Let be the response variable with categories . The covariates are denoted by a -dimension row vector .
For a stratified clustered sample design, each observation is represented by a row vector, , where
is the stratum index
is the cluster index within stratum
is the unit index within cluster of stratum
denotes the sampling weight
is a -dimensional column vector whose elements are indicator variables for the first categories for variable . If the response of the th unit of the th cluster in stratum falls in category , the th element of the vector is one, and the remaining elements of the vector are zero, where .
is the indicator variable for the category of variable
denotes the -dimensional row vector of explanatory variables for the th unit of the th cluster in stratum . If there is an intercept, then .
is the total number of clusters in the sample
is the total sample size
The following notations are also used:
denotes the sampling rate for stratum
is the expected vector of the response variable:
Note that , where 1 is a -dimensional column vector whose elements are .