I have received notice that there is another error in the CPS files. This error involves place of birth. The error has been corrected but we do not have the new release of the data yet. We have a new version of the outgoing rotation files which corrects the previously described errors (August 9 notice) but has the error described below. Our March 1995 CPS file still has the errors described in the August 9 notice, as well as this error.

In January 1995, the Census Bureau introduced a new set of answer screens to the place of birth question in the CPS CAPI instrument. These items are numbered 1-42. The screen containing items 36 and higher---for example, "South Korea," "Thailand," "Vietnam" was omitted. Many of the Southeast Asian countries have names that begin with letters after "S."

The following illustrates the magnitude of this error using three of the affected Asian countries:

                   '94 CPS'       '95 CPS'       '96 CPS'
 
Korea     Old       533,000        335,700         62,100
          New       533,000        517,830        550,340
 
Vietnam   Old       496,000        189,150         46,867
          New       496,000        447,770        740,050
 
China     Old       565,000        597,980      1,076,700
          New       565,000        515,690        801,380
 

In the table above, the "New" row provides the corrected figures, in years for which there was an error. The "Old" row provides the incorrect figures, in years for which there was an error.

The new corrected data are much more consistent with known immigration trends (and 1990 census data) than the original CPS data. There is a slight inconsistency between the '95 and '96 CPS estimates that arises from the new race edits introduced in 1996, but that is minor compared to the country of birth coding problem.

Source: Communication with Dianne Schmidley and Greg Weyland, Bureau of the Census