Common Core of Data (CCD)

4. SURVEY DESIGN

TARGET POPULATION

The target population includes all public elementary and secondary schools, LEAs, and SEAs throughout the United States, including the District of Columbia, the Department of Defense Education Activity², BIE schools, Puerto Rico, and the four U.S. Island Areas.

SAMPLE DESIGN

The CCD collects information from the universe of state-level education agencies.

Data Collection and Processing

Through the 2005–06 collection, CCD data were voluntarily obtained from administrative records collected and edited by SEAs during their regular state reporting cycle. In 2006–07, CCD nonfiscal data reporting became mandatory for SEAs.

In 2007–08, reporting CCD nonfiscal data to EDFacts, a new data collection system, became mandatory for SEAs.   SEAs now annually submit over 100 data files through the EDFacts Submission System (ESS).

The school year (SY) 2015–16 collection was the first CCD collection processed completely through EDFacts' Partner Support Center (PSC) and the new Data Management System (DMS). The PSC used the DMS to reach out to SEAs and resolve any data issues. When errors in data were found, SEAs were encouraged to resubmit the data file through ESS.

Reference dates. Most data for the nonfiscal surveys are collected for a particular school year (September through August). The official reference date is October 1st or the closest school day to October 1st. Special education, free-lunch eligibility, and racial/ethnic counts may be taken on December 1st or the closest school day to that date. Student and teacher data are reported for the current school year, whereas through 2005–06, data for high school graduates, other completers, and dropouts reflected the previous year. Fiscal data are for the previous fiscal year; FY 98 data represent the 1997–98 school year.

Data collection. The ways in which CCD data are collected have evolved with the advancement of technology. In the early days of the collection, survey instruments were usually distributed to the states in January. Starting in the 2001–02 collection, downloadable PC software was used. In 2004–05, a web-based data collection application was developed and put into use. A state CCD coordinator, appointed by the Chief State School Officer, is responsible for overseeing the completion of the surveys (often, different coordinators are responsible for the fiscal and the nonfiscal surveys). To ensure comparable data across states, NCES provides the CCD coordinator with a set of standard critical definitions for all survey items. In addition, data conferences and training sessions are held at least yearly. The state's data plan identifies any definitional differences between the state's recordkeeping and the CCD's collection as well as any adjustments made by the state to achieve comparability. Counts across CCD surveys may not be identical, but differences should be consistent and the state is asked to describe the reason for any discrepancy.

NCES provides the state with general information collected during the previous survey on each district and school (e.g., name, address, phone number, locale code, and type of school/district). This information must be verified as correct by the CCD coordinator or recoded with the correct information. The coordinator must also assign appropriate identification codes to new schools and agencies and update the operational status codes for schools and agencies that have closed.

Beginning with the 2005–06 school year, the CCD nonfiscal data were collected through the U.S. Department of Education’s Education Data Exchange Network (EDEN). States reported data to EDEN through multiple file groups that fall into various reporting schedules throughout the year. Although states could report data outside the collection period and could revise their reported data at any time in EDEN, NCES extracted the data files from EDEN on the cutoff dates of data submission. The data resubmitted by states after the files were extracted may or may not be included in the CCD final release file. In 2007–08, EDFacts superseded EDEN as the data collection system used to collect CCD nonfiscal data.

Data for the CCD fiscal surveys and the TCS are collected by the U.S. Census Bureau. The data are compiled into prescribed formats and submitted by the SEAs. The closing date for the current year’s data is the Tuesday following Labor Day. Corrections to submitted fiscal data are accepted until October 1st, however, only corrections that lower a state’s current expenditure per pupil are accepted after the “Labor Tuesday” deadline for use in the formula for allocating Title I and other Department of Education funding to state and local school systems.

Editing. Completed surveys undergo comprehensive editing by NCES and the states. Where data are determined to be inconsistent, missing, or out of range, NCES contacts the SEAs for verification. Review and editing of data for collections prior to SY 2015–16, were done by the Census bureau. Census prepared exception reports that were annotated and sent to SEAs and used as a basis for resolving data anomalies. Starting with SY 2015–16, NCES is processing nonfiscal data through its Data Management System, which provides an online application for exception reports and the exchange of information to resolve data issues. Records are processed through a post–edit check to replace blanks and nonmeaningful zeroes with meaningful responses. After editing, final adjustments for missing data are performed.

Beginning in SY 2009–10, NCES implemented a revised editing methodology for agency and school data to improve the precision of edits in identifying errors. The revised methodology used data from multiple years and was applied to both the reported data items and related ratios. More information on these "multi-year" edits is included in the documentation of the CCD data files (https://nces.ed.gov/ccd/stnfis.asp). Prior to SY 2015 –16 SEAs were allowed greater latitude in reporting counts for zero, missing, and not-applicable values.  With the implementation of DMS and increased standardization of guidance in the file specifications provided to SEAs, EDFacts has required greater precision from SEAs in distinguishing between zero, missing, and not-applicable values.

Estimation Methods

NCES estimates missing values to improve data comparability across states. Only state–level data are estimated on a regular basis. Missing values in the Public School Universe and Local Agency Universe Surveys are generally left as missing, with a few exceptions.

There are two basic estimation methods: imputation and edits. Imputation is performed when the missing value for a data item is not reported at all indicating that subtotals and totals containing the category are underreported. Imputation assigns a value to the missing item, and the subtotals and totals containing this item are increased by the amount of the imputation. Edits correct and adjust reported data. They are used to correct cases in which a value reported for one item incorrectly contains a value for one or more additional items not reported. For example, a state might not differentiate between kindergarten teachers and prekindergarten teachers, reporting “–1” (missing) for prekindergarten teachers and a value representing the count of staff for both categories as kindergarten teachers. CCD survey staff edits these two responses by reducing the amount reported for kindergarten teachers and adding that amount to prekindergarten teachers. The total count for teachers would not be affected by this edit. As well, edits may also be used to change reported data to prevent identification of individuals. Totals and subtotals are recalculated after all imputations and edits have been performed. Last, if it is not possible to impute or adjust for a missing value, the item is set to 0 (“no occurrences”), –1 (“missing”), or –2 (“not applicable”).

Most data elements in the files have a companion cell with a flag indicating whether the data contents were reported by the state (R) or placed there by NCES using one of several methodologies: edit (A); imputation based on the prior year’s data (P); imputation based on a method other than the prior year’s data (I); totaling based on the sum of internal or external detail (T); or combining with data provided elsewhere by the state (C). There is also a code (D) to flag for first–year disaggregated.

Estimating state–level nonfiscal data. NCES imputes and edits some reported values for student and staff counts at the state level (including the District of Columbia). Imputations for prekindergarten students are performed first, followed by staff imputations and then other edits. In the 2015–16 collection, there were a number of cases (13 distinct values in 7 states) where the same values in the same states had been imputed by the same method for three or more years in a row. NCES decided to leave these values as missing rather than impute for a fourth year in a row.

Estimating state–level fiscal data. NCES also imputes and edits revenue and expenditure data. The federal standard (see Allison, Honegger, and Johnson 2009) is used in the edits to distribute expenditure and revenue data. Edits are also used to distribute direct state support expenditures to specific objects and functions. In some cases, local revenues from student activities and food services are imputed.

Future Plans

Because it is an ongoing annual survey, the CCD engages in continuous planning with its data users and providers.

Top

 

Top