Indicator 23 examines the expectations of 1981–82, 1991–92, and 2003–04 12th-graders by several characteristics. The three surveys used for this indicator differed slightly in how they constructed variables for race/ethnicity and socioeconomic status (SES) and in whether they imputed missing data. This supplemental note describes these survey differences to provide contextual information for the comparisons made between years in indicator 23. The surveys are the following:

• High School and Beyond Longitudinal Study of 1980 Sophomores (HS&B-So:80/82);
  
  
• National Education Longitudinal Study of 1988 (NELS:88/92), “Second Follow-up, Student Survey, 1992”; and
  
  
• Education Longitudinal Study of 2002 (ELS:02/04), “First Follow-up, Student Survey, 2004.”

EDUCATIONAL EXPECTATIONS

Each of the three surveys asked students: “As things stand now, how far in school do you think you will get?” and gave them a choice of responses. Although the wording of the possible responses was not identical in all three surveys, the responses were collapsed into four broader categories with equivalent meaning: High school diploma or equivalent or less (no postsecondary experience); some college, including vocational/technical (including postsecondary credits but no credentials, certificates, and associate’s degrees—i.e., any postsecondary experience less than a bachelor’s degree); bachelor’s degree; and graduate or first-professional degree.

RACE/ETHNICITY

The HS&B and the NELS surveys had five categories for race/ethnicity: Hispanic or Latino (of any race) plus four categories among non-Hispanic respondents (White, Black or African American, Asian/Pacific Islander, and American Indian/Alaska Native). The ELS questionnaire also included a sixth category: “more than one race, non-Hispanic.” Respondents in the two earlier surveys who would have identified themselves as multiracial presumably chose one of the available categories or did not respond to the question about their race. Therefore, comparing responses of any of the race categories over time may be misleading because of this inconsistency. (The categories “more than one race” and American Indian/Alaska Native categories are not shown separately due to the small number of cases.) The effects of this change in definitions are unknown, but they are likely to be minor: only 4 percent of the weighted ELS:2002 sample were in the “more than one race” category.

SOCIOECONOMIC STATUS

The SES variable was constructed similarly for each of the three surveys, but some differences exist. First, in NELS and ELS, five items were equally weighted to create the composite variable: father’s educational attainment, mother’s educational attainment, father’s occupation, mother’s occupation, and family income. However, the HS&B data omitted mother’s occupation and used only the other four items to create the SES variable. Second, HS&B relied on student reports for the variables used to create the SES variable, while NELS and ELS used parent reports and substituted student reports when parents’ data were unavailable; ELS imputed data that were still missing. Finally, HS&B estimated family income by incorporating both reported income and household belongings, while NELS used data on family income where available and turned to household belongings only if income was not reported. For more information on other differences among the SES-related variables used in the three datasets, see Appendix H of the ELS:02/04 data file documentation, available at http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2006344.

IMPUTATION

In addition to the differences in variable definitions, the ELS data used for indicator 23 include imputed responses, while data from the NELS and HS&B surveys do not include imputed responses. Imputations are estimates of likely responses for cases where actual responses are missing. Imputations are extrapolated logically from respondents’ answers to other items, to the extent possible. When logical inference is not possible, widely accepted statistical methods are used to assign likely responses based on characteristics of the case being imputed and responses from people with similar characteristics. For information on the possible effects of imputation in ELS (including of the SES composite), see appendix C of the ELS:02/04 data file documentation, available at http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2006344.