Identify the statistical procedure or model being used
Count the total number of independent observations or data values
Subtract the number of constraints, parameters, or estimated quantities
For a one-sample mean, use n – 1
For a two-sample independent t-test, use n1 + n2 – 2
For a paired t-test, use n – 1
For a chi-square goodness-of-fit test, use k – 1 – number of estimated parameters
For a chi-square test of independence, use (r – 1)(c – 1)
For a one-way ANOVA, use k – 1 for between-groups and N – k for within-groups
For regression, use n – p – 1 for residual degrees of freedom
For a contingency table, use rows minus 1 times columns minus 1
For a variance estimate, use n – 1
For a confidence interval or test, match the degrees of freedom to the specific formula being applied