Problems with the March 1995 CPS

August 9, 1996

April, 1997: A corrected 2nd Edition of these data has been received. Please see the Update Notice for revised edition information.

There is an unusually high error rate in several variables in the March 1995 CPS data that have been distributed by the Census Bureau.

According to Greg Weyland at the Census Bureau, data errors in the March 1995 files include mistakes in the household identifiers and in the variables for citizenship, substate geography, month in sample (for Hispanics), and health insurance.

This error affects researchers using the original March 1995 CPS files, the Unicon uniform March CPS CD-ROM, and the 1994 and 1995 outgoing rotation files from the Bureau of Labor Statistics.

The Population Studies Center has the "bad" versions of the data. We will notify users when we get the "replacement" versions.

Source: Communication with Unicon (Grabowski), ICPSR (Rockwell and Austin) and Bureau of Labor Statistics (Montgomery)