"Classification systems change from census to census for a variety of reasons. The change from 1970 U.S Census to 1980 U.S Census classification was so dramatic that studying the changes and making comparisons are too complicated and expensive. Treating the actual census results as unknown, we simulated a new Census data base reflecting the real situation in 1970 & 1980 classification systems. One of our objective is to explain the process by which codes change so that the researchers can better understand how the new data bases were created. The second objective is to show how this newly created data base is then used to study the comparability of the two classification systems. In this project we do not attempt any estimative or predictive inference. We simply simulate the industry and occupation codes in the U.S. Census public-use samples via a model similar to the one used for multiple imputation."