Personal tools
You are here: Home Research K-12 Education School Dropouts The National Dropout Data Collection System: Assessing Consistency

The National Dropout Data Collection System: Assessing Consistency

Authors: Phillip Kaufman
Date Published: January 13, 2001

Unfortunately, while a great deal of time and resources are being devoted to measuring one educational outcome—the academic achievement of students in school— less is being devoted to measuring the complementary outcome—how many students complete high school.
Related Documents


One measure of the success of a country’s educational system is the level of education achieved by its young adults. In this country, the primary standard of educational level has been high school graduation. Indeed, one of the oldest series of data collected by the Federal government is the proportion of the population that has completed high school. These data show that there has been remarkable progress in the last half-century in high school completion rates. Rates increased from 38 percent of all 25 to 29 year olds in 1940 to around 86 percent in the early 1980s and have remained constant since (U.S. Department of Education, 2000). As a consequence of this progress, during the last half century, high school completion became an expectation for young people in this country (Dorn, 1996).

However, the current enthusiasm for high school exit exams has prompted concerns about the adverse effects that these exams might have on high school completion (Hebert and Hauser, 1999). Some have thought that the implementation of such tests will make it more likely that students will drop out of school rather than face the consequences of failure on these exams. For example, Bonsteel and Rumberger warn that the twin horns of the end of social promotion and the beginning of high school exit exams will greatly increase the number of high school dropouts (Bonsteel and Rumberger, 1999).

Obviously, measuring the impact of exit exams requires reliable statistics on recent trends in high school completion and dropout. Unfortunately, despite decades of collecting data on completers and dropouts, this still remains no easy task. The relatively limited resources that go into the collection of high school completion and dropout data at the federal level produce data that provide more heat than light on some rather basic questions on high school completion—how many students drop out in any given year and how many students complete high school.
Completion and dropout rates can vary dramatically depending on the data source. As I will show below, these differences in rates arise because:

  • Different rates are based on different populations;
  • Different rates are derived from different methods;
  • and Rates based on survey methods generally have large sampling errors.

Because of these factors, reported rates differ significantly from one anther and are not easily “translatable” into one another. This may lead to the appearance that they give different answers and make it difficult to policy makers to sort out the magnitude of the dropout “problem”. In this paper I will describe the difficulties we currently face in providing high school dropout and completion data on a consistent and timely basis. I will review the data that are currently being collected on high school completion and dropout and attempt to explain why different sources provide different answers to the basic questions about high school completion in this country. After briefly describing the national data and the state-by-state breakouts of the national data, I will compare the different rates derived from these different data sources and attempt to reconcile some of the differences in reported rates.

In the discussion, I have two specific recommendations to improve data collection on high school completion. My first recommendation is to support the development of the new American Community Survey—a new household survey that has the potential to give much more accurate estimates of high school dropouts and completers at the state and local level. As these estimates will be derived from the same survey instrument, they will provide comparisons across states (and within states) based on common definitions of dropout and completion.

My second recommendation is to supplement existing data systems with longitudinal surveys of students. As I will argue below, not only do longitudinal data have the potential of providing more accurate estimates of dropouts and completers, but they also provide the context in which one can understand the changes in the process of dropping out (perhaps due to changes in graduation policies).

In compliance with the UC Open Access Policy, this report has been made available on eScholarship:

Document Actions

Copyright © 2010 UC Regents