How to use Statistical Hypothesis Test Calculator
- Choose the supported test. Use sample SD for one-sample t and known population SD for z; Welch compares two independent group means against a zero difference.
- For a mean test, enter the required summaries and choose the alternative before interpreting data. For chi-square, enter observed and expected counts as two tab-separated columns without a heading.
- For chi-square, use 2 to 100 categories, expected counts at least 5 and matching observed/expected totals. Enter how many distribution parameters were estimated; the test uses an upper tail.
- Enter significance level as a fraction, such as 0.05 for 5%, and choose display precision.
- Calculate and review statistic, degrees of freedom, p-value and decision alongside the sampling assumptions. A threshold decision is not an effect-size or causal conclusion.
Example: Statistical Hypothesis Test Calculator
Test whether a sample mean of 10 differs from null mean 9, with sample SD 2, n = 25 and a two-sided one-sample t test.
Options
- Test and summaries
- Choose the supported test before entering its required summaries. Welch compares independent means without assuming equal variances.
- Alternative and alpha
- Choose the direction and significance threshold before interpreting the result. Changing the tail after seeing the output changes the question being tested.
Supported inputs and limits
- Only one-sample t, known-population-SD z, two-independent-mean Welch t and chi-square goodness-of-fit tests are supported.
- Alpha is 0.000001 to 0.25.
- Sample sizes are whole numbers up to 1,000,000, with at least two for t/Welch and one for z.
- SD must be positive; numerical degrees of freedom above 1,000,000 are refused.
- Chi-square uses 2 to 100 categories in at most 10,000 characters.
More details
Observed counts are nonnegative integers; expected counts are at least 5 and totals must match within the numerical tolerance.
Degrees of freedom subtract one plus the entered estimated-parameter count.
The tool does not provide paired tests, weights, contingency-table independence tests or multiple-comparison correction.
Mean tests need suitable independent sampling and distribution assumptions.
A p-value is not the probability that a hypothesis is true, and failure to reject does not establish equality.
Display precision cannot prevent numerical underflow or repair an unsuitable model.
Where your input is processed
This tool does not upload your inputs. See the limits for anything saved on your device.
A small p-value does not describe the size of an effect
The p-value measures compatibility with the null model through the selected test statistic and tail. It does not give the probability that the null is true. Review assumptions, sample design and effect size alongside the result; a threshold decision alone cannot establish cause or practical importance.
Questions about Statistical Hypothesis Test Calculator
Does a large p-value prove the null hypothesis?
No. It means this supported test did not provide sufficient evidence against the stated null at the chosen threshold.
Does Welch require equal variances?
No. It uses the two entered variances and its own degrees-of-freedom estimate.
Can I choose expected counts after inspecting a result?
Expected counts and any fitted parameters need a defensible model; changing them after testing changes the question.