
An official website of the United States government
A longitudinal survey obtains information from the same respondents at multiple points in time. A cross-sectional survey, by comparison, obtains information from respondents at a particular point in time. Cross-sectional surveys that are repeated monthly, quarterly, annually, or with some other periodicity usually include a different sample of respondents each time they are administered. Many of the repeated cross-sectional surveys sponsored by BLS include a mix of new and experienced respondents each time they are administered, rather than selecting an entirely new sample for each administration.
Longitudinal surveys like the NLSY79 and NLSY97 are useful for studying changes that occur over long periods of time, such as the number of job changes or unemployment spells that people experienced over some segment of their lives, the number of times they moved to a different county or State, or the number of years in which their family income was below the poverty threshold. For a variety of reasons, cross-sectional surveys are not very useful for obtaining reliable information about such long-term changes.
Longitudinal surveys are also useful for examining cause-and-effect relationships. Cross-sectional surveys of the labor market have shown, for example, that workers who have been with their employers a longer period of time have higher earnings than workers who have less tenure with their employers. Cross-sectional surveys are not very useful, however, for determining whether longer tenure causes higher pay, higher pay causes longer tenure, or there is no cause-and-effect relationship at all. Longitudinal surveys have been used to examine whether the statistical correlation between tenure and earnings exists because workers become more productive as they gain seniority and are paid more for that higher productivity or, conversely, because highly paid workers tend to stay with their employers for longer periods, rather than seeking employment elsewhere.
NLSY97 variables (NLSinfo.org)
NLSY79 variables (NLSinfo.org)
NLSY79 Child and Young Adult variables (NLSinfo.org)
NLS Mature Women variables (NLSinfo.org)
NLS Young Women variables (NLSinfo.org)
NLS Older Men variables (NLSinfo.org)
NLS Young Men variables (NLSinfo.org)
No. The National Longitudinal Surveys are designed to provide information about changes that occur in people's lives over time. The cross-sectional surveys that BLS sponsors, such as the Current Population Survey of U.S. households and the Current Employment Statistics survey of nonfarm businesses and government agencies, are designed to provide monthly, quarterly, and annual historical estimates of labor market activity.
No. The National Longitudinal Surveys are designed to represent specific birth cohorts at the national level. The surveys cannot provide representative estimates for States, counties, or metropolitan statistical areas. NLS data files with geographic variables are available on a restricted basis for authorized researchers to use, but the permitted uses do not include producing estimates for States or other geographic areas. The permitted uses of NLS geographic variables include linking respondents with publicly available information about their local labor markets and communities, identifying instances of geographic migration, or estimating statistical variance more precisely by accounting for the effects of the geographic clustering in the NLS sample designs.
No. All of the National Longitudinal Surveys collect information about the industries and occupations in which respondents work, but the surveys are not suited for producing estimates of industry or occupational employment or wages. The National Longitudinal Surveys only represent specific population groups based on their year of birth, rather than the entire population across all age groups. Other BLS surveys, such as the Current Population Survey, Current Employment Statistics survey, Occupational Employment Statistics survey, and National Compensation Survey, are designed to provide estimates for specific industries or occupations.
Researchers can obtain NLS public-use data files and documentation for free at NLSinfo.org.
NLS User Services at the Ohio State University Center for Human Resource Research (CHRR) distributes a variety of NLS documentation designed to inform the research community on the content and current status of the surveys. They offer general publications, technical manuals, and user's guides that are available to the public. Those wishing more information on NLS documentation should visit the NLSinfo.org or contact: NLS User Services (E-mail: usersvc@postoffice.chrr.ohio-state.edu)
Many NLS publications are available online in Portable Document Format (PDF) at the NLS Publications Center Work and Family reports, NLS Discussion Papers, and several reports on youth labor are available. In addition, the quarterly NLS newsletter is also available in PDF at NLSnews.htm.
The NLSY79 public-use files that include data from round 1 (1979) through round 27 (2016-2017) are now available. The round 27 confidential files are also available.
The NLSY97 public-use files that include data from round 1 (1997-1998) through round 18 (2017-2018) are now available. The round 18 confidential files are also available.
The NLSY79 Child and Young Adult public-use files that include data collected from 1986 to 2014 are available for researchers to use.
The public-use files for the NLS of Older Men that include data from 1990 and all prior years, the NLS of Young Men that include data from 1981 and all prior years, and the NLS of Mature Women and Young Women that include data from 2003 and all prior years are available for researchers to use. No future collection of these cohorts is planned.
After the field period ends, data must be processed before it can be released. Information on field periods is found below:
NLSY97 NLSY79 NLSY79 Child and Young Adult
To protect respondent confidentiality, the NLS public-use files do not include geographic variables such as state, county, and metropolitan area. Such variables, when combined with the rich longitudinal records of respondents' significant life events, would create an unacceptable risk that someone could use the data to identify individual respondents. Instead, BLS has established a licensing system through which legitimate researchers at universities and other research organizations in the United States can use NLS data with geographic information at their own facilities, provided that the research project and physical and electronic security measures described in the NLS geocode application are approved by BLS.
NLSY79, NLSY79 Young Adult or NLSY97 Geocode files include the state, county, and metropolitan area of residence for each respondent in each survey year. Geocode data is also available for any of the Original Cohorts.
To protect the confidentiality of respondents, the Bureau of Labor Statistics (BLS) only grants access to geocode files for researchers in the United States who agree in writing to adhere to the BLS confidentiality policy and whose projects further the mission of BLS and the NLS program to conduct sound, legitimate research in the social sciences. Applications from abroad cannot be accepted. Applicants must provide a clear statement of their research methodology and objectives and explain how the geocode data are necessary to meet those objectives. Researchers who are granted access to NLS geocode files may use them at their own facilities, provided that the facilities meet BLS security requirements. More.
The geocode application document is available online at www.bls.gov/nls/geocodeapp.htm.
The Bureau of Labor Statistics (BLS) has opportunities available on a limited basis for researchers from colleges and universities, government, and eligible nonprofit organizations to obtain access to BLS data files not available on public-use or geocode files. These confidential files are available for use only through the BLS Restricted Data Access program on statistical research projects approved by BLS. The files may be accessed at the BLS National Office in Washington, DC or in a Federal Statistical Research Data Center (FSRDC). Access to data is subject to the availability of BLS and FSRDC space and resources. These data files include:
The Zip Code and Census Tract files:
Available for NLSY79 and NLSY97 respondents. Geocode data with zip code or census tract variables included.
The 1996 NLSY97 School Survey:
All public and private schools with a 12th grade in the 147 nationally representative primary sampling units (PSUs) used for the NSLY97 sample construction. School characteristics specifically targeted to gain information on school-to-work programs.
The 2000 NLSY97 School Survey: Sample of all schools in the original 1996 NLSY97 school survey. In addition, vocational education school in the PSUs are included in the sample. Where NLSY97 respondents have moved to secondary schools with a 12th grade outside the 147 PSUs, those schools also are included.
For more information about how to request access to these and other confidential data files, please see BLS Restricted Data Access at www.bls.gov/bls/blsresda.htm.
Please note there is a different process for obtaining confidential data about the original cohorts than the standard Geocode application used for the NLSY79, NLSY79 Young Adult, and NLSY97 cohorts. Information about the Original Cohorts Geocode data can be found at www.bls.gov/nls/request-restricted-data/request-original-cohorts-geocode-data.htm.
The NLS program takes its legal and ethical obligations to protect the confidentiality of respondents very seriously. Without the trust and cooperation or respondents, the NLS program could not continue to be such a rich source of data for researchers and policymakers. More information about protecting the confidentiality of NLS respondents is available at: www.bls.gov/nls/handbook/2005/nlshc9.pdf.
The respondent identification variable permits users to merge other variables for individual respondents. For instance, if a user forgets to select a variable needed for analysis or later decides to add more variables, the respondent ID allows the user to merge the newly extracted variables with those extracted previously. For this reason, users should include the respondent ID in their list of variables each time they extract NLS data.
The respondent ID appears in the first round of data for each NLS cohort. This is variable R00001.00 in each of the cohorts, except for the Child and Young Adult data files, where the respondent ID variables are C00001.00 (for children under age 15) and Y00001.00 (for young adults). A respondent's assigned ID number remains the same across all survey years.
Linking NLSY79 mothers to their children can be done by using the sequential identification variables for the mother and the child. To merge the files, save the child ID (C00001.00) and mother ID (C00002.00) from the NLSY79 Child and Young Adult file. Then save the mother ID (R00001.00) from the main NLSY79 file. The mother's ID will be the same in both files. The child ID is provided for all children regardless of their age in any survey year. The child ID is composed of the first 5 digits of the mother's ID plus a 2-digit code (01-11). This 2-digit code generally but not always indicates the child's birth order. The child ID allows users to link children not only with their mothers but also with any siblings on the NLSY79 Child and Young Adult file. Children with the same first 5 digits in their IDs have the same mother. Appendices E and F of the NLSY79 Child and Young Adult User's Guide provide sample SPSSx and SAS programs to assist users in merging files. See the NLSY79 Child and Young Adult User's Guide (PDF) on NLSinfo.org.
It is not necessary to link NLSY97 youth variables with parent variables because the parent variables already are embedded within the record for each youth. For example, looking only at survey year 1997, questions that begin with a "Y" were asked of youth respondents. Questions that begin with a "P" were asked of parents. The parent variables begin at variable R0541100.
When you set up the extraction, choose to generate SPSS statements. The extraction software will create an SPSS syntax file named filename.sps. This contains the data dictionary, variable names, and variable labels. The software creates another ASCII file with the data, which is called filename.dat. Use the syntax file to read the data into the Data Editor.
Go to the folder where the data were extracted and double-click on filename.sps. This will open two screens in SPSS: the Data Editor and the Syntax Editor. The Data Editor will be blank. In the Syntax Editor you should see the data dictionary, variable names, variable labels, some missing values command, and a descriptives command.
The first line of SPSS syntax is the File Handle statement and it tells SPSS where to find the filename.dat file. The handle statement looks something like this: file handle pcdat/name='default.dat'/lrecl=xxx.
Specify the path to the .dat file. Suppose the filename.dat file is on the C: drive, in a folder called NLSY79, then edit this line to: file handle pcdat/name='C:\NLSY79\default.dat'/lrecl=xxx.
On the tool bar of the Syntax Editor you will see the Run button. Click on this button and choose "All." SPSS will run the syntax file, which will read the data from the filename.dat file into the Data Editor. The results of the descriptives command will appear in the SPSS viewer. Once the data have been read into the Data Editor, you will be able to use all available SPSS commands.
Also, you may want to think about the missing values command. The SPSS code produced by the extraction includes a couple of lines telling SPSS the nonresponse codes (-1 thru -5) values are "missing," and SPSS will not treat the missing values as valid. The resulting frequencies will match the codebook. You may want to recode the missing values command to suit your purposes.
Currently, there are no plans to hold New User Workshops. Information about a workshop, if funded, will be found here in early spring of each year. Additional information would be sent out using the BLS e-mail information system. To subscribe please sign up for e-mail notifications using the orange "Subscribe to NLS Updates" box found at the left of every page.
A BLS news release published in August 2019 examined the number of jobs that people born in the years 1957 to 1964 held from age 18 to age 52. The title of the report is "Number of Jobs Held, Labor Market Activity, and Earnings Growth among the Youngest Baby Boomers: Results from a Longitudinal Survey." The report is available on the BLS web site at: www.bls.gov/news.release/pdf/nlsoy.pdf.
These younger baby boomers held an average of 12.3 jobs from ages 18 to 52. (In this report, a job is defined as an uninterrupted period of work with a particular employer.) On average, men held 12.5 jobs and women held 12.1 jobs. For additional statistics on the number of jobs held, see the tables at: www.bls.gov/nls/tables/news-releases-files/nlsy79/number-of-jobs-held-ages-18-to-52.xlsx .
One limitation of the NLSY79 is that it does not reflect the labor market behavior of people who are not in that particular cohort; that is, people who are older or younger than the baby boomers in the survey or who immigrated to the United States after the survey began in 1979.
Another way to examine job changing is with statistics on workers' tenure with their current employer. Such statistics for all workers age 16 and older are available from the Current Population Survey (CPS). For more information on CPS tenure data, see the web site at www.bls.gov/cps/home.htm. You also can send e-mail or call (202) 691-6378.
Each month, the Bureau of Labor Statistics (BLS) provides unemployment estimates derived from the Current Population Survey (CPS), a monthly cross-sectional survey. While the CPS can provide "snapshots" of labor market behavior, the National Longitudinal Survey of Youth 1979 (NLSY79) gives valuable insight into spells of unemployment, particularly repeated spells, that other surveys cannot identify. Longitudinal surveys track the same individuals over time and complement the information provided by cross-sectional surveys like the CPS, giving us a more complete picture of the labor market. The NLSY79 has tracked the labor market activities of younger baby boomers over a considerable segment of their lives. The survey includes a nationally representative sample of people born in the years 1957 to 1964 who were living in the United States when the survey began in 1979.
Estimates from the NLSY79 show that these baby boomers experienced an average of 5.8 spells of unemployment from age 18 to age 52. Consistent with CPS estimates that show higher unemployment rates for people with less education, the NLSY79 shows that people with less education experienced more unemployment spells over time. High school dropouts experienced an average of 8.1 spells of unemployment from age 18 to age 52, while high school graduates experienced 6.5 spells and college graduates experienced 4.2 spells. In addition, 34.4 percent of high school dropouts experienced 10 or more spells of unemployment, compared with 24.3 percent of high school graduates and 6.6 percent of college graduates. For more information on the number of unemployment spells that baby boomers in each level of educational attainment experienced from ages 18 to 52, see the table at: www.bls.gov/nls/tables/news-releases-files/nlsy79/percent-distribution-unemployment-spells-by-education-ages-18-to-52.xlsx.
Just as the CPS estimates show higher unemployment rates for people in their late teens and early twenties compared with those who are middle aged, the NLSY79 shows that younger baby boomers experienced fewer unemployment spells as they aged. On average, these baby boomers experienced 2.9 spells of unemployment from ages 18 to 24, falling to 1.6 spells from ages 25 to 24, 0.8 spells from ages 35 to 44, and 0.6 spells from ages 45 to 52. For more information, see table at : www.bls.gov/nls/tables/news-releases-files/nlsy79/percent-distribution-unemployment-spells-ages-18-to-52.xlsx.
The Bureau of Labor Statistics (BLS) never has attempted to estimate the number of times people change careers in the course of their working lives. The reason we have not produced such estimates is that no consensus has emerged on what constitutes a career change. A few examples may help to illustrate the difficulty of defining careers and career changes. Take the case of a BLS economist who is promoted to a management position. Before the promotion, she spent most of her time conducting economic research. After the promotion to the management position, she still may conduct research, but she also spends much more time supervising staff and reviewing their research, managing her program's finances, and attending to a variety of other management tasks. This promotion represents an occupational change from economist to manager, but does it also represent a career change? It depends on how you define a career change.
Did a construction worker who decided to start his own home-remodeling business experience a career change? What about a newspaper reporter who became a TV news anchor? Each of these examples involves a change in occupation, industry, or both, but do they represent career changes? Most people probably would agree that a medical doctor who quits to become a comedian experienced a career change, but most "career changes" probably are not so dramatic.
What about the case of a web site designer who was laid off from a job, worked for six months for a lawn-care service, and then found a new job as a web site designer? Might that example constitute two career changes? If not, why not? Is spending six months at the lawn-care service long enough to consider that a career? How long must one stay in a particular line of work before it can be called a career?
Until a consensus emerges among economists, sociologists, career-guidance professionals, and other labor market observers about the appropriate criteria that should be used for defining careers and career changes, BLS and other statistical organizations will not be able to produce estimates on the number of times people change careers in their lives.
Bureau of Labor Statistics (BLS) worklife estimates do not use NLS data. The last worklife estimates from the BLS were published in 1986 as Worklife Estimates: Effects of Race and Education, Bulletin 2254. (PDF 1.32 MB, 37 printed pages.)
From time to time, staff at the Bureau of Labor Statistics are asked something along the lines of the following question:
"I read in a recent article that, according to the Bureau of Labor Statistics, out of 100 people that start working at the age of 25, by the time they turn 65, 60 percent depend on Social Security or charity, 29 percent are deceased, 4 percent can afford to retire and 1 percent is wealthy. It goes on to say that 95 percent of people age 65 or older cannot afford to retire. I have been to the Bureau of Labor Statistics web site and have been unable to find the documentation for this information. Can you help me locate it?"
A brief search of the Internet does indeed turn up several references like this that are attributed generally to the Bureau of Labor Statistics, but none of these references ever cites a specific Bureau of Labor Statistics report or news release. Neither the Bureau of Labor Statistics nor any other agency of the U.S. Department of Labor has ever produced any statistics or reports that support the statement. The statement includes imprecise language and value judgments that would not meet Bureau of Labor Statistics quality standards. For example, the Bureau of Labor Statistics and other Federal statistical agencies do not define terms like "depend on Social Security," "afford to retire," and "wealthy."
Several of the assertions in the statement are incorrect or misleading. For example, research from the Social Security Administration shows that Social Security was the sole source of income for 21 percent of "units" age 65 or older in 2012, although Social Security accounted for at least half of total income for 57 percent of units age 65 or older (See Table 8.A1). (The report defines a unit age 65 or older as either a married couple living together and at least one spouse was age 65 or older or an unmarried person age 65 or older.) On average, Social Security accounted for 35 percent of total income in 2012 for units age 65 or older (see table 10.1). See the report at www.ssa.gov/policy/docs/statcomps/income_pop55/.
Statistics from the National Center for Health Statistics show that, of the people who lived to be age 25, about 85 percent of them reached age 65. In other words, the death rate is about 15 percent, not 29 percent. See the National Center for Health Statistics web site at www.cdc.gov/nchs/deaths.htm.
Last Modified Date: January 16, 2020