You spent a year rerunning someone else's study, with a bigger sample and an analysis plan fixed before the first participant arrived. The effect came out smaller than the original, and now you have to share the paper in a field where the original authors are probably in your network.
A replication is one of the most useful papers a researcher can write, and one of the easiest to promote badly. Framed one way, it reads like a takedown. Framed another, it reads like a footnote. Neither is fair to the work.
Say Which Kind of Check You Ran
Start by naming what you actually did, because readers will not agree on the words. The National Academies report on reproducibility and replicability notes that disciplines use these terms in different and sometimes contradictory ways, and it adopts definitions meant to apply across fields.
| Term | What it means in the National Academies report | Plain phrasing for a post |
|---|---|---|
| Reproducibility | Consistent computational results using the same input data, computational steps, methods, code and conditions of analysis | "We reran the original analysis on the original data." |
| Replicability | Consistent results across studies aimed at the same scientific question, each of which has obtained its own data | "We ran the study again with new participants." |
| Generalizability | The extent to which results apply in contexts or populations that differ from the original one | "We tested whether it holds in a different population." |
A post that says "we tried to reproduce" when you collected new data invites the wrong argument. One clause of plain description up front saves a long thread later.
Know the Backdrop Your Readers Bring
Many readers will meet your paper with large replication projects already in mind. In psychology, the Open Science Collaboration replicated 100 experimental and correlational studies from three journals. Ninety-seven percent of the original studies had statistically significant results, against 36% of the replications, and replication effects were half the magnitude of the original effects.
In the social sciences more broadly, Camerer and colleagues replicated 21 experiments published in Nature and Science between 2010 and 2015. They found a significant effect in the same direction for 13 (62%) of them, and the replication effect sizes averaged about 50% of the originals.
Those numbers are context, not a verdict on your study. The National Academies report is explicit that non-replicability occurs for a number of reasons that do not necessarily reflect that something is wrong, and that working out which reasons apply takes further investigation. Your post can say that a smaller effect is a common outcome without implying anything about the original team.
Report a Degree, Not a Verdict
The same report suggests that assessing replicability may not give a binary pass or fail answer, and that the answer may best be expressed as the degree to which one result replicates another. That is the single most useful framing for your posts and summaries.
The Open Science Collaboration's own abstract shows why. It reports, side by side, that 36% of replications had statistically significant results, 47% of original effect sizes were in the 95% confidence interval of the replication effect size, and 39% of effects were subjectively rated to have replicated the original result. The headline rate depends on the criterion, so name yours.
In practice that means putting the original effect size and yours side by side, each with its interval, and saying in a sentence which criterion you set in advance. Readers can then judge the size of the gap themselves.
A Worked Example
Say, as a hypothetical, that you replicated a published educational psychology study reporting that a short classroom writing exercise improved students' test scores. Your registered analysis found an effect in the same direction, much smaller than the original, with an interval that includes effects close to zero.
- Takedown: "Classroom writing effect fails to replicate." It turns a degree into a verdict and makes the original authors the story.
- Footnote: "We conducted a replication of an earlier study. Results were mixed." True, and nobody reads past it.
- Fair and readable: "We reran a published study of a short classroom writing exercise with a much larger sample and an analysis plan registered in advance. The effect pointed the same way but was far smaller, and our interval includes effects close to zero. Here is what that means for anyone using the exercise in a classroom."
The third version names the check, gives the direction and the size, admits the uncertainty and ends with why a reader should care. If your result was null rather than smaller, the guide to communicating null results covers the wording in more depth.
Bring the Original Authors In
The large projects did not treat original authors as opponents. In the Camerer study, the replications followed analysis plans reviewed by the original authors and pre-registered before the replications were run. The Association for Psychological Science describes its Registered Replication Reports as multi-lab replications of important experiments published along with comments by the authors of the original studies.
You can borrow the same courtesy when you promote your paper. Send the original authors a short note and a link before you post, name and link their paper in everything you share, and point readers to their response if they write one. Disagreement about the science is fine; surprise is what makes it personal.
It also helps to reuse the case you made to the journal. At the time of writing (November 2026), PLOS One asks authors of a study that replicates previous work to provide a sound scientific rationale and to discuss the existing literature. That rationale already explains why the question deserved a second look, and it makes a better opening line than the result.
A Checklist Before You Share
- Name and link the original paper.
- Say which check you ran: reproduction, replication or a test of generalizability.
- Give the original and replication effect sizes side by side, each with its interval.
- State the criterion you used to judge replication, and whether it was set in advance.
- Describe what differed (sample, setting, materials) without claiming it explains the gap unless you tested it.
- Tell the original authors before you post, and link their response if there is one.
- Link your data, code and registered plan; if the replication was a Registered Report, the guide to promoting a Registered Report covers the two announcement moments.
Where Amplify Fits
A replication lives or dies on careful wording, and most of that wording is already in your paper. Amplify generates a video abstract, a vertical short, an audio brief and a one-page infographic from the PDF you upload and nothing else, with no open-web filler. Every artifact carries the title, the authors and a link back to the paper of record.
Each format arrives as three variants you can edit line by line or regenerate, so if a draft slides from "smaller effect" into "failed to replicate", you change the line. A subject-matter reviewer reads every draft before it can leave your workspace, every claim is traced to a page in the PDF, and nothing publishes until you approve it.
The embeddable widget puts every format beside the paper on your lab site or faculty profile, where readers can read, listen and chat with the paper in multiple languages. The analytics show reads, listens, questions asked and which format moved people, which helps when you later report on the work.
Replications make the record more trustworthy, but only if people read them as carefully as they were run. Sharing yours with the same care is part of the job.
