STRATA
Skip to study
XLVII

SOURCE RECORD / 2015

What survives a replication?

DocumentedR01 · 2026-11-18
Scope of this record

KNOWN refers to the published report in this selected corpus. It is not an estimate for all scientific fields or a judgment about individual researchers.

Statement

The Open Science Collaboration reports replications of 100 studies from three psychology journals, with replication effects smaller on average and success varying by the criterion used. [source]

Equation

Unrecorded

The connection is scoped to this source. No broader cross-domain equality is established by this entry.

Observation

The abstract distinguishes statistical significance, effect-size intervals and subjective replication assessments. These are different evaluations, not interchangeable labels for a single outcome. [source]

Progressive questions

Which outcome and analysis were specified before replication? Are the original conditions accessible? How should an interval shift, a smaller effect or a design difference change the exact claim rather than the reputation of its author?

Source response

The reported replications used original materials when available. The abstract describes weaker effects and differing replication measures; detailed cause attribution requires inspection beyond this seed. [source]

Interpretation

XLVII reading: replication is another linked record with its own inputs, scope and uncertainty. A disagreement should lead to inspectable comparison, not silent deletion of the earlier source.

Conclusion

This source makes reproducibility an empirical object of study. Its sample and evaluation criteria bound the result. [source]

Next experiment

Proposed: select one lawful open dataset and reproduce its published analysis before choosing a changed condition. Record versions, exclusion rules and estimates with uncertainty. Failure to reproduce the reported calculation is distinct from failure of the original scientific hypothesis.

Work through a related example.

V.08 / Replication, reproducibility, and preregistration

Separate three kinds of repeat.

Take a published plot and ask first whether its shared data and code regenerate the figure. Next ask whether new data under the same protocol support a comparable estimate. Finally change one condition to test the proposed mechanism. Record a separate outcome for each task so that one successful repeat does not stand in for all three.

X.12 / Data, code, licenses, and executable research records

Package one figure as a reproducible record.

Take a small plot and include its input file, transformation script, parameter values, dependency versions, and expected output. Add a checksum for the input and a short explanation of the axes. Run it in a clean environment when feasible, recording the actual result separately from the instructions and expected behavior.

Follow earlier and later work.

Before

No linked record.

After

  • Proposed connection · UNKNOWNCompute has a budget

    Editorial research connection to investigate; not a verified correspondence, causal lineage, or mathematical equivalence.

  • Proposed connection · UNKNOWNCompression, reopened

    Editorial research connection to investigate; not a verified correspondence, causal lineage, or mathematical equivalence.

Follow the chain
  1. Compute has a budget · 1 link away · UNKNOWN
  2. Compression, reopened · 1 link away · UNKNOWN
  3. Attention without recurrence · 2 links away · UNKNOWN
  4. Compression → classification · 2 links away · UNKNOWN
  5. Prediction and residual error · 3 links away · UNKNOWN
  6. Dynamic Markov modelling · 3 links away · KNOWN
  7. Information is not meaning · 3 links away · UNKNOWN
  8. Memory as a collective state · 4 links away · UNKNOWN
  9. The chemical compass · 4 links away · UNKNOWN
  10. Erasure has conditions · 4 links away · UNKNOWN
  11. Reversibility and the record · 5 links away · UNKNOWN

Placement.

Technology

This entry records the cited source and a proposed next investigation. It makes no claim of historical abandonment, suppression, or independent reproduction.

Sources.

  1. Estimating the reproducibility of psychological science

    Open Science Collaboration / Science · 349(6251), aac4716; DOI 10.1126/science.aac4716; abstract.

    Publication metadata and abstract inspected; original datasets and methods not reanalyzed. Checked 2026-11-18.

Record history
  1. 2026-11-18 / R1 / SOURCE_SEEDED

    Source-limited editorial seed with explicit inspection scope. Research questions and proposed experiments are not executed results.