In order to obtain a society’s characteristics, information about the society members should be accessible. One method to collect this information is sample survey. Depending on how to choose sampling units, there are two different sampling methods: probability sampling and non-probability sampling. In non-probability sampling approach, parameter estimates are bias, so results in the class of biased estimators should be studied. This approach has been mostly considered in the most of the surveys for the good speed and low costs. One of the non-probability sampling methods is cut off sampling, which is more used in social-economic survey, especially in determining different economical indices (price index) such as industrial establishment survey, in national statistical offices.

In this paper different types of cut-off sampling are discussed. Data drawn from industrial establishment census (establishments that had 10 workers and more in 1385), are used to estimate the added value (in the same year, 1385) for each establishment by using type 3 cut off sampling, and performance of two approaches, deleted and model assisted approaches, are compared. Results of this study show that under certain conditions model-based approach estimator has smaller MSE in comparison with that of the deleted approach. 

Finally, results of the model assisted approach are compared with a classical method of estimation.