Degrees of Freedom
The number of observations that are still free to vary once the parameters have been estimated; in OLS with regressors and an intercept it is .
When to use
It sets which t-distribution to use for critical values and confidence intervals, and the variance estimate divides by it. In F-tests, is the difference in degrees of freedom of the restricted and unrestricted models.
See Lec 5.