Each stage feeds the next. Your study design fills your proposal; your proposal configures your analysis; your analysis is verified before it ever reaches your panel. Nothing gets retyped, and nothing gets assumed.
Describe your objectives and get a real design back — the recommended study type with rationale, sample size from power analysis at three tiers (.80 / .85 / .90, Cohen 1988), the statistical tests that design implies, ethics requirements flagged 🔴 required vs 🟡 best practice, and an editable month-by-month timeline.
Stop hunting for a scale you can legally use. Choose from 24 validated instruments — PHQ-9, GAD-7, PSS-10, SWLS, UCLA-3, Rosenberg, NEP-15, BAS-16 and more — then curate the items down, or upload your own. Each ships with a scoring key and the exact CSV column names your analysis will expect.
Build a full proposal from scratch — pre-filled from your Study Design, section by section in PH HEI format: background, problem statement, framework, RRL, methodology, timeline, ethics, references. Exports as a real .docx with proper heading structure and a hanging-indent reference list.
Upload the proposal you just generated and the pipeline configures itself — study type, variables, and instruments carry straight over. Then map your columns, and the engine picks the test from your data's actual shape (two groups? paired? numeric or categorical?) the way a statistician reads a decision tree. 13 study types — hypothesis testing, quantitative, qualitative, descriptive, cohort, case-control, cross-sectional, case series, quasi-experimental, RCT, psychometrics, diagnostic accuracy, systematic review — plus 9 specialized methods: Kaplan-Meier survival, Cox regression, MANOVA, factor analysis, k-means clustering, Mann-Kendall trend, SEM, Delphi consensus, and Cronbach's α.
Larry recomputes every reported statistic a second time from your raw data and flags any mismatch before you submit. He also checks every citation against your confirmed library, explains any number in plain language, and answers how the product itself works — all in one conversation.
Search Semantic Scholar and CrossRef for real, citation-ready sources with CRAAP-score filtering — finding and verifying sources is always free and unlimited. Upload papers and interrogate them directly to ground your framework and RRL.
The finished paper, its result charts, and the math explainer each export as real .docx files — charts as embedded images, references as a genuine hanging-indent list. Seven citation styles: APA, MLA, Chicago, Harvard, AMA, CSE, Vancouver.
Academic integrity is an engineering problem here, not a disclaimer. Five design decisions make the difference — each one deliberately taken out of the language model's hands.
Every statistic, p-value, effect size and confidence interval comes from a dedicated Python compute engine running against your actual data. The model only writes prose around numbers the engine already returned.
Which test runs is decided by the real shape of your data — the same decision tree a human statistician follows. Never a model's guess at what sounds plausible.
Larry recomputes every reported statistic a second time from raw data and flags any mismatch — a built-in check against silent drift between what was computed and what got written.
Before a study-type recommendation reaches you, the system verifies the model's stated reasoning actually matches the type it picked. A model that explains one design but selects another is rejected, not trusted.
If your data or design doesn't cleanly fit a supported test, the pipeline says so and tells you what's missing — instead of forcing a plausible-looking but wrong analysis through.
The Advisor Validation Gate marks design output as recommendatory only. Effect sizes and instrument choices must be grounded in literature your advisor can verify — the pipeline never pretends otherwise.
Study Design, the Instrument Library, the Proposal Writer, all 13 study types, and Larry's verification gate are included everywhere. Higher tiers buy headroom, never features.
Quotas reset each billing cycle.
A solid starting point for students building their first real study.
Quotas reset each billing cycle.
Built for thesis writers and researchers actively mid-study.
Quotas reset each billing cycle.
For faculty, PhDs, and institutional researchers producing at volume.