Statistics & Probability calculators
Statistics is the discipline of saying how much a limited set of observations can support. These calculators cover the two halves of that job: descriptive tools that summarise the data in front of you, and inferential tools that quantify how far you can generalise beyond it. Each one shows the intermediate quantities — sums of squares, standard errors, degrees of freedom — because those numbers are what let you tell a result you can defend from one that merely looks tidy.
9 calculators in this category
Most used statistics & probability calculators
Describe before you infer
Every analysis should start with the shape of the data. A mean well above the median signals a long right tail; two modes usually mean two populations have been mixed; a standard deviation that rivals the mean says the average is not describing anything typical. These are one-minute checks that routinely change the choice of method.
The habit that prevents most statistical mistakes is plotting the data before computing anything. Anscombe's quartet — four data sets with identical means, variances and correlations that look completely different on a scatter plot — is the standing proof that summary numbers alone cannot tell you whether a model fits.
The square root of n governs everything
Precision improves with the square root of sample size, not with sample size. Doubling a survey shrinks its margin of error by only 29%; halving the margin requires four times the data. That single relationship explains why national polls settle at around a thousand respondents, why A/B tests need far more traffic than teams expect, and why chasing the last percentage point of precision gets expensive very quickly.
It also explains why population size barely matters. Beyond about twenty thousand people, the required sample is essentially flat — sampling a city and sampling a continent cost the same, which is one of the least intuitive true facts in the field.
What the numbers do not say
A 95% confidence interval is a statement about the procedure, not about the particular interval in front of you: repeat the sampling many times and about 95% of the intervals produced would contain the true value. A correlation is consistent with X causing Y, Y causing X, or a third variable driving both. A p-value is the probability of data this extreme if the null hypothesis were true, and nothing more.
Model assumptions carry the same caveats. Normal-based percentiles need roughly normal data. Binomial probabilities need independent trials with a constant success rate. Least squares treats a single mistyped observation as gospel. Knowing which assumption is doing the work is the difference between a calculation and a conclusion.
Every statistics & probability calculator
Statistics & Probability calculator questions
Should I use the sample or population formula?+
Use the sample version, dividing by n − 1, unless your data is a complete census of the group you are describing. Almost all real data is a sample.
How many observations do I need?+
For a survey at ±5% and 95% confidence, about 385 completed responses. The sample size calculator handles other margins, confidence levels and small populations.
When do I need t instead of z?+
When the standard deviation is estimated from a small sample, roughly under 30 observations. The t distribution has fatter tails and produces wider, more honest intervals.
Do these calculators run significance tests?+
The correlation and z-score tools report test statistics and p-values, and the confidence interval tool answers most practical significance questions by showing whether a value of interest falls inside the interval.