NLSY79

The Employer History roster includes information on virtually all employers reported by NLSY79 respondents, with many of the employer characteristics reported for each employer included in a single record. These variables are classified as "XRND" variables, rather than being assigned to a single survey year. Data from each survey year are currently available. The variables will be updated as necessary as revisions are made and with each progressive round.

Structure

In earlier years, the linking process required to build a record of various commonly used employer-specific characteristics that have been reported in different survey years has involved a rather complex process. Researchers would need to use variables indicating the employer number in the previous survey year and establish the link through each preceding year individually. Appendix 9: Linking Employers Through Survey Years details this process.

By contrast, the Employer History roster is designed to help alleviate that more involved process of employer linking for many of the most commonly used employer characteristics. The Employer History data is constructed in a roster structure, consisting of one record for each of the up to 65 jobs (so up to 65 records). Many employer characteristics have been compiled into each employer's record. The record for each employer will include each of the variables listed in the table below. For instance, the record for employer #1 will contain EMPLOYERS_ALL_CPSJOB_[YEAR].01, EMPLOYERS_ALL_UNION_[YEAR].01, EMPLOYERS_ALL_IND_[YEAR].01, EMPLOYERS_ALL_OCC_[YEAR].01, etc. for each survey year that employer #1 was reported.

Data notes

Employer information for additional employers (jobs 6-10) in some of the older survey years may not have been included in the roster. These jobs were often included on separate data tapes, the original data of which has been difficult to recover. Should any of these data on additional employers be recovered in the future, that data will be added to the Employer History roster accordingly and users notified of these additions. These employers comprise a very small proportion of those ever reported for a very small number of respondents.

Locating Employer History variables

The Employer History items can be easily located within the NLSY79 dataset on NLS Investigator by using one of the following search criteria:

• Question Name (enter search term) index, "starts with", enter "EMPLOYERS_ALL" search term
• Area of Interest (pick from list) index, "equals", choose areas of interest beginning with "EMPLOYERS_ALL"
• Reference Number (enter search term), "starts with", enter "E"

Table 1 lists the contents of the Employer History roster. The variables are listed under the Areas of Interest currently assigned. The "variable names" column contains the qnames for each of the variables. Currently there are up to 65 jobs for each respondent included on the roster. This means that at least one respondent has reported 65 jobs over the course of the NLSY79's 27 rounds. Most respondents of course will have a much smaller number of jobs reported since 1979.

Beginning with the release of the 2000 data, the Work History file is incorporated into the main NLSY79 data set. Variables previously located on the separate Work History file can be identified by searching for areas of interest beginning with "Work History." In addition, the reference numbers for work history variables begin with "W." Key work history variables are described below.

Weekly arrays

Week-by-week records of the respondent's labor force status and associated job(s), if employed, and the total number of hours worked each week at any job, if employed, are available. This information is contained in the three arrays described above.

Although data on only up to five jobs are released, data are collected on all jobs. Data for the extra jobs are used to construct summary KEY variables by the work history programs. The number of jobs has exceeded ten for one case in 1991 and 1992, two cases in 1998, and one case in 2000.

Many researchers focus on data for the CPS job. Because the CPS was Job #1 only in select years, researchers should see the "Important Information" box in the Labor Force Status section for an elaboration of this concept.

Employment gaps

Gaps within tenure with a specific employer are reported in association with that employer. They occur between the start and stop dates given for an employer. The respondent does not consider himself/herself completely disassociated from the relevant employer during these periods, although he or she was not actively working for that employer. Specific variables for each gap include start and stop dates; the reason that the respondent was not working; the number of weeks that a respondent was unemployed (looking for work or on layoff) or out of the labor force (OLF or not looking for work); and, for those who were OLF at some time during a gap, the reason they were not looking for work. See the Work Experience section for a discussion of gaps with respect to job tenure.

Gaps between employers are gaps in a respondent's employment during which he or she was not associated with any employer. The specific variables collected with respect to "within job gaps" (see the discussion in the Work Experience section on tenure with a specific employer) are also collected with respect to gaps between employers, with the exception of the reason that the respondent was not working during the gap.

The information collected on reasons for employment gaps allows specific dates to be fixed for unemployed or OLF status only if a respondent was unemployed or OLF for the entire period of the gap. If the respondent was unemployed for part of the gap and OLF for the other part, the number of weeks unemployed and OLF is recorded, but the specific dates of periods for which the respondent was actively looking for work/on layoff and not looking for work are not collected. This prevents the Work History program from assigning specific week numbers to these statuses in the event of such a "split gap." Instead, the number of weeks reported as unemployed is assigned to the middle of the total gap period, with the remainder of weeks at the beginning and end of the gap period being assigned an OLF status. Users examining the week-by-week status array containing labor force statuses should be aware that "split gaps" will appear as a series of "5" codes, followed by a series of "4" codes, followed by another series of "5" codes (5 5 5 5 5 .... 4 4 4 4 4 .... 5 5 5 5 5). Although the start and stop dates for the whole gap will be those actually reported by the respondent, the assignment of the unemployed and OLF statuses will not represent actual dates reported by the respondent. They represent only the number of weeks that a respondent reported having held each status, with the unemployed status being arbitrarily assigned to the middle portion of the gap.

Summary labor force related variables

Variables are constructed summarizing different aspects of a respondent's labor force activity, including total number of hours worked, weeks worked, weeks unemployed, weeks out of the labor force, and weeks in active military service. There are two sets of these variables, referring to each of two time periods--the period since the last interview and the past calendar year (see the Labor Force Status section). Variables are also created indicating the number of weeks since the previous interview and the percent of weeks for which a definite status cannot be determined in constructing the summary variables discussed above. See the Work Experience section for further notes on these variables.

Tracing employers back through contiguous survey years

Of particular interest to many researchers have been the PREV_EMP# and TENURE variables associated with each employer. The PREV_EMP# allows a respondent's association with a given employer to be traced back through contiguous survey years. Using PREV_EMP# and the appropriate start and stop dates, a TENURE variable is constructed for each job reported, which depicts total weeks of tenure with each employer across contiguous survey years. Examine the work history documentation in Appendix 18 of the NLSY79 Codebook Supplement to determine if any such time-saving variable constructions exist. 

Creation of the Work History data

The work history is a complete retrospective up to and including the respondent's most recent date of interview. The questions in these survey sections are constructed to collect a complete history for each respondent, regardless of period of noninterview. For example, a respondent previously interviewed in 1984 and not interviewed again until 1989 will have a complete labor force history as of the 1989 interview, as information for the intervening period will be recovered in the 1989 interview. The Work Experience section contains a discussion of possible discrepancies or inconsistencies in these data. Researchers should be aware that, although such possibilities exist, they have not appeared to be a major factor in the quality or completeness of the work history record.

Be aware that for respondents with simultaneous active military status and civilian employment status, civilian labor force activity will take precedence over military status. For the purposes of constructing the week-by-week status array, the civilian job number will replace the military status code for weeks in which both statuses occur. The order of precedence for various labor force status codes is detailed in the work history documentation (see the discussion of the work history PL/I program in Appendix 18); see also the Work Experience section.

For purposes of constructing the status array and computing the summary labor force activity variables, the work history programs require that specific week numbers be assigned on the basis of the job-specific start and stop dates. In the event that missing data occur in the job-specific start and stop dates, the programs take one of two actions. 

1. If only the day in a given date is missing, the program assigns the number "15," placing these dates in the middle of the month. This allows an approximate week number to be assigned. The possibility still exists, however, that a negative job/gap duration will result because the day is arbitrarily fixed. For example, a start date of 10/-2/90, which indicates a missing day, and a stop date of 10/6/90 would be read by the work history program as 10/15/90 and 10/6/90 respectively. Therefore, when the week numbers are assigned, the arbitrary assignment of "15" as the start day would give an erroneous impression that a job started after it stopped. The status array and computed summary variables will reflect the invalid data in the week numbers.
2. Dates missing a month or year cannot be estimated by the work history program and therefore have invalid missing codes for the week numbers. The status array and other computed variables cannot be calculated for activity within periods for which either or both of the dates have such missing information. These will also register invalidly missing information for any period in which specific dates and week numbers cannot be determined.

Data on respondents' usual earnings (inclusive of tips, overtime, and bonuses but before deductions) have been collected during every survey year for each employer for whom the respondent worked since the last interview date. The amount of earnings, reported in dollars and cents, is coupled with information on the applicable unit of time, such as per day, per hour, per week, or per year. Since 1988, those respondents reporting any unit of time other than "per hour" have been asked a follow-up question on whether they were paid by the hour on that job; if so, an hourly wage rate was collected.

The raw earnings data, collected in the Employer Supplements during each round of the survey and in Section 10 of the 1979 questionnaire, can be found in the variable series 'Rate of Pay Job #1-5' and 'Time Unit of Rate of Pay Job #1-5.' Two sets of variables provide information based on the combined earnings and time unit data. The first set, 'Hourly Rate of Pay Job #1-5,' provides the hourly wage rate for each job as reported. The actual responses of those respondents who report wages with an hourly time unit in the initial earnings question appear in this variable. For those reporting a time unit other than "per hour" or "other" in the initial earnings question, an hourly rate of pay has been calculated. 

A second set of variables based on responses to the initial set of wage/time unit questions, entitled 'Hourly Rate of Pay Current/Most Recent Job,' identifies the hourly earnings for the job identified as the CPS job, that is, the job that the respondent held most recently. Hourly wage rates for those respondents who reported a time unit other than "per hour" can be found in the 1988-93 variables series, 'Paid by the Hour (Time Unit Other than Hourly Previously Reported) Job #1-5' and 'Hourly Rate of Pay (Rate Other than Hourly Previously Reported) Job #1-5.' Table 1 depicts the core set of rate of pay variables present on the NLSY79 combined Main/Work History Data.

Follow-up questions starting in 1986 asked those respondents whose earnings had changed for wage rate and time unit information at the time they first started working for a new employer. In 1986 and 1987, those who were not working for the employer at the interview date were also asked for wage information at the time they left that employer. These data can be found in the following variables: 'Wages Changed Since First Began Working Job #1-5,' 'Rate of Pay When 1st Began Working at Job #1-5,' 'Time Unit of Rate of Pay When 1st Began Working at Job #1-5,' 'Rate of Pay When Last Worked at Job #1-5,' and 'Time Unit of Rate of Pay When Last Worked at Job #1-5.'

Values for wages, business income, bonuses and pension values reported in the Employer Supplements in the 2022 survey have been topcoded for the first time, using a new set of decision rules. These rules were formulated based on the two percent method. There were additional modifications for the treatment of a variety of time units for rate of pay, extreme outlier values and very low frequency variables. [add link]

Time and tenure questions provide information on a respondent's time spent with an employer, time spent away from an employer during which the employment contract was maintained or renewed, and periods of time when the respondent was neither working for an employer nor serving in the active forces.

• Hours worked
• Shift and times worked
• Periods working for an employer including unpaid/paid leave
• Total tenure with employer
• Periods not working or in the military

Hours worked

During each survey, information is collected on:

• the number of hours during the survey week that a respondent worked at all jobs
• the number of hours per day and week usually worked at the CPS job; and
• the usual hours per day and per week worked at the non-CPS Jobs #1-5

Post-1987 surveys differentiated, for each job, the number of hours worked at home from the hours worked at the place of employment. Follow-up questions double check that these at-home hours are included in the total hours reported working. The creation procedures for the 'Hourly Rate of Pay 'variables factor in all values.

Shift and times worked

Data on the type of shift usually worked at the respondent's current or most recent (CPS) job, such as "day shift," "evening shift," or "split shift," have been collected during all but the 1986 and 1987 surveys. Beginning with the 1986 survey, information on the actual times (24 hour clock) that a respondent usually began and ended the CPS job is available. Coding categories for the 1988 and 1989 shift variables were collapsed. All respondents were asked a follow-up question beginning in 1990 on the actual or usual time work began and ended during the last week (or most recent week) they worked.

Periods working for an employer including unpaid/paid leave

Periods of time that a respondent worked for a specific employer can be identified through information collected on dates that employment began and ended. Those respondents with a stop date for an employer are asked for information on the reason the job ended and whether a new job was lined up before the respondent left this job. Interrupted employment with an employer of one week or more during which the respondent was either on unpaid leave or unpaid vacation is also detailed. The total number of such separate within-job gaps is identified, along with the specific start and stop dates and the main reason for not working, such as "on strike," "on layoff," "quit job but returned to same employer," "job ended but began again," "attending school," "health problems," "childcare problems," "pregnancy," and so forth.

Information also is available on whether the respondent was looking for work during "some," "none," or "all" of the weeks during a period of not working. For those respondents who indicate that "some" weeks were spent looking for work or on layoff, information is available on the number of weeks duration for each period of unemployment or layoff, the total number of weeks spent looking and not looking, and the main reason he or she was not looking for work. Users should note that the specific week numbers during which the respondent was unemployed versus out of the labor force within a single gap are not available if both statuses apply. Information is collected only on the total number of weeks that were spent in either of the two nonemployed labor force statuses. For example, an eight-week within-job gap that occurred between week number 152 and week number 160 may identify two of those weeks as weeks of unemployment and six of those weeks as out of the labor force. It will not designate the specific week numbers 154 and 155 as the precise two weeks that the respondent was unemployed as opposed to out of the labor force.

During post-1987 surveys, female respondents (only) are asked for information on the total number of separate periods of paid leave from an employer which were taken due to either pregnancy or birth of a child. Start and stop dates are collected for each period of leave.

Types of information collected about periods of paid and unpaid leave are summarized in Table 1.

Total tenure with employer

The variable series 'Total Tenure (in Weeks) with Employer as of Interview Date Job #X' is created from

• the start (or last interview) and stop dates for an employer
• a match, if any, of employers identified during the current interview as the same employer from a previous interview year

Tenure in weeks with an employer is first calculated for the period between the date of last interview and the current interview date (tenure since the date of last interview). Next, a match with employers reported during the previous interview is attempted. If such a match is established, the total tenure in weeks with the matched employer from the previous interview is added to the total weeks tenure with that employer since the date of last interview. This creates a cumulative tenure in weeks with that employer from the time the employer was first reported up to the most recent week worked. Tenure with employers for whom no match exists in the previous interview is simply the tenure with that employer since the date of last interview. Further information on matching employers can be found in Appendix 9: Linking Employers through Survey Years and in Appendix 18: NLSY79 Work History Data included in the NLSY79 Codebook Supplement.

Tenure is cumulative through contiguous survey years, beginning on the first date the respondent reports working for the employer (the start date) and ending on the date the respondent reports leaving the employer (the stop date). If a respondent started working for an employer before or on the date of last interview, tenure since the date of last interview is added to the already existing time to reflect total tenure. If the respondent is working for the employer on the current interview date, the stop date is set to the current date. If the respondent reports working for this employer at the next interview, tenure between interviews is added to the total tenure figure.

For example, imagine the following scenario. A respondent starts working for COMPANY A on July 7, 1989, and is interviewed on July 31, 1989. She is still working for that company on August 1, 1990, when she is interviewed again. She continues to work for the same company until May 15, 1991 when she leaves. She is interviewed again on August 15, 1991. A cumulative tenure in number of weeks would be constructed in three stages:

• T1 = [Tenure from July 7, 1989 - July 31, 1989 at the 1989 interview]
• T2 = T1 + [Tenure from August 1, 1989 - August 1, 1990 at 1990 interview]
• T3 = T2 + [Tenure from August 2, 1990 - May 15, 1991 (time worked for COMPANY A between 1990 and 1991 interview, before respondent left)]
• TENURE WITH COMPANY A (at 1989 interview) = T1
• TENURE WITH COMPANY A (at 1990 interview) = T2
• TENURE WITH COMPANY A (at 1991 interview) = T3

Periods not working or in the military

During each survey, information is collected on up to six periods since the date of last interview during which a respondent is not affiliated with an employer or in the active military. Data available for each period of not working include month, day, and year that the period began and ended; the specific week numbers associated with the period of not working; whether any of these weeks (coded as "none," "some," or "all") was spent looking for work or on layoff; and, the number of weeks spent looking for work or on layoff, the number of weeks the respondent was out of the labor force, and the main reason that the respondent was not looking for work. All respondents with between-job gaps are asked the questions on periods of not working.

Regularly fielded sections of NLSY79 instruments have collected information on the occupation of respondents' current/last job, jobs since last interview, military job, vocational/technical or government training programs, type of job to which they aspired, and, for those unemployed and out of the labor force, the kind of occupation they were seeking or planned to seek. In addition to this respondent-specific information, data on occupations are also available for other family members, including the spouse and parents of the respondent. Finally, the 1980 school survey collected data on the types of vocational/technical training offered within a respondent's high school.

Verbatim responses to open-ended questions eliciting information on kinds of work or training are entered directly into the survey instrument by the interviewer and subsequently coded using one or more occupational coding schemes. The occupational classification systems listed in Figure 1 have been used to code occupations within the NLSY79. Background information on the development of the 1980 classification system and the relationship between the 1970 and 1980 coding categories is available in the 1989 Census publication listed in this section's references.

References

Duncan, O.D. "A Socioeconomic Index for All Occupations." In Occupations and Social Status, A.J. Reiss, Jr. et al. New York: Free Press, 1961.

U.S. Census Bureau. 1970 Census of Population: Alphabetical Index of Industries and Occupations. Washington, DC: U.S. Government Printing Office, 1971.

U.S. Census Bureau. 1980 Census of Population: Alphabetic Index of Industries and Occupations. Washington, D.C.: U.S. Government Printing Office, 1981.

U.S. Census Bureau. The North American Industry Classification System (NAICS) [standard used by Federal statistical agencies in classifying business establishments for the purpose of collecting, analyzing, and publishing statistical data related to the U.S. business economy]. Retrieved from https://www.census.gov/naics/ 2007.

U.S. Department of Defense. "Occupational Conversion Manual: Enlisted/Officer/ Civilian, Defense Manpower Center." Arlington, VA: DOD 1312.1-M, 1977.

The NLSY79 is unique in that respondents are chosen from both the civilian and military populations. Since most surveys focus on one or the other, researchers rarely can compare outcomes simultaneously for both groups. Funding by the U.S. Department of Defense in the early years of the survey and continued interest by BLS has enabled the NLSY79 to collect a large amount of data on military occupations, training, wages, and testing scores.

Although funding cutbacks reduced the size of the military sample in 1985, military questions continue to be a part of every round of the NLSY79. Researchers will find that the questionnaires from 1979 to 1985 contain substantial information on military experience. While questionnaires from 1986 on contain less information, the interviews continued to ask respondents about key variables such as military enlistment, pay, and training. Researchers should note that respondents age 16 and under at the 1979 interview were not asked any military service questions; this group was asked three questions concerning attitude toward military service and the possibility of enlisting in the future.

NLSY79 military members consist of two groups. The first group is a special oversample of members of the Armed Forces. This group, which in 1979 included 1,280 respondents, was reduced to 201 respondents in 1985 because of funding cutbacks. The second group consists of NLSY79 respondents who joined the military while part of the sample. For example, in 1979 (R00431.) 508 respondents stated that they would "definitely try to enlist in the Armed Forces in the future." 

Table 1 shows the number of NLSY79 respondents who are in the active military by year and the number of individuals who have enlisted in any branch of the service since the last interview. Researchers need to understand the difference between active and reserve duty. Large portions of the "Military" section are either skipped or answered depending on a respondent's active or reserve duty status. Many people believe that active duty personnel are in full-time military jobs while reserve duty are part-time military jobs, but this is not the case. While many reservists serve two weeks a year, a number of reservists are employed full time, year round by the Armed Forces. A more complete picture of military service is gained by examining data on both active and reserve personnel.

The NLSY79 contains more than 1,600 variables pertaining to life in the Armed Forces. The following sections explore some of these variables in more depth. Researchers should note that military information can be combined with other NLSY79 data to provide useful insights into residence characteristics, marital status, fertility, and schooling while an individual serves in the Armed Forces.

Military occupation

Questions about military occupations were asked in 1979-1985. In each year, respondents in the military were asked to report their primary military occupation (for example, R16324. in 1985) and their secondary occupation (R16337. in 1985). These questions were part of a section that also asked for information about how much training the respondent had recently received. Researchers who use the Census Bureau's 3-digit occupation codes should note that while military occupations also are coded with 3-digit codes, the classification scheme is completely separate. CHRR coded military occupations with codes developed by the Department of Defense (1977). Researchers interested in a respondent's Military Occupation or Specialty (MOS) should search for MOS for Army, Marine Corps, and National Guardsmen. Respondents in the Navy or Naval Reserves are classified by their "Primary Rating," while Air Force and Air Force Reserves respondents are classified by their "Air Force Specialty Codes (AFSC)."

Military training

One focus of the NLSY79 military section is training. In 1980 members of the active Armed Forces were asked why they enlisted in the military (R02516.). The most important reason cited by the majority (217 respondents out of 993) was "To get trained in a skill that will help me get a civilian job when I get out." To understand military training, surveys prior to 1986 asked each respondent about the primary and secondary job for which they were trained.

Pre-1986 surveys also asked military respondents about the number of weeks of formal training received in the military, the amount of on-the-job training, and the amount of formal schooling. Each survey also contains two questions that explore the usefulness of military training for civilian life. One question asks if the respondent is doing the same kind of work in civilian life as in the military; the second asks if the respondent uses any skills learned in the military in any civilian jobs. Researchers interested in more details on how military training is transferable to civilian work should see Mangum and Ball (1986).

Military pay and bonuses

The NLSY79 contains a large amount of information on military pay and bonuses. During the early years of the survey, pay information was collected for individuals in the military, individuals in the reserves, and individuals who had separated from the military. Additional information was gathered on the amount of educational and enlistment benefits received. Finally, for individuals who left the Armed Forces, some interviews contain information on the primary reason for separation. A number of respondents stated low pay as their primary reason for leaving the military. Table 2 summarizes pay variables for members of the military.

ASVAB administration

During the summer and fall of 1980, NLSY79 respondents participated in an effort of the U.S. Departments of Defense and Military Services to update the norms of the Armed Services Vocational Aptitude Battery (ASVAB). For more details on the AFQT and ASVAB, refer to the Aptitude, Achievement & Intelligence Scores section.

Reference

Mangum, Stephen and Ball, David. Occupational Skill Training and Transferability: How Does the Military Fare? Proceedings of the Thirty-Ninth Annual Meeting, Industrial Relations Research Association (1986) 412-21.

U.S. Department of Defense, Manpower Reserve Affairs and Logistics. Occupational Conversion Manual. Alexandria, VA: D.M.D.C., 1977.

Date of interview labor force status

A series of date of interview (DOI) status variables (listed at the top of this page) has been created for survey years 1980-2022. They are described in more detail in NLSY79 Appendix 29. The absence of the CPS section in the 2000–2004 and survey years since 2008 prevented the creation of a strictly comparable set of ESR variables for all survey years. The date of interview variables (DOI_EMPLOYED, DOI_HOURS_WORKED, DOI_RETIRED, DOI_DISABLED) have been created in order to provide a consistent constructed set of variables from comparable data to represent the respondent's status at each date of interview.

NLSY79 CPS sections

The CPS sections of the NLSY79 instruments replicate the questions asked in the monthly Current Population Survey (CPS) of American households conducted by the U.S. Census Bureau for the U.S. Department of Labor. The primary purpose of the monthly CPS is to collect up-to-date information about the number of persons in the country who are employed, unemployed, or not looking for work during a given survey week. Results from the CPS surveys, released in the Bureau of Labor Statistics' monthly publication Employment and Earnings, provide detailed information classified by age, sex, race, and a variety of other characteristics, on the employment and unemployment experiences of the U.S. population. In the NLSY79 data, three series of variables provide information on respondents' current and weekly labor force status. A series of variables titled "ACTIVITY MOST OF SURVEY WEEK" (present for 1979-1993) and "EMPLOYMENT STATUS RECODE" (present for 1979-1998 and 2006) reflects the respondent's labor force status during the week before the survey week. The WEEKLY LABOR STATUS array variables contain the labor force status for each week since January 1, 1978. These point-in-time indicators are complemented by a set of summary measures, which provide a count of the total number of weeks a respondent occupied a given labor force status (such as working, unemployed, out of the labor force, in the active Armed Forces). Data are also available on the work experience of respondents' parents, spouses, and other household members.

Survey week labor force status

The following sets of variables on each respondent's labor force status during the week preceding the survey week are available for the survey years as specified:

Activity Most of Survey Week (1979-1993)

The 'Activity Most of Survey Week' variables reflect each civilian respondent's reply to the survey question "What were you doing most of last week?" "Last week" refers to the full calendar week (Sunday through Saturday) preceding the date of interview. The following seven categories of responses have been coded from each year's survey: 

1. working
2. with a job-not at work
3. looking for work
4. keeping house
5. going to school
6. unable to work
7. other

Definitions for each of these activities are intended to be consistent with those used in the CPS survey. Figure 1 provides the set of instructions provided during one survey year to NORC interviewers for coding respondents' labor market status, while Figure 2 gives definitions of key CPS labor force concepts. The main survey week activity question was followed by a second question, which sought to identify those respondents who did some work in addition to a main activity other than working (such as "looking for work" or "going to school"); this follow-up question was asked of all respondents except those who indicated that they were working or were unable to work. It is to these two groups, those whose primary activity during the survey week was working and those who responded that they indeed did some work even though it was not their primary labor force activity, that the series of questions on hours worked was administered.

Work for Pay or Profit Last Week (1994-1998 and 2006)

Beginning in 1994, the CPS section underwent significant changes as outlined later in this chapter. The 'Activity Most of Survey Week' question was replaced with "Last week, did you do any work for pay or profit?" Respondents can answer yes, no, retired, disabled, or unable to work. For those answering no, follow-up questions determine whether the respondent has a job from which he or she is temporarily absent, such as on vacation, sick leave, labor dispute, military duty, and so forth. Follow-up questions also probe for more information about disabilities or other reasons a respondent is unable to work. Note that the previously used follow-up question no longer exists that sought to identify those who did some work in addition to a main activity other than working.

Employment Status Recode (1979-1998 and 2006)

Created variables called 'Employment Status Recode' (ESR) are available for each survey year through 1998 and again in 2006. These variables recode the survey week activity of civilian NLSY79 respondents into labor force status measures consistent with those developed for the CPS. This recalculation not only factors in the respondent's reported survey week activity but also takes into account variables such as hours worked, whether working for pay, whether looking for work, what the respondent is doing to find work, whether and why he or she was absent from a job, and so forth. Added to the seven labor market status categories listed in (1) above is an eighth category, "in the active forces." The algorithm changed in 1994 in an attempt to match the new CPS algorithm as closely as possible.

Employment Status Recode-Collapsed (1979-1998 and 2006)

A collapsed version of ESR is available that classifies all NLSY79 respondents into one of the following four labor market activity categories:

• employed ("working" or "with a job not at work")
• unemployed
• out of the labor force ("keeping house," "going to school," "unable to work," or "other")
• in the active forces

CPS changes in 1994

Beginning in 1994, the "Current Labor Force Status" (CPS) section was changed to ensure that the NLSY79 matched changes that occurred in the Current Population Survey. This survey underwent a major revision in January 1994, thus causing a revision of the corresponding NLSY79 section.

The Census Bureau and Bureau of Labor Statistics revised the national CPS for four major reasons: 

• Because the last major CPS revision occurred in 1967, research suggested that the wording of many CPS questions was dated and response lists no longer reflected typical answers. For example, the old set of CPS responses did not have childcare problems on the list of reasons why a respondent was absent from work in the last week.
• In 1979, the National Commission on Employment and Unemployment Statistics, or Levitan Commission, had suggested a number of changes to U.S. labor force classifications. The revised CPS implemented many of the recommendations, such as tightening the definition of discouraged workers.
• Research in survey methodology suggested better ways of asking questions. For example, a new question was inserted before occupation and industry questions that checked if the respondent changed jobs or employers since the last survey. This extra question dramatically reduced the number of spurious job changes recorded.
• Advances in computer technology made it desirable to switch from a paper-and-pencil instrument (PAPI) to a computer-administered instrument (CAPI). Switching to CAPI eliminated a variety of data transcription problems, automated the survey's skip patterns, and allowed answer checks during the interview instead of during a post-processing phase.

NLSY79 users will notice a number of changes when they compare the 1994 NLSY79 CPS section with earlier years. First, there are many more data items is 1994 and subsequent years. The 1993 data set contains data for 87 CPS items, while the 1994 version contains 228 data items. Not all of these new questions contain useful data since a number are internal machine checks. 

Moreover, even though there are more data items, some respondents actually answer fewer questions in the new revised CPS section. For example, the 1994 NLSY79 CPS section contains information on 50 disabled individuals. These individuals answered only a small number of the section's questions. However, for many respondents the revised CPS section contains more in-depth responses. Additional categories of answers were added to many questions, such as how respondents search for work and the number of jobs a multiple-job respondent holds.

Finally, NLSY79 researchers should know that the 1994 CPS section increased the likelihood that a respondent would be classified as unemployed. For example, data from the BLS parallel survey suggest that the revised CPS's introduction increased national unemployment rates by 0.5 percent. Moreover, this increase primarily occurred among women, not men.

Weekly labor force/military status

The detailed information on employment dates and gaps between jobs collected during the regular NLSY79 surveys has permitted the construction of weekly Labor Force Status variables for each NLSY79 respondent for the period January 1, 1978, through the most current survey date. In the event that a respondent is not interviewed for one or more surveys, he or she is asked to provide retrospective information at the first reinterview point in order to maintain a continuous set of variables in the Work History data. Respondents on active military duty were not asked CPS questions. These Labor Force Status variables enable researchers to determine whether, during any week since January 1, 1978, a respondent was

• working
• associated with an employer
• unemployed
• out of the labor force
• not working
• on active military duty (for some survey years)

Because these weekly labor force variables are constructed from actual start and stop dates and information on employment gaps within and between jobs, the coding categories and meanings differ from the survey week and date of interview variables described above. These coding categories are defined in Figure 3. Users should note that the number of weeks in each array is greater than the actual number of weeks filled in to provide a "cushion" when creating the Work History data. The extra weeks found in these arrays contain no valid data. See the Work History section for further details.

NLSY79 job search questions provide data that show what methods were used, how intensively respondents searched, and the outcome of these searches. The survey provides data on topics not found in many other key labor market data sets like the Current Population Survey (CPS). For example, the NLSY79 provides details on job offers rejected while searching, the desired characteristics of the job being sought, and whether the respondent searched while employed in another job. While every round of the NLSY79 contains questions on job search activities, researchers interested in this area should examine the 1982, 1986, and 1987 surveys, which contained special sections with a large number of job seeking questions.

Who searches

The NLSY79 asks respondents in various surveys who are both working and not working if they are searching for work. To find most of the questions which determine if a respondent is seeking work, search for questions which contain the phrase "looking for work." This phrase captures all of the employed job search questions plus all non employed job searchers from 1979 to 1993. To capture not employed job seekers after 1993, use the phrase "done anything to find work."

Methods of job search

Respondents are asked not only if they searched for work but are also asked the specific types of activities they used to find work. Coded activities usually include: checking with a State employment agency; checking with a private employment agency; contacting an employer directly; contacting friends or relatives; placing or answering an ad; and looking in a newspaper. Depending on the specific year and set of survey questions the list of coded activities is sometimes longer than those just mentioned and includes other job search activities like: using a school placement service; taking a civil service exam; contacting an organization such as CETA or the Urban league; checking with a labor union; or asking a teacher for job leads. In the 1986 and 1987 surveys, job seekers who were not employed were asked to tell interviewers the top three job-search methods they used. Respondents also ranked these job search methods, allowing researchers to know which method was tried the most.

In 1981, the NLSY79 questionnaire contained an extensive set of time use questions. The time use questions not only covered daily activities but asked roughly 3,300 respondents how long they spent using specific methods of finding work. The job seeking time use section, which begins at R05173., tracks the number of minutes over the last seven days the respondent used employment services, asked relatives about finding work, answered ads, and a variety of other specific techniques. Combining the results from all these questions results in the total number of minutes spent searching for work in the past week.

Duration of job search

Questions for both employed and not employed job seekers in the CPS portions of the questionnaire refer to job search activity over the past four weeks. Even though the questions are only asked to respondents who recently searched, the total amount of time spent searching could be much longer than four weeks. If a respondent states that they were seeking work, the NLSY79 questionnaire probes for how long the respondent has been looking for work. While the specific format changes depending on the survey year, researchers can convert all answers into the number of weeks spent searching. While the mean length is under 2 months for those employed and around 3 months for those respondents not employed, users are cautioned that a small but significant number of respondents have very long job searches, with durations around two years.

Why was the search started and stopped

The NLSY79 has extensive information on why respondents started searching and why they stopped. While the exact list of answers varies depending on the survey year, the questionnaires prior to 1994 includes a single question that asks job seekers who were not employed why they were searching for work. Job seekers could state they began a search because they lost their job, quit their previous job, left school, wanted to help the family with money, or a host of other reasons. The reasons were also asked in selected surveys to employed job seekers.

Beginning in 1994, there is no longer a single question that asks seekers who were not employed why they were looking. Instead, the single question is replaced by a sequence (see for example the 1994 sequence R45541.00 - R45543.00), which first asks the respondent's activities before starting their job search, followed by questions which ask respondents how their last job ended. While this sequence of questions does not completely replicate the single question set of answers, combing the information from the sequence provides a relatively close match.

Given that the NLSY79 is a longitudinal survey that tracks, month-by-month, the respondent's labor market status (see the work history section for details on this month-by-month tracking), it is relatively easy to see when a job seeker found a new position. However, not all job seekers find employment. The NLSY79 also contains some information on why people are no longer searching for work. For example, in the 1996 CPS section, selected respondents were asked why they were no longer looking for work. Responses included: chance for advancement improved in their current job, could not find a better job, pay increased at their current job, working condition improved at their current job, relocation plans changed, career plans changed, the respondent's financial situation improved, and a layoff ended.

Special sections

The NLSY79 has periodically added a number of special job search sections to the questionnaire. In the 1982 survey a section entitled "job finding" was added. This section asked very detailed questions about how the respondents found their current primary job. The section contains the standard high level questions such as: what was the main reason you were looking for work and what methods did you use to find work?

Beyond these standard questions, the section contains unique job search questions. Respondents are asked to specify exactly how many employers they directly applied to for work; how far they traveled to look for work; and if they moved to take the new job. Additionally, respondents were asked if any specific person helped them to get a job. If a specific person did help them, that person's sex, relationship to the respondent, and degree of help are all quantified by the survey. While most of the job search questions are found under the "Job Search" area of interest, other questions, particularly those that track job search by unemployed respondents, are found in the Current Population Survey areas. These areas are labeled "CPS," "CPS 1994," "CPS 1996," and "CPS 1998."

In the 1986 and 1987 surveys the "gaps between jobs" section was expanded for the male portion of the cohort. For each gap between jobs, respondents were asked about all types of methods used to find work. Then, month-by-month within that gap, respondents were asked the specific search methods used during that shorter time period. Respondents who stated more than three methods were asked to list the three most frequently used.

Finally, for each of the top three methods the respondent was asked how many job offers each method produced, the highest wage for each method, and if the offer was accepted. If the person rejected the job offer they were asked why. A partial list of reasons for rejection includes unsuitable working conditions, too many hours, better offer provided, and transportation difficulties.

The employer supplement sections in the 1994 to 2000 surveys contain a number of specific questions about how the respondent found the particular job on which the supplement focuses. Respondents are first asked if they were looking for work when they were offered the job. Then respondents who were looking for work were asked which job search method led to their being offered the job. Finally, respondents were asked if they turned down any other offers before accepting this job and how much those other offers paid.

Hypothetical jobs

The NLSY79 from 1979 to 1982 included questions about hypothetical job offers. These questions can be found by looking for the word "hypothetical" in the Question Text search. The set of questions in 1979 asked the respondent would they work at $2.50 per hour, $3.50 per hour and $5.00 per hour washing dishes, doing factory work, work cleaning, checking out groceries in a supermarket, cleaning up the neighborhood, making hamburgers, and working in a national park. Additional questions from 1979 to 1982 asked how much money it would take for a person to switch employers if the job was in the same general field. If the person was interested in switching employers the survey asked how many days per week the respondent wanted to work and how many hours per day they would like to be employed at the hypothetical job.

These questions provide information about the respondent's reservation wage, or the minimum pay needed to cause the individual to either switch jobs or join the labor force. Combining this with general labor market indicators found on the NLSY79 data set, such as the local area unemployment rate, enables researchers to understand how the local labor market conditions affect an individual's work force decisions.

Three sets of job satisfaction measures have been collected for employed respondents during select survey years. In addition, a job satisfaction index can be created using items collected during the 1979-1982 and 1988 surveys.

1. Global Job Satisfaction: During each survey year, respondents employed since the last interview are asked to rate, on a four-point scale from "like it very much" to "dislike it very much," how they feel about their current/most recent (CPS) job. This question provides a general or 'global' indication of a respondent's current job satisfaction. Beginning in 1994, this question was asked about each job.
2. Satisfaction with Government Jobs Program: Respondents who were working at a job in conjunction with their participation in a Federally funded employment and training program were asked, during the 1979-1987 survey years, how satisfied/dissatisfied they were with their entire experience in the jobs program.
3. Facet-Specific Job Satisfaction Scale: During the 1979-1982 and 1988 surveys, wage and salaried workers, as well as those self-employed in incorporated businesses, were presented with a series of descriptive statements about the pay, working conditions, promotion opportunities, supervisors, coworkers, and so forth at their current job and were asked to rate each statement on a scale from "very true" to "not at all true."

Constructing a Job Satisfaction Index

A job satisfaction index can be constructed for the 1979-1982 and 1988 survey years by coupling six of the facet-specific job satisfaction ratings listed above with the global job satisfaction measure and a question that asked respondents whether, given the freedom to make such a choice, they would take another job or keep the job they had at the survey point. This scale, a shortened form of the job satisfaction scales of the University of Michigan's Quality of Employment Surveys, developed by Quinn (1973), provides a reliable indicator of job satisfaction.

To construct the full seven-item scale, raw scores for each item listed in Table 1 should be converted to z scores for each respondent. The scores can be multiplied by 100 to remove decimals and combined to obtain an unweighted average of the seven z scores. The resulting scores for the satisfaction index are either positive or negative numbers that can be interpreted as deviations from the mean for the total sample of survey respondents.

References

Quinn, R.B. and Mangione, T.W. "Jobsat '72 and its Kinfolk--a Brief Manual." In The 1969-1970 Survey of Working Conditions: Chronicles of an Unfinished Enterprise. Ann Arbor, MI: Survey Research Center, Institute for Social Research, University of Michigan, 1973.

These variables are limited to the 1979 and 1982 data collections for the NLSY79. Descriptions of the broader range of job characteristic data available for the NLSY79 can be found in the Jobs & Employers section. The JCI was preceded by an instrument developed by Hackman and Oldham known as the Job Diagnostic Survey (JDS), dimensions of which are also incorporated in the JCI, although in a simpler format. Comparisons of the JCI and JDS by Dunham et al. have shown that both scales tend to collapse to a one-dimensional scale measuring job complexity. Therefore, the JCI was shortened by selecting one scale item that loaded strongly on each of the dimensions of job complexity shown to be important in earlier research. In their 1976 article, Sims et al. reported the necessary factor analysis scores used to obtain the abbreviated scale. Question and reference numbers for the seven items that compose the shortened JDI scale are listed in Table 1.

References

Dunham, Randall B.; Aldag, Ramon; and Brief, Arthur P. "Dimensionality of Task Design as Measured by the Job Diagnostic Survey." Academy of Management Journal 20,2 (June 1977): 209-23

Hackman, J.R. and Oldham, J.R. "Development of the Job Diagnostic Survey." Journal of Applied Psychology 60 (1975): 159-70.

Sims, Henry R.; Szilagyi, Andrew; and Keller, Robert. "The Measurement of Job Characteristics." Academy of Management Journal 26,2 (June 1976): 195-212.