Questions, answered.

No. You describe your study in plain language — your problem statement and objectives — and upload your dataset. thericerca infers the design, chooses the appropriate test for each objective, and explains every result in plain English alongside its formal statement. Knowing statistics helps you interpret the findings, but it isn’t required to get a correct, defensible analysis.

Tabular data as CSV, TSV, Excel (.xlsx/.xls), or SPSS (.sav) files. On upload, thericerca profiles each column — its type, distribution, and missingness — and uses that profile to recommend and run analyses. Your original file is stored immutably and content-addressed, so the exact bytes you uploaded are the exact bytes analysed.

Descriptive statistics and frequencies; correlation (Pearson, Spearman, point-biserial); group comparisons (t-tests, one-way ANOVA with gated post-hoc comparisons, and their non-parametric counterparts — Mann–Whitney, Kruskal–Wallis); association (chi-square, Fisher’s exact); linear, multiple, logistic, ordinal, and Poisson regression; reliability (Cronbach’s alpha); and exploratory factor analysis. It picks among them based on your variables and objectives.

A complete report — Statistical Tests Used, Results in plain language beside each formal statement, a detailed Discussion with Conclusions and Recommendations, and an appendix of the exact Python that produced every figure. You can download it as HTML, Word (DOCX), or PDF.

No — and this is the point of the whole design. Every statistic is computed by real scientific libraries (scipy, statsmodels) in an isolated sandbox, never generated by a language model. Before a report is issued, a numeric-claim validator traces each number in the prose back to the computed results; anything it can’t verify flags the report rather than shipping it.

A deterministic rules engine reads the measurement level and role of each variable and your objective, then selects the appropriate test — a chi-square for two categoricals, a correlation for two continuous measures, ANOVA for a continuous outcome across three or more groups, and so on. It also checks each test’s assumptions and falls back when they fail (for example, Pearson to Spearman on non-normal data), noting the reason in the report.

Optionally. A language model can help map your plain-English objectives to a statistical plan and phrase the write-up. It never computes a statistic. thericerca runs its full deterministic path with or without AI assistance — the model only enriches the wording when AI assistance is enabled.

It’s a rigorous instrument, not a replacement for judgment. thericerca automates choosing and running the correct test with fidelity and hands you a defensible report — but the interpretation of findings in your specific context still benefits from domain and statistical expertise. It is not professional, legal, or clinical advice.

The feasibility assessment — the full plan, the tests it will run, and the cost — is free, before you spend anything. Running the analysis draws from credits, which you buy in packs. You only pay to run, and you always see the estimate first.

No. You can create an account, upload data, and see the complete feasibility plan and its cost without entering a card. Payment happens only when you choose to run a paid analysis.

Passwords are hashed with bcrypt and checked against a strength policy; sessions use short-lived access tokens with rotating, single-use refresh tokens that are revoked on sign-out. New accounts verify their email, and sensitive endpoints are rate-limited. See the Privacy & Cookies notice for how data is stored.

Your datasets and reports are private to your account. They are used to run the analyses you request and to generate your reports — nothing else. We don’t sell your data or use it for advertising. The full detail is in the Privacy & Cookies notice, and you can manage your privacy choices at any time.

Only if AI assistance is enabled, and even then only the study description and computed results needed to phrase the write-up — never used to compute statistics. When AI assistance is off, nothing leaves for a model at all. You control this in your privacy choices.

Yes. Every report downloads as HTML, DOCX, or PDF, and you can request access to or deletion of your account data. Start from Your privacy choices or email [email protected].