solvedwithstats

An interactive statistical toolbox. SWS does; SBS explains.

Type in your numbers and get a correct, transparent answer — computed entirely in your browser, no account, no tracking, no server round-trip. Every tool links to a companion explainer on Solved By Stats.

Tools

• Estimation

  Confidence Interval Calculator

  t-interval for a mean (from raw data or a summary) and Wilson / normal intervals for a proportion. Shows its work.

• Hypothesis testing

  Welch's Two-Sample t-test

  Compare two independent groups with unequal variances — t, Welch–Satterthwaite df, p-value, CI, and Cohen's d. Raw data or summary stats.

• Study design

  Sample Size & Power Calculator

  How many subjects do you need — and what difference could your study actually detect? Sample size, power, and margin of error for means and proportions. Shows its work.

• Describing data

  Descriptive Statistics Calculator

  Summarise a list of numbers — mean, median, mode, quartiles, spread and skew — and learn which ‘typical value’ to report. Shows its work.

• Uncertainty

  Uncertainty & Error-Bar Visualiser

  See how sample size and spread shape an estimate's plausible range — and why overlapping error bars don't prove ‘no difference’.

• Sampling

  Sampling-Distribution Simulator

  Draw thousands of samples from a population — bell-shaped, skewed, flat or two-humped — and watch their means pile into a bell centred on the truth, or run one average and watch it settle. Seeded, so every run is reproducible.

• Correlation & causation

  Correlation, Causation & Confounding Simulator

  Watch a third variable manufacture a strong correlation out of nothing — then hold it fixed and see what survives. Raw and partial correlations side by side, with the full working shown.

• Significance

  P-Value & Significance Simulator

  Run a thousand experiments in a world where nothing is going on and watch 1 in 20 still cross the p < 0.05 line — then read one experiment honestly: what a p-value is, and what it is not.

• Regression

  Regression Explorer

  Fit the least-squares line to a known truth buried in noise and watch the signal get recovered — then audit the fit in the residual view, drag the line with one far-away point, and meet Anscombe's quartet: four datasets, one summary.

• Model complexity

  Bias–Variance Simulator

  Fit models of rising complexity to dozens of fresh samples from a known truth and watch the error split itself — bias² falling, variance climbing, one interior degree winning. The tradeoff, computed.