To run a procedure on an input data set that contains observations that occur multiple times, you can use a variable in the data set to represent how frequently observations occur and specify a FREQ statement with the name of that variable as its argument (variable) when you run the procedure.
When you specify a FREQ statement in other SAS procedures, they treat the data set as if each observation appeared n times, where n is the value of variable in the observation. However, PROC MI treats the data set differently because as PROC MI imputes each missing value in each observation, it generates only one imputed value for that missing value. That is, when you specify a FREQ variable, each imputed observation (with its imputed value in place of the missing value) is treated as if it appeared n times. In contrast, if an observation actually occurs n times in the data set, the missing value at each occurrence is imputed separately, and the resulting n observations are not identical.
PROC MI uses only the integer portion of each value of variable; if any value is less than 1, PROC MI does not use the corresponding observation in the analysis. When PROC MI calculates significance probabilities, it considers the total number of observations to be equal to the sum of the values of variable.