APPENDIX I User Notes This section will contain information relevant to the Current Population Survey, March 1995 file that becomes available after the file is released. The cover letter to the updated information should be filed behind this page. User notes will be sent to all users who purchased their file or technical documentation from the Census Bureau. CURRENT POPULATION SURVEY, MARCH 1995 User Note 1 (April 1994 - August 1995) I. ESTIMATES FOR METROPOLITAN AND NONMETROPOLITAN AREAS During the period April 1994 through June 1995, the Bureau of the Census is systematically introducing a new sample design for the Current Population Survey (CPS) based on the results of the 1990 Decennial Census. During this phase-in period, CPS estimates are being made from two distinct sample designs; i.e., the old 1980 sample design and the new 1990 sample design. While most CPS estimates are thought to be minimally affected by this mixed sample, metropolitan/nonmetropolitan estimates and related CPS geographic estimates are subject to greater error and variability. The causes of these errors are differences in the metropolitan area definitions used in each design, differences in coverage, errors in geographic recoding, and changes in CPS sample areas (including redefining sample areas to correspond to the updated metropolitan definitions.) Using the results from each Decennial Census, the Office of Management and Budget (OMB) revises its official definitions for metropolitan areas. This usually involves adding counties or groups of counties to existing metropolitan areas or designating new metropolitan areas based on certain criteria. As a result, each new CPS sample design represents a larger metropolitan population than its predecessor. The 1980 sample design is based on projected metropolitan area definitions as of June 30, 1983; the 1990 design uses anticipated definitions as of June 30, 1993. During this period, the total number of metropolitan areas increased from 318 to 323. This situation is exacerbated since the mix of the designs (that is, the percent of each design in sample) changes during the phase-in period as more new-design households are introduced. For CPS microdata files containing data collected during the period April 1994 through May 1995, the 1990 design sample was recoded to reflect 1983 metropolitan area definitions. Despite efforts to make the data from the two sample designs consistent, the net result of the different metropolitan definitions in combination with the other changes mentioned above is increased error in metropolitan, nonmetropolitan and related estimates during this phase-in period. As an example, estimates from the January 1995 CPS show an increase of only 1 percent in total civilian noninstitutional population aged 16 and over for metropolitan areas between the 1980 and 1990 designs, but the estimated central city population for the 1990 design is 6 percent higher for the 1990 design. Therefore, the Bureau recommends that users exercise extreme caution when analyzing any tabulations or comparisons using these or related variables during this period. Another constraint on the use of geographic data from this time period is that not all geographic codes on the CPS file were recoded to reflect both the 1980 and 1990 metropolitan definitions. Given below are two lists of variables on the Pre- April 1994 public use files. The first list shows which variables were recoded and are on the files. The second lists contains the variables that have been removed from the files. April 94-May 95 Non-March Files March 1995 Variable Location Location RECODED VARIABLES MSA/PMSA FIPS CODE CHARACTERS 97-100 CHARACTERS 44-47 MSA STATUS CODE CHARACTERS 112-113 CHARACTER 57 CENTRAL CITY STATUS CODE CHARACTERS 108-109 CHARACTER 58 INDIVIDUAL CENTRAL CITY CODE CHARACTERS 110-111 CHARACTER 285 VARIABLES DELETED STATE RANK CHARACTERS 95-96 CHARACTERS 42-43 PMSA RANK N/A CHARACTERS 48-49 MSA/CMSA RANK CHARACTERS 101-103 CHARACTERS 50-52 MSA SIZE CHARACTERS 104-105 CHARACTER 55 CMSA FIPS CODE CHARACTERS 106-107 CHARACTERS 53-54 MSA/PMSA SIZE N/A CHARACTER 56 With the exception of the metropolitan size classifier, all previous breakouts of specific metropolitan tallies can be produced with the data appearing on the April 1994-May 1995 files. II. MATCHING OF HOUSEHOLDS DURING THIS REDESIGN PERIOD As was the case with previous redesigns of CPS, we will be revising the geographic identifiers on the CPS public use files to incorporate the new sample and utilize the updated (June 30, 1993) metropolitan area definitions. We plan to introduce these new definitions on the September 1995 public use file. During this changeover, however, we must take steps to prevent the matching of households containing different sets of detailed geography. We do this in accordance with Title 13, United State Code, to protect the confidentiality of our respondents. Specifically, we will revise our household identification numbering scheme to preclude matching across time frames with different sets of geographic variables. This changeover, along with revisions to the internal census household numbering scheme, have revised the methodology for some matching of households across CPS public use files. In general, this follows the pattern of ten years ago in that data for the three separate time frames are only matchable (within the constraints of the CPS rotation pattern) to itself. These time frames are: 1. Data collected in or prior to May 1995. 2. Data collected in June, July, or August of 1995. 3. Data collected in or after September 1995. Given below are procedures for matching households using data collected in 1994 or 1995: A. 1993 to 1994: Continue using the 12-digit household identification number as before. B. January 1994 through May 1995:* `Use either the FIPS or Census State Code in combination with the 12-digit household ID number and the serial suffix. C. June 1995 through August 1995:* Use the household ID number in combination with state and serial suffix for household matching. Also note that during this three month period, there will be no substate geographic identifiers carried on the CPS public use files. D. September 1995 Files Forward: Use the household ID number to match households. If you have any questions regarding the changeover to the new geographic identifiers and matching of households, call Greg Weyland on (301) 457-3806. See the May 1994 issue of the Bureau of Labor Statistics' issue of Employment and Earnings or call Donna Kostanich on (301) 457-4275 for further information regarding CPS sample redesign. *You will also need the use of these three fields if you want to place the file in a household sort or group household records together. October 1995 CURRENT POPULATION SURVEY, MARCH 1995 User Note 2 (March 1994 - March 1995) Individuals attempting to match the March 1994 Annual Demographic File to any other file will find it impossible to do so. This is because of an error discovered in the program used to create the household identification number (found in Columns 320-331 of the household record) A correction file will be available by October 15 on the internet or on a floppy. Contact Greg Weyland on (301) 457-3806 if you have any questions. October 1995 CURRENT POPULATION SURVEY, MARCH 1995 User Note 3 The March 1995 CPS questionnaire utilized a somewhat more detailed set of health insurance questions that were designed to take better advantage of computer-assisted interviewing. Since the questions were revised, caution should be used in making comparisons between March 1994 and March 1995 health insurance estimates. For the purpose of this file, the health insurance items were recoded to their equivalent fields based on the set of questions that had previously been used. At a later time, the full set of new health insurance fields will be released. Though the new questions did not appear to have a noticeable effect on overall health insurance estimates (neither the number nor the rate of persons without health insurance was different from last year), the new questions did appear to have an effect on individual types of coverage. For example the number of persons with military health care increased as a result of the fact that the field used to define this coverage (CHAMP) also contains persons with "miscellaneous" government insurance. In addition, the number of persons with employer-provided coverage, based on either their own coverage or that of a dependent (COV-GH) is also larger than last year, probably the result of a more straightforward set of questions about this type of coverage (we now ask separate questions about employer and other types of private health insurance plans, whereas previously we asked about employer coverage as a subset of private coverage). A couple of other differences should be noted: 1) the children's health insurance fields (CH-HI and CH-MC) are now consistent with the other health insurance recodes that cover both adults and children (COV-HI and MCAID), and 2) given the new questionnaire, it is possible for a small number of persons covered by their own employer plan not to be asked whether their employer paid all, part, or none of the cost of that plan (if this coverage was reported in the new questions as "other" health insurance). Thus, the number of persons with an entry in HIPAID is slightly less than the number of persons with a "1" in HIEMP. In addition, the field HIPAID has only two values this year. A "1" means that the employer paid all or some of the cost of the plan and a "2" means that the employer paid none of the cost of the plan. October 1995 CURRENT POPULATION SURVEY, MARCH 1995 User Note 4 Source and Accuracy Statement. The Source and Accuracy Statement for Current Population Survey, March 1995 is attached. Earned Income Tax Credit. The person record variable Earned Income Tax Credit (EIT-CRED) is now updated, as a result of the 1994 tax law changes, to include single filers who can claim the earned income tax credit. Detailed Health Insurance Fields. When creating the March 1995 CPS microdata file, health insurance recodes were created to coincide as closely as possible with the health insurance recodes that have been on previous March files. However, as User Note 3 indicates, the health insurance questions were somewhat more detailed in March 1995, as the questions were restructured to take better advantage of computer-assisted interviewing. Hence, the following variables were added to the data dictionary. Begin Position Household Record: HMCARE 274 HMCAID 275 HMCHAMP 276 HHI-YN 277 Family Record: FFNGCARE 251 FFNGCAID 256 Person Record: P-MVCARE 643 P-MVCAID 648 EMCONTRB 653 Please replace pages 9-7, 9-13, and 9-35 in your technical documentation with the attached revised copies to reflect these additional variables. A new file containing these additional health insurance variables and the corrected amounts for the earned income tax credit variable is available on compact disc-recordable (CD-R) at no additional cost from Customer Services, Administrative and Customer Services Division, Bureau of the Census, Washington, DC 20233. Phone: (301) 457-4100. New Detailed Health Insurance Fields. Some users may require more detailed health insurance recodes than those currently contained in the March 1995 CPS file. We created an extract file, mar95ext.dat, that is available on the Internet. File Access. To access this extract file go to the Census Bureau Internet home page http://www.census.gov. Then select /ftp/pub/housing. Once you are in this directory, select mar95ext.dat. You can use ftp to transfer the file to your computer. This file is also available in a compressed (PKZIP) format (version 2.04C). The file, mar95ext.zip, transfers significantly faster than mar95ext.dat. However, users will need PKZIP software to use this file. A record layout for the file (either format) is in the file, nhirl.txt. For more information on accessing this file, copying it, and matching the extract to the full March file, contact: Richard Denby at (301) 763-8174, or e-mail (rdenby@census.gov). For information on the data contained in the extract file, contact: Chuck Nelson at (301) 763- 8029, or e-mail (cnelson1@census.gov). March 1996