high schools, however, students’ experiences are spread over six or seven (or more) teachers each year, meaning that the appropriate unit of analysis is not the classroom but rather the school. In addition to cognitive assessments, NELS:88 also collected data from sampled students about their activities and attitudes with paper-and-pencil surveys. When appropriate, we note how students’ responses might inform school-level investigations. All childlevel reports of behaviors on ECLS-K are made by teachers or parents.
E&S survey data are available at both the district and the school levels, and Table D-1 shows the data that are available at the lowest level of aggregation, which is generally the school, but for some measures it is individual classrooms. Virtually all E&S data are broken down into multiple student categories (see the table footnotes). Throughout the table, the symbol * between two student characteristics indicates that the measure is broken down by more than one student characteristic (e.g., gender and race and ethnicity). For example, “Total Student Enrollment Gender*Grade” indicates that the E&S survey provides information about the number of boys and girls in each grade. Similarly, “Grade*Race” indicates how many black eighth graders or white third graders are enrolled in a particular school.
The ECLS-K measures listed in Table D-1 are from the first grade file, measured in 1999–2000. NELS:88 measures are from the second follow-up, which occurred during the 1991–1992 school year, when most students were in twelfth grade. It is important that researchers planning to create combined E&S/NELS:88 data files use E&S survey data from the appropriate year: for example, the 1992 E&S survey data for those using data from the second NELS:88 follow-up. Much of the aggregate information available on the E&S survey that is unavailable at the school level on ECLS-K or NELS:88 is available about individual sampled students, although researchers should be judicious in making school-level aggregates from student-level data. Although within-school student samples were randomly drawn, the numbers for which some aggregates are drawn is small, and they are all in a single grade. Furthermore, many similar (and even identical) school-level measures are also available for the ECLS-K base year (when sampled children were in kinder