Trends in High School Dropout and Completion Rates in the United States

Appendix A: Guide to Sources

Common Core of Data

The Common Core of Data (CCD) is the primary database of the National Center for Education Statistics (NCES) on public elementary and secondary education in the United States. It is a comprehensive, annual, national statistical database of all public elementary and secondary schools and school districts containing data designed to be comparable across all states. This database can be used to select samples for other NCES surveys and provide basic information and descriptive statistics on public elementary and secondary schools and schooling in general.

The CCD collects statistical information annually from approximately 100,000 public elementary and secondary schools and approximately 18,000 public school districts (including supervisory unions and regional education service agencies) in the 50 states, the District of Columbia, the Department of Defense Education Activity (DoDEA), the Bureau of Indian Education (BIE), Puerto Rico, American Samoa, Guam, the Northern Mariana Islands, and the U.S. Virgin Islands. Three categories of information are collected in the CCD survey: general descriptive information on schools and school districts; data on students and staff; and fiscal data. The general school and district descriptive information includes name, address, phone number, and type of locale; the data on students and staff include selected demographic characteristics; and the fiscal data pertain to revenues and current expenditures.

The EdFacts data collection system is the primary collection tool for the CCD. NCES works collaboratively with the Department of Education’s Performance Information Management Service to develop the CCD collection procedures and data definitions. Coordinators from state education agencies (SEAs) submit the CCD data at different levels (school, agency, and state) to the EdFacts collection system. Prior to submitting CCD files to EdFacts, SEAs must collect and compile information from their respective local education agencies (LEAs) through established administrative records systems within their state or jurisdiction.

Once SEAs have completed their submissions, the CCD survey staff analyzes and verifies the data for quality assurance. Even though the CCD is a universe collection and thus not subject to sampling errors, nonsampling errors can occur. The two potential sources of nonsampling errors are nonresponse and inaccurate reporting. NCES attempts to minimize nonsampling errors through the use of annual training of SEA coordinators, extensive quality reviews, and survey editing procedures. In addition, each year SEAs are given the opportunity to revise their state-level aggregates from the previous survey cycle.

EDFacts

EdFacts is a centralized data collection through which SEAs submit PK–12 education data to the U.S. Department of Education (ED). All data in EdFacts are organized into “data groups” and reported to ED using defined file specifications. Depending on the data group, SEAs may submit aggregate counts for the state as a whole or detailed counts for individual schools or school districts. EdFacts does not collect student-level records. The entities that are required to report EdFacts data vary by data group but may include the 50 states, the District of Columbia, the DoDEA, the BIE, Puerto Rico, American Samoa, Guam, the Northern Mariana Islands, and the U.S. Virgin Islands. More information about EdFacts file specifications and data groups can be found at https://www2.ed.gov/about/inits/ed/edfacts/file-specifications.html.

EdFacts is a universe collection and is not subject to sampling error, although nonsampling errors such as nonresponse and inaccurate reporting may occur. ED attempts to minimize nonsampling errors by training data submission coordinators and reviewing the quality of state data submissions. However, anomalies may still be present in the data.

Differences in state data collection systems may limit the comparability of EdFacts data across states and across time. To build EdFacts files, SEAs rely on data that were reported by their schools and school districts. The systems used to collect these data are evolving rapidly and differ from state to state.

In some cases, EdFacts data may not align with data reported on SEA websites. States may update their websites on schedules different from those they use to report data to ED. Also, ED may use methods for protecting the privacy of individuals represented within the data that could be different from the methods used by an individual state.

Current Population Survey

The Current Population Survey (CPS) is a monthly survey of about 54,000 households conducted by the U.S. Census Bureau for the Bureau of Labor Statistics. The CPS is the primary source of labor force statistics on the U.S. population. In addition, supplemental questionnaires are used to provide further information about the U.S. population. The March supplement (also known as the Annual Social and Economic [ASEC] supplement) contains detailed questions on topics such as income, employment, and educational attainment; additional questions, such as items on disabilities, have also been included. The October supplement contains questions on school enrollment and school characteristics.

CPS samples are initially selected based on results from the decennial census and are periodically updated to reflect new housing construction. The current sample design for the main CPS, last revised in July 2015, includes about 74,000 households. Each month, about 54,000 of the 74,000 households are interviewed. Information is obtained each month from those in the household who are 15 years of age and over, and demographic data are collected for children 0–14 years of age. In addition, supplemental questions regarding school enrollment are asked about eligible household members age 3 and over in the October CPS supplement.

In January 1992, the CPS educational attainment variable was changed. The “Highest grade attended” and “Year completed” questions were replaced by the question “What is the highest level of school ... has completed or the highest degree ... has received?” Thus, for example, while the old questions elicited data for those who completed more than 4 years of high school, the new question elicited data for those who were high school completers, i.e., those who graduated from high school with a diploma as well as those who completed high school through equivalency programs, such as a GED program. For more information about how the variable changes affected the calculation of event and status dropout rates, see Kaufman, Alt, and Chapman (2004).

A major redesign of the CPS was implemented in January 1994 to improve the quality of the data collected. Survey questions were revised, new questions were added, and computer-assisted interviewing methods were used for the survey data collection. Further information about the redesign is available in Current Population Survey, October 1995: (School Enrollment Supplement) Technical Documentation at https://www.census.gov/programs-surveys/cps/technical-documentation/complete.html. Beginning in 2003, the race/ethnicity questions were expanded. Information on people of Two or more races were included, and the Asian and Pacific Islander race category was split into two categories—Asian and Native Hawaiian or Other Pacific Islander. In addition, questions were reworded to make it clear that self-reported data on race/ethnicity should reflect the race/ethnicity with which the responder identifies, rather than what may be written in official documentation.

The estimation procedure employed for monthly CPS data involves inflating weighted sample results to independent estimates of characteristics of the civilian noninstitutional population in the United States by age, sex, and race. These independent estimates are based on statistics from decennial censuses; statistics on births, deaths, immigration, and emigration; and statistics on the population in the armed services. Caution should be used when comparing population estimates (e.g., the number of 18- to 24-year-olds) from CPS data over long periods of time (e.g., 10 or more years) since CPS data reflect the latest available Census-based controls. For instance, 2012–2017 CPS data reflect Census 2010-based controls, while CPS data from 2003–2011 reflect Census 2000-based controls. Thus, the estimates of levels for data collected in 2012 and later years will differ from those for earlier years by more than what could be attributed to actual changes in the population. These differences could be disproportionately greater for certain population subgroups than for the total population. Nevertheless, the most recent change in population controls had relatively little impact on summary measures such as averages, medians, and percentage distributions.

The generalized variance function is a simple model that expresses the variance as a function of the expected value of a survey estimate. Methods for deriving standard errors and examples can be found within the CPS technical documentation at https://www.census.gov/programs-surveys/cps/technical-documentation/complete.html. Standard errors were estimated using replicate weight methodology beginning in 2005 for March CPS data and beginning in 2010 for October CPS data. Those interested in using CPS household-level supplement replicate weights to calculate variances may refer to Estimating Current Population Survey (CPS) Household-Level Supplement Variances Using Replicate Weights at http://thedataweb.rm.census.gov/pub/cps/supps/HH-level_Use_of_the_Public_Use_Replicate_Weight_File.doc.

Dropouts

Each October, the CPS includes supplemental questions on the enrollment status of the population age 3 years and over as part of the monthly basic survey on labor force participation. In addition to gathering the information on school enrollment, with the limitations on accuracy as noted below under “School Enrollment,” the survey data permit calculations of dropout rates. Both status and event dropout rates are tabulated from the October CPS. Event rates describe the proportion of students who leave school each year without completing a high school program. Status rates provide cumulative data on dropouts among all young adults within a specified age range. Status rates are higher than event rates because they include all dropouts ages 16 through 24, regardless of when they last attended school.

In addition to other survey limitations, dropout rates may be affected by survey coverage and exclusion of the institutionalized population. The incarcerated population has a high dropout rate. Dropout rates for the total population might be higher than those for the noninstitutionalized population if the prison and jail populations were included in the dropout rate calculations. On the other hand, if military personnel, who tend to be high school graduates, were included, it might offset some or all of the impact from the theoretical inclusion of the jail and prison populations.

Another area of concern with tabulations involving young people in household surveys is the relatively low coverage ratio compared to older age groups. CPS undercoverage results from missed housing units and missed people within sample households. Overall CPS undercoverage for October 2017 is estimated to be about 11 percent.

CPS coverage varies with age, sex, and race. Generally, coverage is larger for females than for males and larger for non-Blacks than for Blacks. This differential coverage is a general problem for most household-based surveys. Further information on CPS methodology may be found in the technical documentation at https://www.census.gov/programs-surveys/cps.html.

Further information on dropout data may be obtained from

Joel McFarland
Annual Reports and Information Staff
National Center for Education Statistics
550 12th Street SW
Washington, DC 20202
[email protected]

Educational Attainment

Reports documenting educational attainment are produced by the Census Bureau using the March CPS supplement (ASEC). Currently, the ASEC supplement consists of approximately 70,000 interviewed households. Both recent and earlier editions of Educational Attainment in the United States may be downloaded at https://www.census.gov/topics/education/educational-attainment/data/tables.All.html.

In addition to the general constraints of CPS, some data indicate that the respondents have a tendency to overestimate the educational level of members of their household. Some inaccuracy is due to a lack of the respondent’s knowledge of the exact educational attainment of each household member and the hesitancy to acknowledge anything less than a high school education.

Further information on educational attainment data from CPS may be obtained from

Associate Directorate for Demographic Programs—Survey Operations
Census Bureau
U.S. Department of Commerce
4600 Silver Hill Road
Washington, DC 20233
[email protected]
https://www.census.gov/programs-surveys/cps.html

School Enrollment

Each October, the CPS includes supplemental questions on the enrollment status of the population age 3 years and over. Currently, the October supplement consists of approximately 54,000 interviewed households, the same households interviewed in the basic CPS. The main sources of nonsampling variability in the responses to the supplement are those inherent in the survey instrument. The question of current enrollment may not be answered accurately for various reasons. Some respondents may not know current grade information for every student in the household, a problem especially prevalent for households with members in college or in nursery school. Confusion over college credits or hours taken by a student may make it difficult to determine the year in which the student is enrolled. Problems may occur with the definition of nursery school (a group or class organized to provide educational experiences for children) where respondents’ interpretations of “educational experiences” vary.

For the October 2017 basic CPS, the household-level nonresponse rate was 13.8 percent. The person-level nonresponse rate for the school enrollment supplement was an additional 9.9 percent. Since the basic CPS nonresponse rate is a household-level rate and the school enrollment supplement nonresponse rate is a personlevel rate, these rates cannot be combined to derive an overall nonresponse rate. Nonresponding households may have fewer persons than interviewed ones, so combining these rates may lead to an overestimate of the true overall nonresponse rate for persons for the school enrollment supplement.

Although the principal focus of the October supplement is school enrollment, in some years the supplement has included additional questions on other topics. In 2010 and 2012, for example, the October supplement included additional questions on computer and internet use.

Further information on CPS methodology may be obtained from https://www.census.gov/programs-surveys/cps.html.

Further information on the CPS School Enrollment Supplement may be obtained from

American Community Survey

The Census Bureau introduced the American Community Survey (ACS) in 1996. Fully implemented in 2005, it provides a large monthly sample of demographic, socioeconomic, and housing data comparable in content to the Long Forms of the Decennial Census up to and including the 2000 long form. Aggregated over time, these data serve as a replacement for the Long Form of the Decennial Census. The survey includes questions mandated by federal law, federal regulations, and court decisions.

Since 2011, the survey has been mailed to approximately 295,000 addresses in the United States and Puerto Rico each month, or about 3.5 million addresses annually. A larger proportion of addresses in small governmental units (e.g., American Indian reservations, small counties, and towns) also receive the survey. The monthly sample size is designed to approximate the ratio used in the 2000 Census, which requires more intensive distribution in these areas. The ACS covers the U.S. resident population, which includes the entire civilian, noninstitutionalized population; incarcerated persons; institutionalized persons; and the active duty military who are in the United States. In 2006, the ACS began collecting data from the population living in group quarters. Institutionalized group quarters include adult and juvenile correctional facilities, nursing facilities, and other health care facilities. Noninstitutionalized group quarters include college and university housing, military barracks, and other noninstitutional facilities such as workers and religious group quarters and temporary shelters for the homeless.

National-level data from the ACS are available from 2000 onward. The ACS produces 1-year estimates for jurisdictions with populations of 65,000 and over and 5-year estimates for jurisdictions with smaller populations. The 1-year estimates for 2017 used data collected between January 1, 2017, and December 31, 2017, and the 5-year estimates for 2013–2017 used data collected between January 1, 2013, and December 31, 2017. The ACS produced 3-year estimates (for jurisdictions with populations of 20,000 or over) for the periods 2005–2007, 2006–2008, 2007–2009, 2008–2010, 2009–2011, 2010–2012, and 2011–2013. Three-year estimates for these periods will continue to be available to data users, but no further 3-year estimates will be produced.

Further information about the ACS is available at https://www.census.gov/programs-surveys/acs/.