Free Research Calculators and Assessment Tools

They are free, and there is no catch to find. Nothing here asks for an email address, nothing is held back until you book a call, and no result is watermarked or truncated. They run entirely in your browser: what you type into the sample size calculator or paste into the heterogeneity calculator is never sent to us, because there is nowhere for it to be sent. We built them for our own statisticians, editors and medical writers, and the reason they are open is straightforward — a researcher who can check our arithmetic is a better client than one who cannot.

No. Every tool on this page is JavaScript running on your own machine. There is no server call, no analytics event carrying your inputs, no local storage and no file upload anywhere in any of the twelve. You can confirm that by opening your browser’s network tab while you use one, or by saving the page and running it with your connection off. For anything involving patient-level or commercially confidential data that is the only guarantee worth having, and it is the reason these were built this way.

Cite the method, not the tool. Every result panel names the formula or the published source it used — Schoenfeld for the log-rank event requirement, DerSimonian and Laird for τ², Altman and Bland for recovering a confidence interval from a p-value — and that primary reference is what belongs in your methods. If a reviewer asks how a number was produced, the panel gives you enough to reproduce it in R or by hand, which is the standard any calculation in a paper should meet.

The arithmetic is exact to the formula, and we check it against independent implementations before it ships. What limits it is the formula, not the code: these are the standard closed-form answers for two independent groups, and they do not cover cluster randomisation, interim analyses, non-inferiority margins, repeated measures, competing risks or multiple primary outcomes. Each of those changes the number materially, and none of them can be fixed by inflating the answer afterwards. If your design has one of them, the calculator will give you a number that is precisely wrong — use it to scope the conversation, then have the calculation done properly.

The reporting-guideline finder answers that for seventeen designs, and it carries two warnings worth knowing even if you never open it. TRIPOD 2015 has been replaced by TRIPOD+AI, which applies to regression models as well as machine-learning ones, so citing the 2015 statement is now out of date. And CONSORT 2025, which supersedes CONSORT 2010, does not cover trials of AI interventions — those still need the 2020 CONSORT-AI extension, which is written against the older checklist. Both are places where a confident wrong answer would cost an author a revision round.

It checks the criteria that can be checked mechanically — a career window, a host country, a consortium composition — and shows you the published rule beside the verdict so you can see what it is working from. It cannot assess a track record, and it does not know what a specific call has derogated. Three of the rules in it moved during 2026: both ERC career windows widened for the 2027 calls, NIH restructured how international collaborations are funded, and Wellcome’s geographic eligibility narrows on 29 October 2026. We date the tool for that reason. Treat it as a first filter, then read the call text before committing six weeks to writing.

No, and it would be dishonest to imply otherwise. A sample size calculator cannot tell you whether the effect you entered is clinically worth detecting, which is the part a reviewer actually challenges. A test chooser cannot see that your observations are clustered within clinics. The heterogeneity calculator will report I² for five studies that should never have been pooled. What these tools do well is make the size and shape of a problem visible early, while the design can still change — and that is usually worth more than the calculation itself.

Yes. Several of these started as internal calculators for a client project — a screening estimator built to scope a review for a hospital library, a timeline planner built for a research office that kept missing its own deadlines. If your department needs a version with your own rates, your own thresholds, or your institution’s branding and workflow behind it, that is a piece of work we take on. So is the reverse: auditing a calculator you already use, because a spreadsheet that has been passed between three postdocs is a common source of the error nobody can find.

Tell us, with the inputs you used. Every tool states its method precisely so that disagreements are resolvable rather than a matter of opinion, and we would rather hear about an error than have it sit there. If you are right, we fix it and say so on the page. If the disagreement turns out to be about which formula is appropriate for your design — which is the more common outcome — that is a more interesting conversation, and one worth having before the analysis rather than after review.

Yes, and the ones that get built are the ones researchers actually ask for. The twelve here came from questions that kept arriving: how many participants, which test, which checklist, how long will screening take, is this journal real, does the date hold. If there is a calculation you do repeatedly by hand or in a spreadsheet you no longer trust, tell us what it is. It is a short walk from a recurring question to a tool that answers it for everybody.