Appendix A: Methodology and Technical Notes

General Notes

Source of data. The numbers of high school diploma recipients, and the student membership data that comprise the denominator for the Averaged Freshman Graduation Rate (AFGR), are taken from the Common Core of Data (CCD) State Nonfiscal Survey file for all AFGRs presented in this report, with the following exception: numbers of graduates by gender reported in table 8 are based on data taken from the CCD Local Education Agency Universe Survey file. The dropout data are reported on the CCD Local Education Agency Universe Survey file. State–level dropout data are created by aggregating local education agency level (LEA) data to the state level. The 2006–07 membership data that were used to create the enrollment base (denominator) for the 2006–07 event dropout rate were taken from the CCD School Universe survey. The grade-level membership data were aggregated to the LEA level from schools associated with the LEA.

The 2006–07 data were reported through the U.S. Department of Education’s EDFacts system. With the exception of New York and California, all states reported their dropout and high school data through EDFacts; New York and California both reported their graduation data through EDFacts but reported their dropout data directly to the NCES CCD staff. The District of Columbia, Puerto Rico, and the U.S. Virgin Islands also reported all their data through EDFacts. American Samoa and the Northern Marianas reported through the CCD online collection maintained by the U.S. Census Bureau and Kforce Government Solutions. Guam, the U.S. Department of Defense dependent schools (overseas and domestic) and the Bureau of Indian Education did not report graduates or dropouts for the 2006–07 school year.

Missing data. When reporting results, NCES treats missing data within individual states differently than it treats missing data across all states, the District of Columbia, and other jurisdictions as a whole. An individual state is considered to have missing data if an item is reported for less than 80 percent of eligible students. If information is missing for some but no more than 15 percent of eligible students across the 50 states and District of Columbia, NCES calculates totals and identifies them as "reporting states" totals (rather than totals for the United States).

EDFacts accepted blank responses in 2006–07 school year reports and did not require that states distinguish among missing, not applicable, and "zero" values. Every effort was made to correctly identify responses as missing, not applicable, or zero after the fact, but it is possible that some blank responses may have been categorized incorrectly.

Totals. "United States" total in tables 1 is limited to the 50 states and the District of Columbia. This total includes data for all 50 states and the District of Columbia. "Reporting states” totals in the remaining tables are limited to the 50 states and the District of Columbia. Because not all, but at least 85 percent of eligible students in the 50 states and the District of Columbia are represented in the graduation and dropout counts, these tables present a "reporting states" total. "United States" and "Reporting States" totals do not include data from the Bureau of Indian Education, Department of Defense dependents schools (overseas and domestic), Puerto Rico, or the other jurisdictions of American Samoa, Guam, the Commonwealth of the Northern Mariana Islands, or the U.S. Virgin Islands. See "Missing data" (above) for more information.

Protecting the confidentiality of dropout and high school graduation data. It would be possible under some conditions to identify an individual student who dropped out of school or who was not awarded a regular diploma at the end of 12th grade. For example, if a state had 10 twelfth grade White, non-Hispanic students enrolled and only 1 White, non-Hispanic high school graduate, that graduating student could infer that all of the other White, non-Hispanic students had failed to graduate. The same situation could occur with dropouts. For example, if a state reported 15 White, non-Hispanic students enrolled in grade 9 in 1 year and 15 White, non-Hispanic ninth grade dropouts, an outside observer could infer that all of the original students had dropped out. (These would be inferences because the CCD cannot distinguish between students who fail to graduate or who drop out and students who transfer out of state or into private schools.) The dropout and high school completion data were modified in order to prevent the identification of any 12th grade student who did not receive a regular high school diploma or any student who dropped out of school. There were few threats to confidentiality in the state-level data. Some reported data were changed to "missing" and some reported numbers of graduates were increased slightly. These changes resulted in a minimal distortion of information.

Data quality. There is variation in the degree of rigor with which the states or school districts verify their data. Those states that collect dropout or graduation data through student-level records systems are better able to verify students’ enrollment and graduation status than are those agencies that collect aggregate data from schools and districts. NCES does not audit state reports. NCES will ask a state to confirm or revise data if the numbers appear to have changed from the prior year considerably more for that state than others, or if the data appear unlikely on the basis of internal evidence. An example of this latter condition would be a state in which the number of graduates exceeded the number of grade 12 students.

Notes on High School Graduation Data

Differences in definitions of "graduate." State and local policies can affect the numbers of graduates reported. There are differences in what a high school diploma represents across states. The CCD defines a regular diploma as the high school completion credential awarded to students who meet or exceed coursework and performance standards set by the state or other approving authority. However, some states award regular diplomas to all students who meet completion requirements, regardless of the extent to which these requirements address the state or district’s academic standards. For example, some states have in-school General Education Development (GED) programs that require fewer credit hours than a regular high school track, but lead to the award of regular diplomas.

Other states award some form of alternative credential to students who meet some, but not all, requirements. For example, special education students who complete their individual education programs or regular education students in some alternative programs may receive a certificate of completion. As a result of different policies, students who receive a certificate of completion in one state might be awarded a regular diploma in another.

Calculating the AFGR. The AFGR is the number of regular diploma recipients in a given year divided by the average of the membership in grades 8, 9, and 10, reported 5, 4, and 3 years earlier, respectively. For example, the denominator of the 2006–07 AFGR was the average of 8th-grade membership in 2002–03, 9th-grade membership in 2003–04, and 10th-grade membership in 2004–05. Ungraded students are prorated into these grades. Averaging these three grades by NCES provides an estimate of the number of first-time freshmen in the class of 2003–04 freshmen in order to estimate the on-time graduation rate for 2006–07.

The method used to create the enrollment base for the AFGR was changed slightly in the 2004–05 school year. Before 2004–05, the enrollment base used the total enrollments by grade. Beginning with the 2004–05 files, the race/ethnicity detail for each grade was summed to form the enrollment base. The change resulted in a slightly more accurate enrollment base, but had little effect on the AFGRs.

Treatment of ungraded students. Although the AFGR denominator is based on enrollments by grade, some states report ungraded students. To adjust for this, the data for ungraded enrollment counts were redistributed across grades in proportion to the graded enrollment of the state, based on the race/ethnicity and gender of the ungraded students and the race/ethnicity and gender of the graded students where this information was available. The redistributed counts of ungraded enrollments were added to the reported enrollment counts for the eighth, ninth, and tenth grades. In this procedure, the ungraded enrollment is subtracted from the total enrollment to get the total graded enrollment; the reported enrollment in each grade is used to compute the proportion of graded students in each grade; these proportions are multiplied by the count of ungraded students to allocate the ungraded students across the individual grades; the counts of ungraded students allocated to each grade are added to the reported enrollments by grade. Ungraded enrollments account for approximately 1 percent of enrollments each year.

Cautions in interpreting the Averaged Freshman Graduation Rate. Although the AFGR was selected as the best of the available alternatives, several factors make it fall short of a true on-time graduation rate. First, the AFGR does not take into account any imbalances in the number of students moving in and out of the nation or individual states over the high school years. As a result, the averaged freshman class is at best an approximation of the actual number of freshmen, where differences in the rates of transfers, retention, and dropping out in the three grades affect the average. Second, by including all graduates in a specific year, the graduates may include students who repeated a grade in high school or completed high school early and thus are not on-time graduates in that year.

While the AFGR is a reasonable proxy at the aggregate national or state level, the potential effects of three factors should be taken into account when interpreting the results for individual states. First, if more high school students moved out of a population than transferred in during the high school years, the number of graduates in the numerator would be smaller and the estimated graduation rate would be lower than the actual on-time rate for that group of freshmen. On the other hand, if more high school students moved into a population than moved out during this 4-year period, the number of graduates in the numerator would be increased and the estimated on-time graduation rate would be higher than the actual rate for that group of freshmen. This can lead to estimated graduation rates of more than 100 percent for small groups; such cases have been adjusted to 100 percent in this report.

Second, including the estimate of eighth-graders from the previous year in order to remove the effect of freshmen who were retained, and thus are not first-time freshmen, ignores the fact that in some cases there may be real change in the number of eighth-graders relative to counts of ninth-graders due to transfers between public and private schools. If more students transfer to public schools at this point, using a count of eighth-graders that does not include those students would serve to artificially decrease the estimated number of ninth-graders, and as a result increase the graduation rate for that class. Conversely, if more students were to transfer out of public schools between the eighth and ninth grades, using the eighth-grade count that includes students leaving the population would artificially increase the estimated number of ninth-graders and in turn, decrease the graduation rate.

Third, there may be a tradeoff between the adjustment for retentions and grade specific differences in the number of dropouts. The use of the 10th-grade enrollment count also helps to dampen the effect of 9th-grade retentions, but ignores the fact that 9th-grade dropouts result in a smaller 10th-grade population. Excluding these ninth-grade dropouts lower the estimate of freshmen and as a result increase the graduation rate.

Missing and suppressed data. The Department of Defense, the Bureau of Indian Education, and Guam did not report high school graduation data for 2006–07. In addition, Kentucky and New York did not report graduation data by race/ethnicity or gender. The District of Columbia reported graduation data by race/ethnicity or gender for a very low percentage of its overall graduate counts and the data were therefore suppressed. Nevada did not report membership by race/ethnicity or gender in 2004–05. The 10th-grade enrollment from 2004–05 is required to calculate AFGR by race/ethnicity for 2006–07.

Notes on Dropout Data

Definition of a dropout. The CCD provides an event dropout number. An event dropout number represents the proportion of students dropping out each year. According to the CCD definition, a dropout is an individual who

• was enrolled in school at some time during the previous school year;
• was not enrolled at the beginning of the current school year;
• has not graduated from high school or completed a state- or district-approved education program; and
• does not meet any of the following exclusionary conditions: transfer to another public school district, private school, or state- or district-approved education program; temporary absence due to suspension or school-approved illness; or death.

The following statements apply for the purpose of this definition:

• The school year is the 12-month period of time from the first day of school (operationally set as October 1), with dropouts from the previous summer reported for the year and grade in which they fail to enroll.
• Individuals who are not accounted for on October 1 are considered dropouts.
• A school completer is an individual who graduated from high school or completed a state- or district-approved educational program upon receipt of formal recognition from school authorities. A state- or district-approved educational program may consist of special education and district or state-sponsored GED preparation.

Defining the school year. Not all states follow a fall-to-fall school year. The CCD dropout count is based on an October–September school year in which a student’s dropout status is determined at the beginning of the year. Some states follow a July–June calendar in which a student’s dropout status is determined at the end of the school year. Dropout rates in states that follow an alternative reporting calendar are comparable to rates for states that follow the October–September calendar (Winglee et al. 2000) and therefore data for states that follow alternative reporting calendars are published in the CCD data files.

Between-year (summer) dropouts. The CCD definition accounts dropouts to the grade and school year for which they do not meet their obligation. Students who complete 1 school year but fail to enroll in the next school year are counted as dropouts from the school year and grade for which they failed to return. For example, a student completing 10th-grade in 2005–06 who does not enroll the next year would be reported as an 11th-grade dropout for 2006–07.

GED programs. Students who leave high school to enroll in adult education/GED preparation programs are reported as dropouts, unless the district tracks these students and reports as dropouts those who fail to complete the program. Students who have received a high school equivalency by October 1 are not dropouts regardless of where they prepared for the test, if the GED is an accepted high school credential in the state.

Calculating the event dropout rate. The event dropout rate for a given grade is the number of dropouts from that grade divided by the number of students enrolled in that grade at the beginning of the school year. For example, the 10th-grade event dropout rate for 2006–07 is the number of 2006–07 10th-grade dropouts divided by the number of students in 10th-grade at the beginning of the 2006–07 school year. Ungraded students are prorated into the denominator. The high school event dropout rate aggregates dropouts and students in membership for 9th- through 12th-grades.

Treatment of ungraded students in counts of dropouts and enrollments for dropout rates. Dropout counts are reported by states to the CCD by grade (grades 7–12). Ungraded students who drop out of school are assigned by the LEA or state to the grade in the dropout count that most closely matches the grade they would have been enrolled in based on their age. Counts of ungraded student enrollments are prorated into graded enrollment counts in order to have denominators for the calculation of dropout rates that reflect the placement of ungraded dropouts in the graded numerators. The proration process is the same as that used for the AFGR.

Missing and suppressed data. Vermont suppressed all dropout counts less than 10 prior to submitting these data to EdFacts for the 2006–07 school year. This resulted in more suppressed data values than actual counts. Due to the extent of missing data, NCES suppressed the remainder of the dropout data for Vermont, thus making it impossible to report dropout counts or rates for Vermont in 2006–07. New York did not report dropout data by grade, only dropout data in aggregate for grades 9 through 12. Pennsylvania did not report dropout data for grades 9 and 10. The race/ethnicity and gender reporting of dropouts for the following states did not represent 80 percent or more of the total student population for any of the 5 reported race/ethnicities: Maryland, New York, and Vermont. Pennsylvania reported dropout data representative of 80 percent or more of the Hispanic and Black, non-Hispanic high school population but not for any of the other 3 race/ethnicities.

Top