Appendix A.2. High School and Beyond Longitudinal Study

The High School and Beyond Longitudinal Study (HS&B) is a national longitudinal survey of individuals who were high school sophomores and seniors in 1980. The base-year survey (conducted in 1980) was a probability sample of 1,015 high schools with a target number of 36 sophomores and 36 seniors in each school. A total of 58,270 students participated in the base-year survey. Substitutions were made for nonparticipating schools—but not for students—in those strata where it was possible. Overall, 1,122 schools were selected in the original sample and 811 of these schools participated in the survey. An additional 204 schools were drawn in a replacement sample. Student refusals and absences resulted in an 82 percent completion rate for the survey.

Several small groups in the population were oversampled to allow for special study of certain types of schools and students. Students completed questionnaires and took a battery of cognitive tests. In addition, a sample of parents of sophomores and seniors (about 3,600 for each cohort) was surveyed.

HS&B first follow-up activities took place in the spring of 1982. The sample for the first follow-up survey included approximately 30,000 people who were sophomores in 1980. The completion rate for sample members eligible for on-campus survey administration was about 96 percent. About 89 percent of the students who left school between the base-year and first follow-up surveys (e.g., dropouts, transfer students, and early graduates) completed the first follow-up sophomore questionnaire.

As part of the first follow-up survey of HS&B, transcripts were requested in fall 1982 for an 18,152-member subsample of the sophomore cohort. Of the 15,941 transcripts actually obtained, 1,969 were excluded because the students had dropped out of school before graduation, 799 were excluded because they were incomplete, and 1,057 were excluded because the student graduated before 1982 or the transcript indicated neither a dropout status nor graduation. Thus, 12,116 transcripts were utilized for the overall curriculum analysis presented in this publication. All courses in each transcript were assigned a 6-digit code based on the Classification of Secondary School Courses (a coding system developed to standardize course descriptions; see http://nces.ed.gov/surveys/hst/courses.asp). Credits earned in each course are expressed in Carnegie units. (The Carnegie unit is a standard of measurement that represents one credit for the completion of a 1-year course. To receive credit for a course, the student must have received a passing grade—"pass," "D," or higher.) Students who transferred from public to private schools or from private to public schools between their sophomore and senior years were eliminated from public/private analyses.

In designing the senior cohort first follow-up survey, one of the goals was to reduce the size of the retained sample, while still keeping sufficient numbers of minorities to allow important policy analyses. A total of 11,227 (94 percent) of the 11,995 people subsampled completed the questionnaire. Information was obtained about the respondents' school and employment experiences, family status, and attitudes and plans.

The samples for the second follow-up, which took place in spring 1984, consisted of about 12,000 members of the senior cohort and about 15,000 members of the sophomore cohort. The completion rate for the senior cohort was 91 percent, and the completion rate for the sophomore cohort was 92 percent.

HS&B third follow-up data collection activities were performed in spring 1986. Both the sophomore and senior cohort samples for this round of data collection were the same as those used for the second follow-up survey. The completion rates for the sophomore and senior cohort samples were 91 percent and 88 percent, respectively.

HS&B fourth follow-up data collection activities were performed in 1992, but only surveyed the 1980 sophomore class. They examined aspects of these students' early adult years, such as enrollment in postsecondary education, experience in the labor market, marriage and child rearing, and voting behavior.

Appendix table A-1 contains the maximum number of HS&B cases that are available for tabulations of specific classification variables used throughout this publication.

The standard error (se) of an individual percentage (p) based on HS&B data can be approximated by the formula

se_p = DEFT [p(100-p)/n]^1/2

where n is the sample size and DEFT, the square root of the design effect, is a factor used to adjust for the particular sample design used in HS&B. Appendix table A-2 provides the DEFT factors for different HS&B samples and subsamples.

In evaluating a difference between two independent percentages, the standard error of the difference may be conservatively approximated by taking the square root of the sum of the squared standard errors of the two percentages. For example, in the 1986 follow-up of 1980 sophomores, 84.0 percent of the men and 77.2 percent of the women felt that being successful in work was "very important," a difference of 6.8 percentage points. Using the formula and the sample sizes from table A-1 and the DEFT factors from table A-2, the standard errors of the two percentages being compared are calculated to be

1.43[(84.0)(16.0)/(5,391)]^1/2 = .714

1.43[(77.2)(22.8)/(5,857)]^1/2 = .784

The standard error of the difference is therefore

(.714² + .784²)^1/2 = (.510 + .615)^1/2 = 1.06

The sampling error of the difference is approximately double the standard error, or approximately 2.1 percentage points, and the 95 percent confidence interval for the difference is 6.8 ± 2.1, or 4.7 to 8.9 percentage points.

The standard error estimation procedure outlined above does not compensate for survey item nonresponse, which is a source of nonsampling error. (Table A-1 reflects the maximum number of responses that could be tabulated by demographic characteristics.) For example, of the 10,925 respondents in the 1984 follow-up survey of 1980 high school graduates, 372, or 3.4 percent, did not respond to the particular question on whether they had ever used a pocket calculator. Item nonresponse varied considerably. A very low nonresponse rate of 0.1 percent was obtained for a question asking whether the respondent had attended a postsecondary institution. A much higher item nonresponse rate of 12.2 percent was obtained for a question asking if the respondent had used a micro- or minicomputer in high school. Typical item nonresponse rates ranged from 3 to 4 percent.

The Hispanic analyses presented in this publication rely on students' self-identification as members of one of four Hispanic subgroups: Mexican, Mexican-American, Chicano; Cuban; Puerto Rican, Puertorriqueño, Boricuan; or other Hispanic ethnicities.

An NCES series of technical reports and data file user's manuals, available electronically, provides additional information on the survey methodology.

Further information on the HS&B may be obtained from

Aurora M. D'Amico
Postsecondary Studies Division
Postsecondary Cooperative System, Analysis, and Dissemination Program
National Center for Education Statistics
1990 K Street NW
Washington, DC 20006
Aurora.D'Amico@ed.gov
http://nces.ed.gov/surveys/hsb/