[ Main Page of Report | Contents of Report ]
This study uses data from the National Longitudinal Survey of Youth--1979 cohort (NLSY79). The NLSY79 is a large, nationally representative, omnibus survey sponsored by the U.S. Bureau of Labor Statistics. (3) Over 12,000 youths ages 14-22 were first interviewed in 1979. They have been re-interviewed annually through 1994 and biennially since. The original sample included over-samples of blacks, Hispanics, economically disadvantaged non-black non-Hispanics, and youth in the military. Over time the economically disadvantaged and military over-samples were discontinued due to budget reductions. The remaining sample has seen remarkably low attrition with over 84 percent having been interviewed in 1998, the eighteenth round of interviewing.
The NLSY79 focuses on labor market behavior with information collected on aspects of the respondents' lives which are thought to influence, or be influenced by, their labor market behavior. The survey routinely collects information on education, job training, marriage, fertility, household composition, health status, income, and assets. In selected years, funding from various government agencies has provided for collecting additional information on such things as alcohol consumption and drug usage.
The advantage of using the NLSY79 is the availability of measures of long-term adult outcomes in a continuous context. This requires a longitudinal survey so data sets such as the Youth Risk Behavior Survey (YRBS) or the Current Population Survey (CPS) are inadequate.
Also, surveys such as the Survey of Income and Program Participation (SIPP) do not follow respondents for a long enough period of time. Other longitudinal surveys of adolescents, such as Add Health, the National Survey of Adolescent Males, or the new NLSY97, began more recently and have not yet followed their respondents beyond the earliest years of adulthood. The youngest respondent in the NLSY79, on the other hand, turned 34 in 1998 (the most recent year of data available at the time of this study). Over the years of the survey, we are able to see the respondents (generally) complete their education, develop their careers, and form families. Also, the NLSY79 has collected information on certain behaviors such as alcohol use and drug use at multiple points in time.
The NLSY79 pioneered asking about sensitive activities in a large omnibus survey. However, these questions did not appear until the respondents were mostly out of their adolescent years. (4) Thus contemporary measures of frequency and intensity are not measured in the years needed for this study. For the most part, we are restricted to a retrospective report on age of initiation. The one exception to this is in reports of delinquent and criminal behaviors. In this case, questions were included in 1980, when the respondents were 15-23 years old. The weakness in this case is that some respondents were no longer adolescents, and it is a one-time measure so that cumulative effects differ across different aged respondents. (5) While these weaknesses limit the analysis in certain ways, the NLSY79 is very in its inclusion of a wide variety of long term adult outcomes.
Like all survey data, the NLSY79 relies on self-reports of socially undesirable behaviors. Thus there may be under-reporting of these behaviors. Mensch and Kandel (1988) find some evidence of under-reporting drug use in the 1984 NLSY79. Also, since we are limited to reports of age of initiation, the length of retrospective recall may distort the distribution. Individuals may recall inaccurately, putting events nearer or further in time than their actual occurrence. However, contemporary reports may include more reporting of less important events (e.g. a single cigarette smoked) and may be more affected by the social desirability of a given response. Thus, retrospective reports of age of initiation may not be worse than contemporaneous reports.
The sample used in this study includes all NLSY79 respondents who are part of the continuing sample. Table 1 shows sample sizes by sex and race/ethnicity.
Sex |
Race/Ethnicity | Sample Size | Percent |
|---|---|---|---|
Male |
Non-Black non-Hispanic |
2529 | 25.33 |
Black |
1524 | 15.26 | |
Hispanic |
981 | 9.82 | |
Female |
Non-Black non-Hispanic |
2495 | 24.98 |
Black |
1477 | 14.79 | |
Hispanic |
980 | 9.81 | |
| Total | 9986 | 100 |
It should be noted that this is the total available sample for the analysis. Any given behavior, outcome, or other variable may be missing for a particular observation. Thus sample sizes will differ depending on what relationship is being analyzed.
[Go To Contents]
In this chapter we provide a general description of the variables and estimation techniques employed in this study. A complete description of how we created variables, the mean and sample size for each variable, and a discussion of our estimation techniques is included in Appendix A.
The NLSY79 offers a wide array of outcomes we can study. We define ten adult outcomes of interest and categorize them into four major domains. The outcome domains and the measures in each domain are:
Health
1. A measure of past-year alcohol abuse or dependence around age 30 (7), defined according to the DSM III-R (8). Alcohol abuse and alcohol dependence measure different problem behaviors. We combine them to capture either type of alcohol use disorder.
2. Whether used drugs (marijuana or cocaine) in the past month around age 30. (9) While simply having used marijuana or cocaine in the past month is not an indication of a drug problem per se, most adults do not use these drugs routinely. We specifically chose not to capture drug use in the past year, a common measure, in order to reduce the likelihood of capturing occasional recreational use.
Economic
1. Ever under the poverty line between ages 25-29. Data limitations restricted the highest age we could consider to 29. We purposely cut off the bottom age to prevent college students from appearing to be in poverty (see Appendix A).
2. The number of years in poverty between the ages of 25-29. This variable distinguishes those who may have temporarily, for any number of reasons, fallen below the poverty line from those who are more regularly in poverty.
3. Ever on welfare (AFDC/TANF or food stamps) between ages 21-33. As an alternative to the poverty measure, we capture welfare receipt. The data allow us to look over a wider range of ages than for poverty.
4. The number of years on welfare between ages 21-33. This variable distinguishes transitory welfare receipt from longer term dependence.
5. The percent of employable time spent employed between the end of formal schooling and age 33. This variable captures attachment to the labor force. In order not to "penalize" women for time spent out of the labor force for childbearing, we calculate the time spent employed as a percent of time in the labor force. We identify the end of formal schooling before starting the count.
6. The age at which achieved at least 2 years working for a single employer since leaving formal schooling. Unlike general labor force attachment, this measure captures how long it took, since leaving formal schooling, to hold a "steady" job. There is not a standard measure of labor market success commonly used, but these two measures are consistent with the literature (Pergamit 1995).
Family Formation
We create a single outcome comprised of six combinations of marriage and fertility outcomes measured at age 33. These are:
Crime
Whether ever been in jail by age 33. This variable is based on the location of the interview (respondents in the NLSY79 are followed anywhere, even into institutions). Only incarcerations as an adult are counted. Since some jail episodes would have occurred between interviews, this outcome is underestimated. However, no national data set measures the lifetime prevalence of serving time in jail; the NLSY79 provides an opportunity not found elsewhere (Freeman 1996).
The NLSY79 offers us a set of risky behaviors that typically begin during or near the teenage years, if they begin at all. We examine five of these behaviors:
Each of these is measured using age of initiation except for delinquency, which is a measure of total number of delinquent and/or criminal acts in 1980. We divide ages of initiation into four age categories:
Combining those who initiate after age 19 and those who never initiate reflects our interest in the effects of adolescent behavior. These two groups include all individuals who did not initiate as an adolescent. Further, since we do not observe respondents' entire lifetimes, we cannot know if they "never" initiate. One caveat throughout our analysis is that "late initiators" and "never initiators" may be very different groups.
Two primary ways of analyzing delinquency appear in the literature. One divides delinquent acts into "personal" and "property" (e.g. Greenberg 1985). Another common method is to break down delinquency by levels of frequency. Nye and Short (1957) suggest four categories: (1) did not commit the act, (2) committed the act once or twice, (3) committed the act several times, and (4) committed the act very often. We adopt this latter approach and define four categories of delinquency based on the number of delinquent/criminal acts engaged in: 0, 1-2, 3-8, and 9 or more.
For our investigation into the impact of adolescent family structure, we use a measure of living arrangements at age 14. While the impact of family structure is likely a dynamic process, differing depending on when and how many changes occurred, we simplify the analysis to a single age around the time that most adolescents are making their choices whether to undertake a risky activity. (10)
We identify six different family types:
To explore the impact of SES, we examine mother's and father's education, categorized as: less than 12 years, 12 years, some college, and college graduate. To explore the impact of having a parent with alcohol problems, we create a single variable that captures if either biological parent was an alcoholic or a problem drinker during the respondent's childhood. (11)
[Go To Contents]
We employ a variety of estimation methods as dictated by the type of outcome variables examined. For the continuous variable measuring the percent of time employed since leaving school, we are able to use ordinary least squares (OLS). For the dichotomous variables alcohol abuse or dependence, used drugs in the past month, ever in poverty between ages 25-29, and ever on welfare between ages 21-33, we employ logistic regressions. In the case of marriage and fertility outcomes, where there are multiple combinations of a single outcome variable, we estimate a multinomial logistic. The variables, "number of years in poverty" and "number of years on welfare," are count variables that are estimated using a negative binomial regression. Finally, to estimate the time since leaving school that is needed to acquire a steady job (lasting at least two years) requires estimating a survivor function which we approximate with the Weibull distribution.
A more detailed description of the estimation methods is included in Appendix A. What is important is the interpretation of the values reported from these regressions. They fall into two categories: marginal effects and odds ratios. We report marginal effects for OLS, negative binomial, and Weibull regressions. For each of these methods, the marginal effect measures the change in the outcome variable for a unit change in an explanatory variable. In the case of a categorical variable, each marginal effect is relative to the effect of an omitted category. For example, we divide age of initiation into four categories. In the regressions, initiation between ages 11-15 is omitted and serves as the reference category.
Odds ratios are reported for logistic and multinomial logistic regressions. Odds ratios measure the relative probability of the estimated outcome among one group relative to the reference group. The choice of reference group is the same across all types of estimations. Odds ratios for multinomial logistics are difficult to interpret because there are multiple equations. As an alternative we create predicted values for each outcome combination for each category of the relevant explanatory variable. Thus, we would have a predicted probability of observing each of the six adult family formation outcomes given each of the four age of initiation categories.
[Go To Contents]
3. For a complete description of the NLSY79 and the entire NLS program, see Bureau of Labor Statistics, NLS Handbook 2000, Washington, DC.
4. The youngest members of the sample may have gotten questions about some risky behaviors in their later adolescent years. For example, the first alcohol questions appeared in 1982 when the respondents were 17-25 years old.
5. The measurement error introduced most likely increases the standard errors of any of our estimates, making it more difficult to detect significant relationships.
6. Note: The analysis sample excludes the military over-sample dropped after 1984 and supplemental economically disadvantaged non-black, non-Hispanic over-sample dropped after 1990.
7. The respondents were 29-32 at the time the outcome was measured.
8. DSM-III-R refers to the American Psychiatric Association's Diagnostic and Statistical Manual of Mental Disorders, Third Edition, Revised. This tool is used by practitioners to help diagnose and treat mental disorders. The manual was updated in 1994 with publication of DSM-IV.
9. The respondents were 28-32 at the time the outcome was measured.
10. This simplification seems reasonable given the scope of this study, however we clearly do not capture the full family context. For example, for an adolescent living with a single parent at age 14, we do not know whether the parents never married; whether a marriage ended in divorce or death, or whether a divorce (or death) occurred recently or earlier in the adolescent's life.
11. We require that the respondent must have lived with the biological parent at least one year.
Main Page of Report | Contents of Report
Home Pages:
Human Services Policy
(HSP)
Assistant Secretary for Planning and Evaluation (ASPE)
U.S. Department of Health and Human Services
(HHS)