An Exploratory Evaluation of the Data from the Pilot Teacher Compensation Survey: School Year 2005-06
NCES 2008-440
April 2008

Appendix A: Methodology and Technical Notes

Common Core of Data survey system. The State Nonfiscal Survey of Public Elementary/Secondary Education, the Local Education Agency Universe Survey, and the Public Elementary/Secondary School Universe Survey are the nonfiscal components of the Common Core of Data (CCD) survey system, while the School District Finance Survey and the National Public Education Financial Survey (NPEFS) are the fiscal components. School finance data are reported annually by state education agencies (SEAs) through the efforts of state CCD coordinators. Participation in the CCD is voluntary. The Teacher Compensation Survey (TCS) is also a part of the CCD survey system. The survey will be considered a research and development effort until more than 45 states are participating in the collection.

Data for the TCS are collected from SEAs through an online reporting system. They are then processed, edited, and verified by the U.S. Census Bureau, the National Center for Education Statistics (NCES), and the Education Statistics Services Institute (ESSI) of the American Institutes for Research (AIR). The school year 2005–06 pilot TCS collection opened on May 23, 2007 and closed on September 4, 2007. Arizona, Arkansas, Colorado, Florida, Iowa, Missouri, and Oklahoma reported data in the school year 2005–06 Pilot Teacher Compensation Survey.

The pilot TCS collects a limited amount of data on each public school teacher in the participating states. The data in this report are based on all full-time teachers (full-time equivalency, or FTE, greater than or equal to 0.9), who teach at one school only, and whose base salaries are for teaching duties only. Teachers who teach at regular schools, charter schools, special education schools, vocational education schools, or other types of schools are included. Arizona and Arkansas did not report whether individual base salaries included work in addition to teaching. Records for these two states were included if their FTE were greater than or equal to 0.9 and base salaries were greater than zero. The TCS universe file contains 509,225 records representing 497,927 teachers. The total number of records meeting the requirements for inclusion in this report was 381,375.

Data quality. Staff at NCES, the U.S. Census Bureau, and ESSI collaborate to edit all CCD data submissions and ask state CCD coordinators to correct or confirm any numbers that appear out of range when compared with other data. If no explanation for anomalous data is provided by the state, NCES may change the data value to missing.

Missing data and data editing. Not all states collect and report all of the data items requested in the pilot TCS. NCES has not performed any imputations for missing data on the pilot TCS, with the exception of missing FTE data in certain circumstances. If FTE was reported as missing, a value of 1.0 was imputed for records where the teacher status indicator showed the teacher to be a full-time teacher in one school only and the teacher's base salary was larger than the 10th percentile of the base salaries of all teachers with the number of years of experience in the state. Other edits to the data included editing the years of experience data in Arkansas, Colorado, Iowa, and Oklahoma so that teaching experience would be counted as of the end of the 2005–06 school year. This was accomplished by adding one to the years of experience in those four states. Another edit was to adjust total salary to equal base salary for those records in which total salary was less than base salary. Teachers who served in more than one school had a record for each school where they taught, and the FTE value reflected the amount of time the teacher worked at that school. These teachers were not included in the analysis. Only data for full-time teachers teaching at one school were used in this report. The age of a teacher was derived from the birth year data reported by the state. Age was calculated by subtracting the birth year from 2006.

Fifth, median, and 95th percentile cutpoints. The 5th percentile is a value such that 5 percent of the observations are less than this value and that 95 percent are greater. The 95th percentile is a value such that 95 percent of the observations are less than this value and that 5 percent are greater. The 5th and 95th percentile cutpoints have been chosen to exclude any outlier data. A median is a number dividing the higher half of a population from the lower half. The median can be found by arranging all the observations from lowest value to highest value and picking the middle one.

Federal range ratio. The federal range ratio is used in this report as an indicator of the difference between teachers with relatively high base salaries and teachers with relatively low base salaries. As used by Berne and Stiefel (1984) and in previous NCES publications (Parish, Matsumoto, and Fowler 1995; Hussar and Sonnenberg 2000), the federal range ratio excludes the top and bottom 5 percent of districts in order to reduce the influence of extreme values. The federal range ratio is the difference between the base salaries of the teacher at the 95th percentile and the teacher at the 5th percentile divided by the amount for the teacher at the 5th percentile.

Comparability of data across states. There are some issues affecting the comparability of these data that have not been resolved. These include inconsistencies in the reported data themselves and differences in reporting. NCES continues to identify these inconsistencies and will work with state education agencies to reduce them as much as possible. A major factor regarding the comparability of the data is differences in salaries and other costs in different areas within states and across the country, i.e., it costs more to hire a teacher in one part of the country than it does to hire a teacher with the same qualifications in another part of the country. This report does not take into account cost differences across the states or within states.

Confidentiality protection. In order to prevent the identification of an individual teacher with certainty, some data items were swapped between records. These changes resulted in a minimal amount of change to the file overall.