National Academies Press: OpenBook
« Previous: SURVEY DESIGN OF THE MARCH CPS AND THE SIPP
Suggested Citation:"Population Coverage." 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 21
Suggested Citation:"Population Coverage." 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 22
Suggested Citation:"Population Coverage." 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 23
Suggested Citation:"Population Coverage." 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 24
Suggested Citation:"Population Coverage." 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 25

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 21 collection and processing procedures, etc.—to determine whether and what kinds of major changes may be needed. The Census Bureau is also exploring the concept of using SIPP, CPS, and administrative records data to improve the quality of income statistics (see further discussion, below). All aspects of the current CPS and SIPP design and data collection procedures affect the quality and utility of the data for microsimulation modeling of income support programs. SURVEY-BASED PROBLEMS This section covers population coverage, household and individual nonresponse, item nonresponse, reporting errors, sample size, and timing of data delivery. Population Coverage It is well known among survey statisticians that household surveys rarely cover the population as well as the decennial census (see Citro and Cohen, 1985; Shapiro and Kostanich, 1988). The CPS and the SIPP are no exceptions to this pattern. Table 1 shows “coverage ratios” for the March 1984 and 1986 CPS and SIPP by age, race, and sex. A coverage ratio is the ratio of the population estimated from the survey, after application of the sampling weights (the reciprocal of the sampling fraction), including adjustments for nonresponse, to the population estimated from the most recent decennial census updated by vital records on births, deaths, and net immigration. Overall, the March 1984 CPS covered an estimated 84 percent of black males and 90 percent of black females, compared with 92 percent of all other males and 94 percent of all other females. The March 1984 SIPP coverage ratios were about the same—85 percent of black males and 92 percent of black females, compared with 91 percent of all other males and 93 percent of all other females. For the CPS, coverage ratios for blacks were slightly lower for March 1986 (compared with the March 1984 CPS)—82 percent of black males and 89 percent of black females in 1986—and about the same in the two years for whites. The March 1986 SIPP showed the same decline for blacks compared with 1984—coverage ratios were 80 percent for black males and 89 percent for black females in 1986—while the SIPP showed coverage improvements for whites between the two years. Black males have the lowest coverage ratios overall, and, by age, black me n in the 20–39 age categories are generally the worst-covered subgroups. The Census Bureau uses ratio-estimation procedures to adjust the survey weights for population undercoverage in both the CPS and the SIPP. The weights are adjusted so that the population estimated from each survey agrees with the updated decennial census-based population estimates by age, sex, race, and

TABLE 1 Coverage Ratios for SIPP and CPS Samples: March 1984 and March 1986 DATABASES FOR MICROSIMULATION: A COMPARISON OF THE MARCH CPS AND SIPP 22

DATABASES FOR MICROSIMULATION: A COMPARISON OF THE MARCH CPS AND SIPP 23

DATABASES FOR MICROSIMULATION: A COMPARISON OF THE MARCH CPS AND SIPP 24 Hispanic origin. SIPP weights are also adjusted to agree with CPS weights by household type.4 However, these ratio adjustments do not correct all coverage errors. First, they do not correct for the undercoverage in the decennial census itself, which is minimal in total—net undercount was estimated to be only a little over 1 percent of the population in 1980—but substantial for some population groups. Thus, in 1980, an estimated 9–10 percent of black children under age 5 were missed, as were an estimated 13 percent of black me n aged 25 to 34 and 16–18 percent of black me n aged 35 to 54. (See Fay, Passel, and Robinson [1988: Tables 3.2, 3.3]. However, recent work by Robinson [1990] suggests that undercount rates ma y have been overestimated by several percentage points for black men aged 35–54, because of errors in estimating the incompleteness of birth registration for these groups.) Second, the ratio adjustments do not correct for differences other than age, sex, and ethnic origin on which the undercovered population might be expected to differ from the covered population. Moreover, the absolute size of the bias for a population subgroup will generally be positively correlated with the proportion of the group that the survey fails to cover. The correlates of undercoverage are not definitively established. However, analysis of the postenumeration survey conducted as part of the 1980 census and of other survey, administrative records, and ethnographic data suggests that the following characteristics in addition to age, race, and sex relate to undercount (see Citro and Cohen, 1985; Fein, 1989): • Household relationship Undercount rates are higher for household members other than the head, spouse, and children of the head. • Marital status Undercount rates are higher for people who are not married. • Household size Undercount rates are higher for one-person households and for very large households. • Income The rate of undercount increases as household income decreases. • Geographic location Undercount rates are higher in central cities of large metropolitan areas. In addition, a very strong correlate of the undercount rate in the 1980 census was an indicator of responsiveness to the census, namely, the rate of 4The process of creating final CPS and SIPP weights for each record is complex and involves several stages, including obtaining the base weight, which is the reciprocal of the sampling fraction; adjusting for household and person nonresponse; adjusting to reduce sampling variance for the PSUs that were grouped into strata and subsampled; and adjusting for agreement with census-based population control totals. Weights vary within households (except that the weights of husbands and wives are adjusted to be the same). The Census Bureau produces household and family weights, which are the weights of the principal person in each case, for use in household- and family-level analyses. It is not clear how these different weights might affect analyses for other kinds of units, such as AFDC units or tax filing units.

DATABASES FOR MICROSIMULATION: A COMPARISON OF THE MARCH CPS AND SIPP 25 return of questionnaires in the mail: district offices that had low mail return rates had high undercount rates. Finally, the evidence suggests that, due to improvements in techniques for developing census address lists, the problem of undercoverage has shifted from one of locating buildings to one of identifying everyone who is associated with a particular household. Thus, the 1950 census evaluation program found that three-quarters of all missed people were in whole households that were missed with the remainder in otherwise enumerated households. The 1970 census evaluation program, in contrast, found that fully one-half of missed people were in otherwise enumerated households (Citro and Cohen, 1985:187). Fay (1989) has analyzed within-household undercoverage in the CPS relative to the decennial census, using results from the match of the April 1980 CPS to the 1980 census that was carried out for the 1980 postenumeration program. His results do not speak to the added undercoverage in the census itself and pose problems of interpretation because of several complexities in the analysis (for example, the difference in concepts of place of residence for college students in the CPS and the census). However, they are suggestive of ways in which weighting adjustments do not adequately compensate for household survey undercoverage. For example, he finds that about one-fourth of adult black men who are counted in the census but not in the CPS are household heads, whose households should be categorized as married couple households in the CPS but instead are categorized as households headed by unmarried women. Another one-half of adult black men who are counted in the census but not in the CPS are members of single female-headed households (such as other relatives or nonrelatives). The patterns are somewhat different for adult Hispanic males who are counted in the census but not in the CPS, more of whom are missing from households identified in the CPS as headed by a married couple than from single female-headed households. The results of initial analyses of undercoverage in the 1990 census and how it compares with the patterns that were evident in earlier censuses have recently become available. In total, coverage appears to have gotten slightly worse: for the population as a whole, the net undercount has been estimated at 1.8 percent in 1990, compared with 1.2 percent in 1980. Also, an estimated 5.7 percent of blacks were undercounted in 1990, compared with 1.3 percent of others; the corresponding figures for 1980 are 4.5 percent and 0.8 percent (Bureau of the Census, 1991: Table 3). Overall, these tentative findings about census and survey undercoverage suggest that the welfare-eligible population is at much greater risk of not being counted than other people. However, assessing the impact of undercount on estimates of costs and caseloads for income support programs from the March CPS or the SIPP is not straightforward. For example, increasing the number of low-income households through an undercount adjustment would presumably enlarge the eligible pool for programs such as AFDC. On the other hand, adding

Next: Household and Individual 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, NAP.edu'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!