When norming tests, it may be preferable to use the matrix sampling technique. The results from the samples may be used to estimate what the distribution of scores would have been if each subject had taken all the items. This paper compares four methods for making these estimates. The sample size made it possible to compare the techniques in a more realistic way than had other studies with much smaller sample sizes. Differences between the criterion distribution and the estimates are tabled. Conclusions and limitations are stated. (Author)