In the1980, 1990, and 2000 censuses and the ACS, information on ancestry was collected via an open-ended question. Because virtually any response was allowed in the census and the ACS, IPUMS-USA includes a much wider range of responses than IPUMS-CPS does. The census and ACS samples also record two responses; these are included in the variables ANCESTR1 (for first response) and ANCESTR2 (for second response). For comparability to IPUMS-CPS, all responses in the census and ACS samples other than the thirteen options given to CPS respondents should be recoded as "other," or analysts should focus on the specific categories present in IPUMS-CPS (e.g., persons of "Russian" ancestry). Some categories which have the same codes in IPUMS-USA and IPUMS-CPS (e.g., "South American") are more inclusive in the CPS, because very few specific countries were included as possible choices for the survey respondents. Thus, all responses for individual South American countries in IPUMS-USA would have to be combined into a catchall "South American" category for comparability with IPUMS-CPS. Similarly, only broad categories like "French" or "German" are included in IPUMS-CPS, while small regional subcategories for these countries are part of the IPUMS-USA database.