National Academies Press: OpenBook
« Previous: Population Coverage
Suggested Citation:"Household and Individual Nonresponse." National Research Council. 1991. Improving Information for Social Policy Decisions -- The Uses of Microsimulation Modeling: Volume II, Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/1853.
Page 26
Suggested Citation:"Household and Individual Nonresponse." National Research Council. 1991. Improving Information for Social Policy Decisions -- The Uses of Microsimulation Modeling: Volume II, Technical Papers. Washington, DC: The National Academies Press. doi: 10.17226/1853.
Page 27

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

DATABASES FOR MICROSIMULATION: A COMPARISON OF THE MARCH CPS AND SIPP 26 “missing men” to some of these households might reduce the size of the eligible pool, depending on their relationship to the AFDC unit, their employment status, and their contribution to the household's resources. The validation experiment conducted by the Panel to Evaluate Microsimulation Models for Social Welfare Programs included crude adjustments for undercount to the March 1984 CPS that formed the database for the experiment and to the March 1988 CPS that was used to generate demographic controls for projecting the former data forward in time (see Cohen et al., Chapter 8 in this volume). The simulations on the adjusted files indeed increased the size of the AFDC-eligible population and, correspondingly, lowered the simulated participation rate by about 4 percentage points (from 76 to 72% on the March 1984 CPS and from 78 to 74% on the March 1988 CPS). Participation rate changes varied by state: some states, like New Hampshire and Indiana, had very little change; others, like Illinois, Michigan, and the District of Columbia, had large changes. (The simulated participation rate dropped by more than 10 percentage points in the District of Columbia and by 8 points in Illinois.) However, this experiment did not implement the kinds of intrahousehold adjustments in composition and other characteristics that are undoubtedly necessary to compensate fully for population undercoverage. House hold and Individual Nonresponse CPS interviewers generally obtain high response rates from households. In recent years, typically, about 4– 5 percent of households have failed to respond to the March CPS, and another 9 percent of people in interviewed households have failed to respond. The Census Bureau adjusts for whole household nonresponse by increasing the weights of responding households. Only a few variables are used in the adjustment, including geographic area, rotation group, and race of the household head (if known). The adjustment for person nonresponse in the CPS uses a procedure that the Census Bureau refers to as a “statistical match” to impute an entire record for each nonrespondent.5 In this procedure, the records are indexed by various characteristics that are available for both respondents and nonrespondents, and a search is made for the respondent donor who best matches each nonrespondent.6 5Alternatives to imputing for person (and item) nonresponse would be to delete households with missing data and reweight the remaining households or to retain all households and leave missing data items blank. However, because many households have some missing data, it is believed to be preferable to perform an imputation so as to preserve as many cases as possible for analysis and provide users with a complete record for each case. 6More commonly, the term “statistical match” is restricted to a match between records from two different survey samples based on a set of common items asked in both surveys (see Cohen, Chapter 2 in this volume, for a review of statistical matching techniques). In statistical matches between surveys, as well as in the Census Bureau's procedure, a goal is to maximize the use of the available information by seeking to match a pair of records by using as detailed a categorization for as many variables as possible and relaxing the criteria for the match (by collapsing categories or omitting match variables) only to the extent necessary. In the case of imputation of entire records for person nonresponse in the CPS, very few variables are available for the match. In contrast, in the case of imputation for item response (see discussion below), the match employs very elaborate matrices of characteristics.

DATABASES FOR MICROSIMULATION: A COMPARISON OF THE MARCH CPS AND SIPP 27 There has been no analysis of the impact of household and person nonresponse or of the adjustment procedures on estimates from the March income supplement. SIPP interviewers initially obtain response rates about as high as in the CPS. However, over the lifetime of a panel, response rates inevitably decline, as people become tired of responding for one or another reason or move and cannot be traced. The estimated cumulative sample loss of households for the 1984 panel, including households not interviewed because of refusal and households not interviewed because they moved and could not be traced, was 5 percent at the first interview wave, 12 percent at the third wave (after 1 year of interviewing), 19 percent at the sixth wave (after 2 years of interviewing), and 22 percent by the eighth and ninth waves. Sample loss figures for the 1985 through 1988 panels were about 7 percent at wave one, 13–15 percent at wave three, and 18–20 percent at wave six. The SIPP experience is typical of other longitudinal surveys, such as the Panel Study of Income Dynamics (PSID), in which sample loss is greatest at the early interviews and climbs slowly thereafter (Nelson, Bowie, and Walker, 1987:7–10). In addition to household nonresponse in SIPP, there is person nonresponse. In the 1984 SIPP panel, 74 percent of the original sample members completed all interviews (or all interviews before leaving the universe), while another 6 percent missed only one interview. Less than 10 percent completed fewer than three interviews.7 The Census Bureau's procedures for adjusting for household and person nonresponse in the SIPP on a cross- sectional basis, that is, for the data records for each interview wave, are similar to those described for the CPS. Household nonresponse is handled by a weighting adjustment, and person nonresponse is handled by the Census Bureau's “statistical match” procedure. (This procedure is applied not only for people who failed to be interviewed and for whom no proxy response is available, but also for people who died, were institutionalized, or left the country within a 2-month period preceding the interview.) It is important to note that the cross-sectional nonresponse adjustments in SIPP make only minimal use of the information that is available from previous waves for many current nonrespondents. In constructing longitudinal files from SIPP panels, the Census Bureau assigns zero weights to those households and individuals who were not interviewed for one or more waves (either because of nonresponse or because they were not original sample members) and reweights the remaining cases. (Original sample members who died or otherwise left the 7The base for these percentages excludes original sample members in households that did not respond at all in the first wave, as no attempt is made subsequently to obtain interviews for these households.

Next: Item Nonresponse »
Improving Information for Social Policy Decisions -- The Uses of Microsimulation Modeling: Volume II, Technical Papers Get This Book
Buy Paperback | $100.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

This volume, second in the series, provides essential background material for policy analysts, researchers, statisticians, and others interested in the application of microsimulation techniques to develop estimates of the costs and population impacts of proposed changes in government policies ranging from welfare to retirement income to health care to taxes.

The material spans data inputs to models, design and computer implementation of models, validation of model outputs, and model documentation.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook,'s online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!