To obtain information for stratification when the stratification variable is not known for the whole population.

To reduce the cost of measuring an expensive variable by using a cheaper auxiliary variable first.

To increase the response rate by sending two surveys instead of one.

To eliminate all non-sampling errors.