Summaries of National-Level Survey Data Sets Relevant to Welfare Monitoring and Evaluation
DECENNIAL LONG FORM
2000 Census Long Form
The 2000 census, like every census since 1960, included a long-form questionnaire that is administered to a sample of households. The long form includes the short-form questions that are asked of all households with additional questions that are included only on the long form. The added questions include ones on total income and income from seven different sources (e.g., wages, Social Security, and public assistance or welfare benefits, etc.) for the previous calendar year for each household member aged 15 or older. Both the short-form and long-form census questionnaires are mandatory.
The sample design for the 2000 census long form was similar to the design used in the 1990 census with some modification. In 1990 the overall sampling rate was about 1 in 6, producing a sample of about 18 million occupied housing units. In 2000, the overall sampling rate was again be about 1 in 6, producing a sample of about 18 million housing units.
Data collection in the census is mainly by self-enumeration, whereby a respondent for each household fills out a questionnaire received in the mail.
Enumerators follow up those households that fail to return a questionnaire and collect the information through direct interviews. Response rates to the census mailout have declined since 1970, when mailout-mailback techniques were first used. In 1990 approximately 60 percent of U.S. households mailed back their long form questionnaires; in 2000 approximately 54 percent mailed back their questionnaires.
As in all censuses, there were uncounted people in 1990; there were also duplications and other erroneous enumerations. The net undercount in 1990 (gross undercount minus gross overcount) was estimated at 1.8 percent for the total population, but there were substantial differences among population groups. For example, the net undercount was estimated at 5.7 percent for blacks and 1.3 percent for nonblacks. The net undercount also varied significantly by age: almost two-thirds of the estimated omitted population consisted of children under age 10 and men aged 25–39 (Robinson et al., 1993:13).
Item nonresponse rates in 1990 were generally higher for income than for most other items. When household income information is missing, the Census Bureau uses statistical techniques to impute data on the basis of nearby households with similar characteristics. On average, 19 percent of aggregate household income was imputed for 1990 (National Research Council, 1995:387).
Processing and release of the long-form sample data will be provided for areas as small as census tracts and school districts. Typically, long-form data products are released beginning in year 2 and continuing through year 3 after the census year.
Additional information can be found at The Census Bureau website for the census: (http://www.census.gov/dmd/www/2khome.htm).
AMERICAN COMMUNITY SURVEY
The American Community Survey is planned to be a large-scale, continuing monthly sample survey of housing units in the United States, conducted primarily by mail. It will include content similar to that of the decennial census long-form sample, including questions on income and its sources and on participation in public assistance programs, such as cash assistance and noncash assistance such as food stamps, and rental subsidies. The income and assistance receipt questions will refer to the 12 months preceding the interview month. It is planned that the ACS will be mandatory, like the census, rather than a voluntary survey. If the ACS is successfully implemented, there will likely be no long form in the 2010 and subsequent censuses.
Development of the ACS began in 1996 when the survey was tested in four sites, in 1997–1998 it was tested in eight states. Beginning in 1999 and extending through 2001, the ACS will be conducted in 31 sites, chosen to facilitate comparison with the 2000 census long-form data for census tracts and other areas. In
25 of the 31 sites, about 0.4 percent of housing units will be sampled each month, which will generate a sample of about 5 percent of housing units for each of the 3 years, or 15 percent for the 3-year period. (For budgetary reasons, the 3-year sample will be about 9 percent in five sites and 3 percent in one site.) Also, for each year from 2000 to 2002, there will be a nationwide survey, using the ACS questionnaire, of about 700,000 housing units.
Beginning in 2003, the full ACS sample will be 250,000 housing units each month for the rest of the decade, for an annual sample size of about 3 million housing units spread across all counties in the nation. Over a 5-year period, the addresses selected for the ACS sample will cumulate to about 15 million housing units, similar to but somewhat smaller than the 2000 census long-form sample size of about 18 million housing units.
Each month’s ACS sample will be drawn from the Census Bureau’s Master Address File (MAF) for the entire nation. The MAF is a comprehensive residential address list developed for the 2000 census that the Census Bureau intends to update on a continuous basis following the census. The current design calls for the ACS to use a sample design similar to that of the 2000 census long form, with higher sampling rates for small governmental units and lower sampling rates for large census tracts. The sampling rates would be applied by systematic sampling from the MAF.
Data collection in the ACS will be conducted by mailing a questionnaire similar to the census long form to all households in the sample. A replacement questionnaire will be mailed to nonresponding households about 3 weeks later. Then, after about another 3 weeks, nonresponding households will be contacted to the extent possible by the use of computer-assisted telephone interviewing (CATI). In the final stage of follow-up, a one-third sample of the remaining nonrespondent households will be drawn, and field representatives will be sent to interview these households in person, using computer-assisted personal interviewing (CAPI) techniques.
Responses were obtained from about 78 percent of the originally designated sample for the four initial ACS test sites—61 percent of occupied housing units responded by mail, another 8 percent responded to the telephone follow-up, and 9 percent responded to the personal follow-up. Because of subsampling at the final stage of follow-up, the weighted response rate in the four initial ACS test sites was over 95 percent.
Item nonresponse rates may be lower in the ACS than in the 1990 census, at least for some items, based on preliminary results from the 1996 ACS test sites (Salvo and Lobo, 1997: Tersine, 1998). On the other hand, the ACS, like other household surveys, may cover the population less well than the census, based on one analysis that found more small households and fewer large households in the 1996 ACS than in the 1990 census. This result could indicate that the ACS is
missing a larger proportion of people in interviewed households than are missed in the census (Ferrari, 1998).1
Publication plans for the ACS, once it is fully implemented, call for the Census Bureau to issue annual reports containing yearly averages of the monthly data for areas with 65,000 or more people. The Census Bureau also plans each year to publish 3-year averages for areas with 20,000–65,000 people and 5-year averages for areas with fewer than 20,000 people.
Although delivery schedules are not known with certainty, yearly averages from the full ACS should be available beginning within a year after the ACS is fully implemented in 2003 (i.e., in 2004). However, 3-year averages will not be available until 2006 at the earliest, and 5-year averages will not be available until 2008 at the earliest. Once sufficient years of data are cumulated to provide 1-, 2-, 3- or 5-year averages as appropriate, each set of averages will be updated yearly. The production goal is to deliver averages within 6 months after the close of a calendar year.
Additional information can be found at The Census Bureau website for the ACS: (http://www.census.gov/acs/www/).
The Current Population Survey is a voluntary monthly labor force participation survey, begun in the 1940s, that includes supplemental questions in many months. For the annual March income supplement, the CPS asks household respondents about income received during the previous calendar year, including income received from public cash assistance programs. The questionnaire also asks about noncash benefit receipt, including Medicaid coverage for household members, food stamps receipt and amount of benefits, energy assistance benefit receipt and amount, free and reduced priced school lunch program benefits for children in the household, and whether the household lives in public housing or receives a housing subsidy.
The monthly CPS sample, beginning in 1996, included about 50,000 households, or 1 in 2,000—a reduction in sample size of about 17 percent from the early 1990s. Part of the CPS sample is changed each month: in the rotation plan—under which each sampled address is in the survey for 4 months, out of the survey for 8 months, and in the survey for another 4 months—three-fourths of the sample addresses are common from one month to the next, and one-half are common for the same month a year earlier.
The CPS uses a multistage probability sample design, which is revised after
each decennial census. A design based on the 1990 census was phased in between April 1994 and July 1995: it included 792 sample areas consisting of about 1,300 counties, chosen to represent all 3,143 counties and independent cities in the 50 states and the District of Columbia.2
The CPS has a state-representative design, which results in larger states generally having larger CPS sample sizes, but with the largest states having CPS sample sizes that are smaller than their proportionate share of the U.S. population and the smallest states having proportionately larger sample sizes. For example, California, with 12.2 percent of the U.S. population, has 9.9 percent of the CPS sample; Wyoming, with 0.18 percent of the U.S. population, has 1.3 percent of the CPS sample. This sample design means that income and program participation estimates in large states are generally more precise than those in smaller states. The largest states, however, have larger relative errors due to sampling variability than would be expected if the CPS sample were allocated to the states in proportion to their population; the reverse holds true for smaller states.3 In fall 1999 the Census Bureau received an appropriation to adjust the March CPS sample size and design so that reliable annual estimates at the state level could be provided of the numbers of low-income children lacking health insurance coverage by family income, age, and race or ethnicity.
Data collection for the CPS is carried out by permanent, experienced interviewers. The first interview and fifth interviews at an address are usually conducted in person; the other six interviews at an address are usually conducted by telephone; CAPI and CATI are used. One household member who is aged 15 or older is allowed to respond for other members.
Response rates in the CPS are high, typically about 94–95 percent of households respond, though they declined by 1–2 percentage points beginning in 1997. each month. However, some interviewed households do not provide information for all members—for this reason, there is little data beyond basic demographic characteristics for about 9 percent of members of interviewed households. In addition, some people who respond to the basic CPS labor force questionnaire do not respond to the March income supplement. To adjust for whole household nonresponse to the basic CPS, the Census Bureau increases the weights of similar responding households. To adjust for person nonresponse to the basic CPS, it imputes a complete data record for another person with similar demographic characteristics.
Like other household surveys, the CPS exhibits population undercoverage at
In January 1996 the number of sample areas was reduced from 792 to 754.
To meet national-level reliability criteria for the unemployment rate, the sample size in a few large states (e.g., California, Florida, New York, Texas) is somewhat longer than what would be required by a state-based design (see the joint Bureau of Labor Statistics and Bureau of the Census CPS website: www.bls.census.gov/cps/mdocmain.html).
higher rates than the census. For March 1994, the ratio of the CPS estimated population to the census-based population control total (all ages) was 92 percent; for black men aged 30–44 years, the coverage ratios were as low as 67–68 percent (U.S. Census Bureau, 1996:Table D-2). It is estimated that about two-thirds of CPS undercoverage is due to missed people in otherwise interviewed households (i.e., people whose existence, let alone any information about them, is not known to the interviewer); the remainder is due to missed housing units because the address was not included in the sampling frame. CPS undercoverage is corrected by ratio adjustments to the survey weights that bring the CPS estimates of population in line with updated national population controls by age, race, sex, and Hispanic origin. Beginning with the March 1994 CPS, the population controls for survey weights reflect an adjustment for the undercount in the census itself. However, the ratio adjustments do not correct for other characteristics, such as income, on which the undercovered population might be expected to differ from the covered population in each adjustment cell.
There is substantial item nonresponse in the March income supplement. About 20 percent of aggregate household income is imputed (about the same percentage as in the census; see National Research Council, 1993:Table 3–6). Imputation techniques are used to provide values for people who fail to respond to the income supplement entirely, as well as for people who fail to answer one or more questions on the supplement.
Publication of detailed official income and poverty estimates from the CPS for the nation as a whole and population groups occurs each year about 6 months after data collection in March. Limited statistics are also published for states on the basis of 3-year averages.
Additional information can be found at The Census Bureau website for the CPS (http://www.bls.census.gov/cps/ads/adsmain.htm).
SURVEY OF INCOME AND PROGRAM PARTICIPATION
SIPP is a continuing voluntary panel survey begun in 1983. From 1983 to 1993, a new sample (panel) of households was introduced each February. Adult members of originally sampled households in each panel were followed and interviewed every 4 months for 32 months, although some panels had fewer than eight interview waves because of budget restrictions, and the 1992 and 1993 panels had ten waves and nine waves, respectively. The 1996 panel, begun in April, followed original sample adults every 4 months for 4 years. A new two-wave panel began in 2000, and a new 3-year panel will begin in 2001.
SIPP is focused on income measurement. The core questionnaire, administered at each interview wave, obtains monthly information on detailed sources and amounts of income from earnings and public and private transfer payments and information for the 4-month period on income from assets. In total, about 56 separate sources of cash income are identified together with benefits from 7 in-
kind programs. Additional detail on program participation and related topics (e.g., child care, health) is collected in various supplements (topical modules).
The SIPP sample covers the U.S. civilian noninstitutionalized population and members of the armed forces living off post or with their families on post. Sample sizes for the 1983–1993 panels varied from 12,500 to 23,500 originally sampled households per panel. The sample size for the 1996 panel was 37,000 originally sampled households; it included households in all states but was not designed to provide reliable estimates at the state level. The sample size for the two-wave 2000 panel was 11,000 households. The sample size for the 3-year 2001 panel is 37,000 households; another larger sized panel will begin in 2004. The 1996 sample included an oversample of addresses in which the residents had family incomes below 150 percent of the poverty level in 1989, based on information from the 1990 census. Proxy characteristics, such as housing tenure and family type, were used for oversampling addresses for which only short-form census information was available. In rural areas, some addresses were oversampled on the basis of 1990 census poverty-related characteristics for the census block in which they were located.
Data collection for SIPP is carried out by permanent, experienced interviewers. The first and second interviews and one interview in each subsequent year of a panel are conducted in person, using CAPI (computer-assisted personal interviewing). Other interviews are conducted by telephone, using CATI (computer-assisted telephone interviewing). Household members aged 15 or older are supposed to respond for themselves, but proxy responses from other householders are accepted. About 35 percent of interviews for adults in each wave are by proxy; over the life of a panel, 60–65 percent of adult sample members have at least one proxy interview (U.S. Census Bureau, 1998).
Response rates to the first wave of a SIPP panel are somewhat lower than CPS response rates: about 5–8 percent of eligible households in the 1983–1991 SIPP panels did not respond to the first interview wave and were dropped from the sample. The first wave nonresponse rate for households in the 1992 and 1993 panels was 9 percent. It was 8 percent for the 1996 panel. By wave 8, the cumulative household nonresponse rate in the 1983–1991 panels was 21–22 percent; in the 1992 and 1993 panels it was 25 percent. By wave 6 of the 1996 panel, the cumulative nonresponse rate was 27 percent. About three-quarters of household nonresponse is due to refusals, and one-quarter is due to losing track of sample household members who move (U.S. Census Bureau, 1998).
People who drop out of SIPP tend to differ from those who stay in the survey: attrition is more likely to occur among young adults, males, minority groups, never-married people, poor people, and people with lower educational attainment (see, e.g., Lamas et al., 1994). There is also evidence that the current noninterview weighting adjustments do not fully compensate for differential attrition across population groups (see, e.g., King et al., 1990).
Like the CPS and other household surveys, SIPP covers the population less
well than the census. Coverage ratios (survey population estimates divided by census-based population estimates) are similar for the CPS and SIPP. SIPP has lower item nonresponse rates than the March CPS: overall, only 11 percent of total regular money income obtained for calendar year 1984 from the first four waves of the 1984 SIPP panel was imputed, compared with 20 percent in the March 1985 CPS. The SIPP and March CPS imputation rates for 1984 for earnings were 10 percent and 19 percent, respectively; for public and private transfers, 12 percent and 21 percent, respectively; and for property income, 24 percent and 32 percent, respectively (Jabine et al., 1990:Table 10.8; see also National Research Council, 1993:Tables 3–4, 3–5).
Data processing for SIPP involves complex operations, particularly to produce calendar-year and longitudinal panel files. Historically, this has often resulted in delays of 1, 2, or more years between collection of data from an interview wave or all waves in a panel and release of data files and publications. There is no regular publication series for SIPP; publications are released on topics of interest, such as program participation, and include estimates for population groups for the nation as a whole.
Additional information can be found at The Census Bureau website for SIPP (http://www.sipp.census.gov/sipp/).
SURVEY OF PROGRAM DYNAMICS
The Survey of Program Dynamics (SPD) is a voluntary study being conducted by the Census Bureau under a requirement of the 1996 PRWORA legislation. The purpose of the SPD is to collect longitudinal data on the demographic, social, and economic characteristics of a nationally representative sample of the U.S. population so that overall evaluations of welfare reforms can be conducted. Congress mandated that the 1992 and 1993 SIPP panels continue to be followed so that the prereform characteristics and well-being of families would be understood. The data from the 1992 and 1993 SIPP panels give 3 years of a longitudinal baseline before the reforms in 1996 (1992–1994 for half the sample and 1993– 1994 for the other half; no longitudinal data from 1995 were collected). SPD will follow the 1992 and 1993 panels of SIPP participants over the years 1996–2001, meaning that, combined, SIPP and SPD will provide 9 years of panel data.
In 1997 there was the SPD Bridge Survey, based on a modified version of the March 1997 CPS questionnaire. A new core SPD questionnaire was developed for the 1998 survey (with the assistance of Child Trends, Inc.). The 1998 survey included a self-administered adolescent questionnaire and retrospective questions on the core topics of jobs, income, and program participation for all persons over the age of 15. The 1999 SPD included a module on child well-being, and the 2000 SPD included a children’s residential history module. Both the adolescent and child well-being questionnaire modules will be included in the 2002 SPD.
The SPD sample consists of all sample persons in the almost 38,000 households that completed all waves of the 1992–1993 SIPP panels (76% of the two
original SIPP samples). The sample size for the 1998 SPD was reduced to approximately 18,500 because of budget constraints. In subsampling the original households, the Census Bureau kept more low-income households and low income households with and without children in the survey than high-income households with and without children.
The SPD data collection is carried out by permanent, experienced interviewers. Computer-assisted interviews, by telephone and in person, are conducted for the questionnaire once a year in May and June. The adolescent questionnaire is a self-administered questionnaire.
Response rates for the Bridge Survey were about 82 percent (about 30,000). The 1998 SPD interviewed 89 percent of eligible households, and the 1999 survey had a response rate of 86 percent. The cumulative attrition rate is high: the beginning 1992 and 1993 SIPP panels had already lost 27 percent of the original SIPP panels, and through the 1999 SPD, the rate is approaching 50 percent. The Census Bureau is planning several steps to address the attrition problem, including interviewing a targeted sample of SIPP and SPD Bridge survey nonrespondents and offering cash incentives to these nonrespondents for completing a survey. Plans to link Social Security Administrative earnings records to SPD households to assess any effects of attrition and to look at employer-side variables have also been made. The Office of the Assistant Secretary for Planning and Evaluation is contributing funds for the Social Security records and SPD/SIPP analysis.
The Census Bureau has explored the degree to which reported data from the SPD differs from data reported from the March CPS, particularly information on program benefit receipt, income, and earnings. There are some statistical differences in measures of these items between these two surveys (Weinberg and Shipp, 2000).
Data from the 1997 survey and preliminary data from the 1998 survey are currently available. A longitudinal file with data from 1992 through the 1998 survey is scheduled to be released in the summer of 2001.
Additional information can be found at The Census Bureau website for SIPP (http://www.sipp.census.gov/spd/).
THE NATIONAL SURVEY OF AMERICA’S FAMILIES
The National Survey of America’s Families (NSAF) is part of the Urban Institute’s New Federalism Project, which is analyzing the devolution of responsibility for social programs from the federal government to the states. The survey is voluntary and is designed to document the well-being of children, their families, and adults under the age of 65 within and across states, as well as changes in the well-being of these populations over time. The survey questions collect data on many benefits programs, including AFDC; Social Security Insurance (SSI); food stamps; Women, Infants, and Children (WIC); and school lunches. The initial survey was conducted from February to November 1997, and the second-
round survey was fielded from February to October 1999. A third round is planned for 2002.
The NSAF sample includes approximately 1,800 families with children under age 18 in each of 13 states (Alabama, California, Colorado, Florida, Massachusetts, Michigan, Minnesota, Mississippi, New Jersey, New York, Texas, Washington, and Wisconsin) where intensive case studies of policies and implementation will be conducted, as well as a sample drawn from the balance of the nation. Together, the 13 states encompass more than half of the nation’s population and represent a broad range of fiscal capacity, child well-being, and approaches to government programs. About 950 families with incomes below 200 percent of poverty are included from each of the 13 states. These low-income households are oversampled because it is anticipated that the policy changes will most affect them. The sample also includes about 1,200 households without children under age 18 in each state. There is overlap between the 1997 and 1999 samples designed to reduce the variance of estimates. Because of the focus on low-income families, the sample includes families without telephones and uses a dual-frame design consisting of a random-digit-dialing component for telephone households and an area sample for households without telephones.
NSAF data collection is conducted primarily by telephone survey using CATI. The interviews average 25 minutes in length for a household without children and 40 minutes for a household with children. Questions are asked about one or two focal children per household, one under the age of 6 and the other between the ages of 6 and 17 years old. The respondent is the household member who is most knowledgeable about the selected children. In households without children under age 18, the respondent is randomly selected from among the adults under the age of 65.
The data collected in 1997 serve as a baseline against which changes can be measured from the 1999 data. In 1997, detailed information was obtained for more than 75,000 adults and 34,000 children in more than 44,000 households. The response rate in 1997 was 70 percent. In 1999, detailed information was obtained for more than 73,000 adults under age 65 and almost 36,000 children in more than 42,000 households.
Data from both the 1997 and 1999 surveys have been published, and data files from both survey rounds are accessible for public use. The Urban Institute has issued a report of the initial results from both rounds of the survey.
Additional information can be found at the Urban Institute’s website for this project (http://newfederalism.urban.org/nsaf/index.htm.)
THE PANEL STUDY OF INCOME DYNAMICS
The Panel Study of Income Dynamics (PSID) has been conducted since 1968 as a longitudinal survey of a representative sample of U.S. men, women, and
children and their family units. PSID is conducted at the Survey Research Center (SRC) of the University of Michigan’s Institute for Social Research. The survey is voluntary and emphasizes the dynamic aspects of economic and demographic behavior. Questions are asked regarding many benefit programs, including AFDC, SSI, food stamps, low income health services, and housing subsidies.
The sample for PSID has grown from 4,800 families in 1968 to 6,434 families in 1999, and is projected to grow to almost 7,400 in 2005. In 1968, it consisted of two independent samples—a cross-sectional sample and a national sample of low-income families. The cross-sectional sample was drawn by SRC as an equal probability sample of households from the 48 contiguous states designed to result in approximately 3,000 completed interviews. The second sample consisted of 2,000 low-income families who had responded to the Census Bureau’s Survey of Economic Opportunity (SEO). The SRC and SEO samples were combined to yield the PSID core sample. In 1990, 2,000 Latino families originally from Mexico, Puerto Rico, and Cuba were added.
Every year from 1968 through 1996, PSID interviewed and reinterviewed individuals from families in the core sample. In 1997, the interview schedule became biennial and the sample was changed in two major ways to keep the study representative of the U.S. population. First, the core sample was reduced from 8,500 families in 1996 to approximately 6,168 in 1997 by reducing the SEO subsample by two-thirds. However, 609 families headed by at least one African American and containing at least one child aged 12 or under were added back into the sample. Second, a refresher sample of post-1968 immigrant families and their adult children was introduced. The Latino sample of 2,000 families that had been added in 1990 was dropped after 1995, and a more representative sample of 441 immigrant families was added in 1997.
In 1997 a Child Development Supplement was added to the core data collection. The supplement interviewed children and parents of children aged 0–12 on a variety of topics concerning the cognitive, behavioral, and health status of the children, as well as measures of the children’s time use and the parents’ or caregivers’ time spent with the children.
From 1968 through 1972, data collection was conducted in face-to-face interviews with paper-and-pencil questionnaires. Since then, the majority of interviews have been completed by telephone. In 1993, PSID started using CATI; in 1999, 97.5 percent of the interviews were conducted by phone, all used CATI. As of 1997, PSID had collected information about more than 60,000 individuals, spanning as much as 30 years of their lives.
In 1968, 76 percent of sampled families were successfully interviewed. The response rate in 1969 was 88.5 percent, but interviews were attempted only with the heads of family units containing adults who were members of families interviewed in 1968. With a minor exception in 1990, no attempt has been made to recontact people who had been lost by attrition from previous years. Since 1969, annual response rates have ranged from 96.9 to 98.5 percent. However, when
attrition is taken into account, the response rate for individuals who lived in the original 1968 households was 56.1 percent as of 1988.
The PSID Data Center is one of the main sources of data dissemination for this survey. As of November 2000, the most recent data available through the PSID Data Center are 1997 early release data (see http://stat0.isr.umich.edu/psid/data-center/data-center.html). The most recent final release data available are from the 1993 survey. In February 2000, the 1999 data were added to the 1984, 1989, 1994, and 1999 wealth files, but as of November 2000 the most recent data in the other PSID supplemental data files are from the 1997 survey (see http://www.isr.umich.edu/src/psid/suppdata.html).
Additional information can be found at the PSID website (http://www.isr.umich.edu/src/psid/).
THE NATIONAL LONGITUDINAL SURVEYS OF YOUTH
The National Longitudinal Survey of Youth 1979 cohort (NLSY79) is one of a set of surveys of cohorts initiated by the U.S. Department of Labor to analyze the sources of variation in the labor market experience of the U.S. population. The first set of surveys, initiated in 1966, consisted of four cohorts, referred to as the older men, mature women, young men, and young women and are known collectively as the NLS original cohorts. The NSLY79 cohort is the fifth cohort. The NLSY79 is a voluntary longitudinal survey of men and women representative of all Americans born in the late 1950s and early 1960s.
NLSY79 gathers data in an event history format, collecting dates for the beginning and ending of important life events, such as employment, marital status, and participation in government assistance programs, including AFDC, food stamps, and cash assistance. It is conducted by the National Opinion Research Center for the Ohio State University Center for Human Resource Research under a contract with the Bureau of Labor Statistics.
The NLSY79 sample is nationally representative of men and women who were born in the years 1957 to 1964 and were living in the United States when the sample was selected in 1978. It does not represent people who were born in the years 1957–1964 and immigrated to the United States after 1978. Three independent probability samples were drawn to represent this population: (1) a cross-sectional sample designed to be representative of the noninstitutionalized civilian population of youth; (2) a supplemental sample that oversamples civilian Hispanic, black, and economically disadvantaged non-Hispanic, nonblack youth and, (3) a military sample designed to represent the population aged 18–21 serving in the military as of September 30, 1978. The original sample included 12,686 young men and women. The oversample of youth enlisted in the military was discontinued after 1984, and the oversample of economically disadvantaged whites was discontinued after 1990. With these two subsamples removed, 9,964 respondents remain eligible for interview. Hispanics and blacks have continued
to be oversampled. In 1986, the NLSY79 was expanded to include surveys of the children born to women in that cohort. The child survey is given on a biennial basis to all children born to NSLY79 mothers and includes cognitive, socio-emotional, and physiological assessments of each child. Demographic and development information are also collected for each child from either the mother or child.
Data collection for NLSY79 was annual until 1994, and biennial starting in 1996. Interviews were conducted in person or by telephone. Some data are collected for the respondent’s spouse as well as the respondent. The response rate for each round of the survey has been over 84 percent.
In 1997, a new cohort of young people aged 12 to 16 as of December 31, 1996, were surveyed. This new cohort of about 9,000 youth is the National Longitudinal Survey of Youth 1997 (NLSY97). The initial round of the annual longitudinal survey interviewed both the youth and the youth’s parent.
NLSY79 data publication occurs biennially. As of November 2000, the most recent data available were from the survey administered in 1998 (see http://stats.bls.gov/nlsdata.htm). Data from the first round of the NLSY97 were released in January 1999, data from the second round were released in May 2000, and data from the third round are scheduled to be released in May 2001.
Additional information can be found at the web site of the Bureau of Labor Statistics (http://stats.bls.gov/nlshome.htm).
Ferrari, P. 1998 1996 American Community Survey vs. 1990 Decennial Census-Household Size and Characteristics by Response Mode. Paper prepared for the American Community Survey Symposium. Bureau of the Census, U.S. Department of Commerce, Washington, DC.
Jabine, T.B., K.E.King, and R.J.Petroni 1990 Survey of Income and Program Participation: Quality Profile. Washington, DC: U.S. Department of Commerce.
King, K.E., S.Chou, M.K.McCormick, and R.J.Petroni 1990 Investigations of the SIPP’s cross-sectional noninterview adjustment method and variables. In Proceedings of the Survey Research Methods Section. Alexandria, VA: American Statistical Association.
Lamas, E., J.Tin, and J.Eargle 1994 The Effect of Attrition on Income and Poverty Estimates from the Survey of Income and Program Participation (SIPP). SIPP Working Paper Number 190. Available at www.sipp.census.gov/sipp/wp190
National Research Council 1993 The Future of the Survey of Income and Program Participation. Panel to Evaluate the Survey of Income and Program Participation. C.F.Citro and G.Kalton, eds. Washington, DC: National Academy Press.
1995 Measuring Poverty: A New Approach. Panel on Poverty and Family Assistance: Concepts, Information Needs, and Measurement Methods. C.F.Citro and R.T.Michael, eds. Committee on National Statistics. Washington, DC: National Academy Press.
2000 Small-Area Income and Poverty Estimates: Priorities for 2000 and Beyond. Panel on Estimates of Poverty for Small Geographic Areas. C.F.Citro and G.Kalton, eds. Committee on National Statistics. Washington, DC: National Academy Press.
Robinson, J.G., B.Ahmed, P.Das Gupta, and K.Woodrow 1993 Estimation of population coverage in the 1990 U.S. based on demographic analysis. Journal of the American Statistical Association 88(423):1061–1079.
Salvo, J.J., and A.P.Lobo 1997 The American Community Survey: Nonresponse Follow-up in the Rockland County Test Site. Paper prepared for the American Community Survey Symposium. Population Division, Department of City Planning, New York, NY. (March).
Tersine, A. 1998 Item Nonresponse: 1996 American Community Survey. Paper prepared for the American Community Survey Symposium. Bureau of the Census, U.S. Department of Commerce, Washington, DC.
U.S. Census Bureau 1996 Poverty in the United States: 1995. Current Population Reports, Series P60–194. Washington, D.C.: U.S. Department of Commerce
1998 Survey of Income and Program Participation Quality Profile 1998, 3rdEdition. Washington, DC: U.S. Department of Commerce.
Weinberg, D.H., and S.S.Shipp 2000 The Survey of Program Dynamics: A Mid-Term Status Report. U.S. Census Bureau. Available: http://www.sipp.census.gov/spd/workpaper/summary10.htm#N_1_