Question 1

Is Clarus a large language model or a chatbot?

Accepted Answer

No. Clarus is a deterministic rule engine. It computes statistics with real numerical libraries (scipy / statsmodels) and the same data always yields the same verdict; it never generates numbers, citations or results, and states unknowns as unknown.

Question 2

Which test does Clarus pick for my data?

Accepted Answer

Clarus infers each variable's role and measurement level from your uploaded dataset, then recommends a fitting method from its library and shows the reasons it matched — checking the method's assumptions before running it and rerouting when they fail.

Question 3

Can Clarus model count outcomes or rates?

Accepted Answer

No — Poisson / negative-binomial regression are not in the library.

Question 4

Can Clarus compute an odds ratio for case-control data?

Accepted Answer

Yes — the odds ratio (2x2) reports the cross-product odds ratio with a Woolf log-OR 95% confidence interval, the valid association measure for case-control designs. Clarus also computes the relative risk (cohort) and the absolute risk difference / number needed to treat.

Question 5

Does Clarus do survival or time-to-event analysis?

Accepted Answer

Kaplan-Meier, log-rank, Cox PH — entirely absent; no censoring role anywhere.

Question 6

Can Clarus fit mixed-effects models for clustered or repeated data?

Accepted Answer

Linear/generalized MIXED-EFFECTS models — MLR & logistic tie_breakers say 'use a mixed-effects model' but none exists.

Question 7

Does Clarus do multivariate methods like PCA, factor analysis or SEM?

Accepted Answer

PCA, EFA/CFA, SEM, MANOVA (multiple outcomes), cluster analysis, discriminant analysis.

The Pro Plan is coming soon.

Unlock more