Methodology & accuracy
StatSolver is a browser-based educational tool for learning common discrete probability models. This page explains what the calculators do, how they are checked, and where their scope ends.
Written and maintained by Mustafa Akilli · Updated September 9, 2026
What the calculators cover
Bernoulli and Binomial
These calculators use the standard binary-trial models. Inputs are validated as probabilities between 0 and 1; the Binomial calculator also validates the fixed trial count and computes the probability for each possible success count.
Geometric
StatSolver uses the trial-number convention: X is the trial on which the first success occurs, beginning at 1. The page states this convention because some textbooks instead count failures before the first success.
Poisson
The Poisson calculator treats λ as the expected event count in the specified interval and validates it as a positive finite value. The guide explains the constant-rate and independence assumptions.
Joint distribution
The joint calculator validates a probability table, computes marginals and moments, checks independence across the table, and derives the distribution of X + Y. It also explains when conditional probabilities are undefined because a denominator is zero.
How correctness is checked
- Regression cases cover ordinary examples, symmetry, dependence, zero variance, negative values, fractional inputs, and large or small probabilities.
- Input hardening checks reject malformed numbers, HTML payloads, division by zero, non-finite values, invalid probabilities, and tampered local history.
- Static site checks verify metadata, internal routes, publisher declarations, `ads.txt`, sitemap entries, and stale legacy branding.
- Educational examples are calculated independently before being published and link to reference material when a formal definition is useful.
Scope and limitations
The calculators implement the formulas described on their pages; they do not infer whether a real-world dataset satisfies an assumption. Independence, constant rates, representative sampling, and measurement quality must be assessed by the person using the model.
Results are provided for education and general information. Verify important academic, scientific, engineering, financial, or professional work with an appropriate source or qualified reviewer.
References
- NIST/SEMATECH e-Handbook of Statistical Methods: binomial distribution
- NIST/SEMATECH e-Handbook of Statistical Methods: Poisson models
- StatSolver Learning Center: choosing a distribution
Corrections and feedback
If you find an arithmetic error, unclear explanation, or broken link, please contact the site author with the page URL and the specific issue.