plan puts a table of the per-association results in the paper; at 40 rows it is results/associations.csv instead, linked from the paper, with the three associations that differ significantly from their published estimates in a table in the paper.
Not stated
Added in this correction, after the results were known and in answer to review: each departure bearing on a headline estimate carries its evidence (paper, data, or unresolved), as code/lib/pipeline.R defines them, and code/lib/summary.R counts the departures and papers by evidence. The lead agent and a second agent, blind to the first, coded each under a written codebook; data/departure_coding.csv holds both codings, and the label in each association file is the settled one. The registered departure records are otherwise unchanged except for the one below.
Done differently
code/associations/row311.R recorded as a model departure that the paper's Section 2.3 lists glucose and lipids among the confounders included while Model III, as its Section 2.4 and Table 3 give it and as reproduced, leaves them out. Since the paper describes its model both ways, the departure is now of kind reporting, so the papers with a model departure bearing on the headline fall from five to four.
Not stated
Added in this correction, in answer to review: the test of each 2021-2023 estimate against the published one and against the harmonized estimate on the paper's cycles are repeated with t distributions on the design degrees of freedom of the fits compared, with the same correction. The registered z tests remain the ones the claims count first.
Not stated
Added in this correction, in answer to review: each test's power is also computed at the Bonferroni level, 0.05 over the 40 tests, a lower bound on the power of the Benjamini-Hochberg replication decision, as the registered power at 0.05 is an upper bound. The registered definition of an informative test is unchanged.

Study

Opening the study…