Sample Design

Sample Changes over Time

Age shift in the child samples. The age distribution of interviewed women in 2018 (between 53 to 62) underscores the fact that most NLSY79 women have reached the end of their childbearing years. Recent survey rounds also mark a continuing shift in the age composition of the child sample from a predominantly younger child group to a more young adult population. As of the current survey round, about 99 percent of the interviewed child sample is age 15 or older and about 95% are age 21 and over. Since very few, if any, children remain to be born in forthcoming NLSY79 survey rounds, a rapid transition towards an even older child population is evident, with the majority of the children in their 20s, 30s, and 40s. Paralleling this shift, the younger component of the overall sample increasingly resides in middle class households and were born to women at older ages. This trend implies that users should exercise caution in undertaking within-sample analyses in which comparisons are made between children at different ages.

Table 2 shows child's year of birth by mother's age at birth of child for all NLSY79 children. Whereas at one time a large proportion of NLSY79 children had been born to adolescent mothers, nearly all of the children and young adults in the current sample had been born to women age 20 and over. In the current round, the entire younger child sample was born to women ages 35 and older and about 12% of interviewed young adults were born to adolescent mothers. 

Table 2. Child's Birth Year by Age of Mother at Birth of Child: All NLSY79 Children

Sample limitations. Table 2 suggests one other caveat for studies that focus on the consequences of earlier, adolescent childbearing for this cohort of women. A modest proportion of the children (300 of the 4,965 interviewed in the current survey) were born prior to the first NLSY79 interview round. If essential explanatory inputs for analysis include pre-1979 points (e.g., employment status in 1977 or early paternal presence in the home), sample size may be temporally constrained because of this left-censoring problem--the unavailability of some data elements for the pre-survey period. All such cases fall in the upper young adult ages, and could affect analyses for young adult children born in 1978 or earlier.

The increasing heterogeneity of the child sample may also be noted in other ways from Table 2. While there has been an increase in sample heterogeneity over the years, users should remain mindful that the oldest and youngest children in the sample are likely to come from families that differ in their socio-economic characteristics. However, it is also becoming increasingly reasonable to generalize from the NLSY79 sample of children to broader representations of selected U.S. child populations; overall, it is worth reiterating that as of this date, the cohort of women have completed essentially all of their childbearing. 

Sample Changes over Time. The increasing heterogeneity of the child sample over time may be noted from Table 3. This table summarizes the age mix as well as the race/ethnic mix of the child sample as it moves forward in time from 1986, the first year of the child interviews, to the current survey round. Over time, there is a gradual transition towards an older average age at interview. Notwithstanding this increase in age, the overall sample has changed very little over time in its racial and ethnic makeup. There has been some counterbalancing between higher minority birthrates and the reality that a higher proportion of the births in recent years are to older, white women. As evident in Table 3, sample sizes have varied over the years, largely reflecting the variations in data collection already noted. The single largest cause of decline from 1988 to 1990 was the removal of the economically disadvantaged white oversample. The slight decline from 1996 to 1998 was related to the capping of interviews in that year at age 20; and the decline from 1998 to 2000 reflected the one-time exclusion of a part of the black and Hispanic oversamples. In the 2002 interview round, there is no age or other sample exclusion, which accounts for the increase in sample size between 2000 and 2002. This increase is most evident in the young adult ages as the larger age cohorts continue to shift from child to young adult. This table shows the number of both children and young adults interviewed, by single year of age and by race/ethnicity, in survey years prior to the current round.

Table 3. Number of NLSY79 Child Interviews by Age and Race/Ethnicity: 1986-2018

Changes in the Young Adult Sample. When the Young Adult Survey was first fielded in 1994, 1,111 older children were identified as eligible to be fielded, and interviews were conducted with 979 Young Adults. The Young Adults interviewed in the 1994 survey round were disproportionately Black or Hispanic and born to younger mothers. In 1998 only, an additional age restriction was imposed: respondents had to be under 21 by the date of interview. A decision was made to retain data from the 15 Young Adults who were inadvertently interviewed even though they had already turned 21. In 2000, approximately 40 percent of black and Hispanic oversample cases between age 15 and 20 were not fielded. Beginning in 2010, respondents over the age of 30 are interviewed every four years.  (The interviewed sample is selected by age as of December 31 of the survey year, so that approximately half of the older young adults are eligible each round. Since 2010, young adults age 31-32, 35-36, 39-40, 43-44, etc. as of December 31 of the target year are not fielded). Table 4 shows the sample of Young Adults interviewed in 2018.

Table 4. Number of NLSY79 Young Adults Interviewed in 2018 by Age and Race/Ethnicity