2008 American Community Survey / Puerto Rican Community Survey Data

The 2008 American Community Survey (ACS) and Puerto Rican Community Survey (PRCS) contain many important changes from previous ACS/PRCS releases. (General information on the ACS and PRCS can be found here.) As always, we encourage users who notice changes to the ACS not reflected in the IPUMS data to email ipums@umn.edu. For a full description of changes to the ACS/PRCS questionnaire, including changes to question wording and order, see this document. Users may also want to familiarize themselves with the likely substantive impacts of these changes by examining the Census Bureau's documentation.

Items Newly Available

  • New items on health insurance (HCOVANY, HCOVPRIV, HCOVPUB, HINSEMP, HINSPUR, HINSCARE, HINSCAID, HINSTRI, HINSVA, HINSIHS). For details, see the ACS health insurance page.
  • New items on family changes in the last year, including marriage (MARRINYR), becoming widowed (WIDINYR), or divorce (DIVINYR). The year in which one was last married (YRMARR) is also new to the IPUMS. MARRINYR has long been available in the 1850-1880 samples; it now makes its appearance in the ACS/PRCS. Additionally, information on the number of times one has been married (MARRNO) is now available for the first time since the 1980 census.
  • New item on veterans' service-connected disabilities (VETDISAB); background is available from the Census Bureau.
  • New indicator of multigenerational households (MULTGEN). This is provided exactly as the Census Bureau released it; users should read the variable description before analyzing it. We plan to provide an IPUMS version of this variable in the future for all years. UPDATE: The IPUMS version of this variable was posted on Jan. 12, 2010.
  • New item on year of naturalization (YRNATUR). Though available in the 1920 1% sample, this has not been available in the ACS before. The variable now indicates the full four-digit year, rather than the last three digits (as was the case previously).

Items With More Detail

  • Additional detail on relationship to head (RELATE). In previous ACS/PRCS questionnaires, the Census Bureau provided only a general in-law category that contained parents-in-law, children-in-law, and siblings-in-law; and the IPUMS used logical rules to break this general category into its three specific types. In the 2008 questionnaire, the parent-in-law and child-in-law relationship categories were provided on the form and can be identified directly from the data, but siblings-in-law are likely to be subsumed under the "other relative" category. The IPUMS attempts to pull probable siblings-in-law out of the "other relatives" category; details on this classification procedure are available here. Additionally, adopted children and stepchildren can now be identified directly from the data, resulting in more detail in the detailed version of RELATE and in the family interrelationship variables STEPMOM and STEPPOP. Because adopted children and stepchildren were previously coded as children of the householder and not as "other relatives", there should be no change in the accuracy of the pointer variables MOMLOC and POPLOC.
  • Additional detail on KITCHEN facilities (SINK, STOVE, FRIDGE)
  • Additional detail on PLUMBING facilities (HOTWATER, TOILET, SHOWER). The coding of HOTWATER, available in earlier years as well, has been changed to accommodate the new data.
  • Additional detail on property value (VALUEH), which has been widened to 7 digits to accommodate increased detail among the most expensive properties. Additionally, the topcode for 2000-2007 data has been changed from $999,998 to $1,000,000.
  • Additional detail on school grade attending (GRADEATT) and educational attainment (EDUC).
  • New ancestry (ANCESTR1, ANCESTR2) codes for "Irish Scotch", "Germanic", and "Mexican Indian".
  • Additional detail on sensory difficulties. The IPUMS variable DIFFEYE (which combined vision and hearing difficulties) has been renamed DIFFSENS. DIFFEYE is now limited exclusively to vision difficulties, and the new variable DIFFHEAR measures hearing difficulties. The universe for these variables is now all persons; previously, it was limited to ages 5+.
  • The IPUMS names for the ACS disability measures have been standardized and now better reflect the content of the questions. However, because of substantial changes to the layout and content of questions, the Census Bureau recommends that researchers not compare disability questions in 2008 and successive surveys to those in pre-2008 surveys. A summary of changes to the disability variables follows; users may also wish to consult the Census Bureau's website.
Concept Old Name New Name Flag Name
Vision or hearing difficulty DIFFEYE DIFFSENS QDIFSENS
Vision difficulty (2008 and beyond only) --------- DIFFEYE QDIFEYE
Hearing difficulty (2008 and beyond only) --------- DIFFHEAR QDIFHEAR
Cognitive difficulty DIFFREM DIFFREM QDIFREM
Independent living difficulty DISABMOB DIFFMOB QDIFMOB

Items Removed Or With Less Detail

  • Detailed information on weeks worked last year (WKSWORK1) is no longer available, but the intervalled version (WKSWORK2) remains.
  • Detailed information on the amount of food stamps received (FDSTPAMT) is no longer available, but the status of recipiency (FOODSTMP) remains.
  • Information on the number of years served in the military (VETYRS) is no longer available, but other information (e.g., VETSTAT) remains.
  • Information on difficulties limiting work (DISABWRK) is no longer available.


UPDATE: The 2008 IPUMS ACS/PRCS data were re-posted on Friday, November 6, after the Census Bureau released new data. HHINCOME, FTOTINC, INCTOT, and INCEARN are now available, and POVERTY is constructed exactly as in previous samples.

Any 2008 PUMS files downloaded from the Census Bureau's website between October 30 and November 3, 2009 have incorrect data for the four income summary variables with the Census Bureau names of FINCP, HINCP, PERNP, and PINCP. DataFerrett data may still be flawed; users should check for a high number of negative values before using any DataFerrett data.

The original note posted on November 4 follows:

Please note that we are not providing certain variables at this point due to apparent errors in the original Census Bureau data files. The affected variables are HINCP (IPUMS equivalent = HHINCOME), FINCP (FTOTINC), PINCP (INCTOT), and PERNP (INCEARN). Briefly, there were unexplained discrepancies in the structure of the various data files released by the Census Bureau. The comma-delimited ASCII files contained only a 3-character code in scientific notation for each of these four income variables, resulting in only 1 real number of income data. For instance, a household income of $33,200 would show up as "3E4" in these variables. The SAS files contained many unexplained blank values for these variables. The data downloaded directly from DataFerrett replaced these blanks with the stated bottom-code value from the data dictionary (-$59,999 for HINCP and FINCP, -$10,000 for PERNP, and -$19,999 for PINCP), resulting in many more bottom-coded incomes than were present in other years.

The Census Bureau released updated data files over the night of Tuesday, November 3. While the revised variables could not be integrated in time for the planned November 4 release, IPUMS will provide them by Friday, November 6. Until then, users can make substitutes by summing income components within individuals (to obtain INCTOT and INCEARN), then summing within families and households (to obtain FTOTINC and HHINCOME). The resulting variables will differ slightly from those in previous samples because of Census Bureau topcoding, but they will be accurate in most cases.

The IPUMS variable POVERTY is constructed using FTOTINC, and so it cannot yet be calculated as in other samples. We have provided it just as the Census Bureau released it; preliminary inspection suggests that it was unaffected by the data problems. The IPUMS updated version of POVERTY will also be released on November 6; it will change only for nonrelatives of the householder whom the IPUMS codes as being in an unrelated subfamily. See the variable description for more information.