The Multivariate Normal Distribution
Many of the inferential methods used in logistic regression depend on the assumption that the estimates have a sampling distribution that is, asymptotically, a multivariate normal distribution.
The multivariate normal is defined by its distribution function that looks very similar to the univariate case. However it involves a random vector and a variance-covariance matrix.
The multivariate normal distribution has some very nice properties that are often behind the scene in output produced by software (including SAS)