Direct answer
Student’s t models a standardized mean when spread is estimated, chi-square models sums of squared standard-normal components, and F models a ratio of scaled chi-square quantities.
Visual explanation
Degrees of freedom reshape distribution families
What this calculation tells you
These distributions account for uncertainty introduced by estimating variation. Degrees of freedom record how much independent information remains after fitted constraints.
Match the statistic and design: mean comparisons often use t, variance inference uses chi-square under normality, and variance ratios or ANOVA use F. Do not select them merely because a critical-value table is available.
Where it is used
Mean inference
Use t-based uncertainty when the classical conditions and estimated spread apply.
Variance inference
Use chi-square relationships under an explicit normal model.
ANOVA and regression
Interpret F ratios with documented model structure.
Education
Connect squared normal variables and variance estimation to distribution shape.
Common situations
- Choosing t rather than z.
- Looking up a variance interval.
- Interpreting an ANOVA F statistic.
- Explaining how degrees of freedom change tail weight.
Start with the statistical question
Match the statistic and design: mean comparisons often use t, variance inference uses chi-square under normality, and variance ratios or ANOVA use F. Do not select them merely because a critical-value table is available.
Animated families show t tails approaching normal as df increases, chi-square shifting and becoming less skewed, and F changing with two degree-of-freedom inputs.
Worked example
A one-sample mean based on n observations has t degrees of freedom n−1 when the sample standard deviation estimates population spread under the classical model.
Assumptions that carry the result
Classical derivations often require independent normal observations or normal residuals. ANOVA and regression add design, linearity, and variance assumptions that a distribution lookup cannot verify.
Interpret the result without overreaching
A correct tail probability under a model does not establish that the design, residuals, independence, or substantive hypothesis is appropriate.
- Using z critical values after estimating spread from a small sample.
- Forgetting that F has numerator and denominator degrees of freedom.
- Treating degrees of freedom as simply sample size in every design.
Practical questions
Frequently asked questions
Does t become normal?
Its distribution approaches standard normal as degrees of freedom increase, though the exact model remains distinct.
Why is chi-square never negative?
It is built from sums of squared standardized quantities.
Why does F need two degrees of freedom?
Its numerator and denominator each contain a separately scaled variance-like quantity.
Further reading
Authoritative sources
Use these primary and professional resources to check definitions, conventions, or requirements that may extend beyond this guide.