Tracking Students Over Time Using Surveys
Larry Mishel and Joydeep Roy question conventional dropout estimates:
The first type of data we examined was from national longitudinal surveys, which follow students over time. We identify the National Education Longitudinal Study (NELS) as the very best source of data because it tracks individual students over time and verifies students' graduation status with transcripts independently obtained from schools. The NELS began with eighth-graders in the spring of 1988 and followed them over the next 12 years, with interviews at regular intervals in 1990, 1992, 1994 and 2000.
The NELS is considered by us and most researchers as the "gold standard." The NELS shows that by 1994, two years after their normal or on-time graduation date, slightly over 82 percent of all students had completed high school with a regular diploma. This number was 85 percent for whites, 95 percent for Asians, 74 percent for blacks and 74 percent for Hispanics. By 2000, when most of these people were age 26, an additional 7 percent of whites, 14 percent of blacks and 9 percent of Hispanics had acquired a GED diploma, which facilitates access to college and the military. These results alone are enough to question the new conventional wisdom that only two-thirds of all students, and only half of all minorities, graduate from high school with a diploma. . . .
The NELS results are confirmed by . . . large-scale longitudinal surveys called the National Longitudinal Survey of Youth conducted by the Bureau of Labor Statistics. Chen and Holzer (2006) show that, for those ages 20-22 in 2002 (including those in prison), the overall graduation rate is the same as in NELS: 82 percent overall, 75 percent for blacks and 76 percent for Hispanics. These data also show an improvement in graduation rates for every race and gender group since 1984, except for black men. The improvements are particularly large and significant for Hispanics, both males and females.
The second type of data we explore is household surveys, which allow us to identify the share of the population at a particular age that has graduated from high school. The Current Population Survey (CPS) results have been questioned because they exclude the institutional (including prison) population and because they undercount low-income blacks. Therefore, we analyzed the decennial census data for 2000. These data include people in institutions (inmates of prisons, nursing homes, hospitals, etc.) and the military and miss very few people in the sample. . . . We find that these census data yield graduation rates very similar to those in the NELS. We also use the CPS and adjust for the increased incarceration rates of black men. . . . Again, the results yield graduation rates far in excess of those claimed in the new conventional wisdom.