1. An Evaluation of Data from the Teacher Compensation Survey: School Year 2006-07 June, 2010 Stephen Q. Cornman Frank Johnson

2. Teacher Compensation Survey (TCS) Initiated in Response to:  Demand for more and better data on teachers’ compensation on a comparative state-by-state basis  Demand for data on total compensation that teachers receive, including benefits

3. Purpose of TCS: Why collect these data?  Teachers are one of the most important components of education—and certainly the most expensive.  Current reports on actual salary data are only available at the state level and are not comparable

4. TCS Data Collection  Administrative records survey  Collect individual level data on each public school teacher  School year 2006-07 and 2007-08 data- 17 states participated  School year 2008-09 data- 23 states committed to participate

5. States Participating in TCS

6. TCS 2007-08 Data Collection  17 states- 1.4 million records  1.12 million unique teachers (34.4% of teachers in US)  approximately 31,300 schools in 5,400 districts Total TeachersTeachers in TCSTeachers not in TCS 3,178,1421,119,7112,058,431 100%35.23%65.57%

7. TCS 2008-09 Data Collection  23 states participating-approximately 1.437 million unique teachers  Approximately 45.2% of teachers in US

8. Teacher Compensation Survey Variables  Dependent variables  Base salary  Total salary  Health benefits  Retirement benefits  Other benefits  Identifier variables  State assigned Teacher ID (use for longitudinal studies)  Linking variables  LEA ID (tie data to NCES Local Education Agency Universe survey)  NCES School ID (tie data to other NCES School Universe survey, e.g., locale codes)  Independent Variables  Experience  Education: highest degree earned  Teacher status  Salary indicator  Demographics: gender, race, age  New teacher in state  New teacher in district  Contract days  FTE

9. Challenges to TCS  Data perspectives differ by State  Variable definitions need to be understandable and consistent  Reconciling data  Carrying teacher ID’s forward  Tracking teachers across state borders  Attracting and retaining volunteer states

10. Data Availability  All states can report base salary  6 of 17 states reported health and retirement benefits data  4 states able to assign consistent teacher identification number

11. Data quality  Variations in State data collection period, variable definitions and response patterns  Snapshot reporting limits data on teachers who joined mid-year or left mid-year  Business rules developed and applied based on data plans, review of state policies, and response pattern consistency

12. Comparison of TCS with Other Sources of Data  FTE counts in the TCS and School Universe are within 4 percent of each other in 14 states  Schools in TCS and School Universe match up well:  31,410 in TCS and 31,087 in School Universe  TCS mean base salary higher than SASS in 15 of 16 states  Mean total teacher salary from TCS data agreed to within 5 percent of average teacher total salary reported by NEA in 11 of 13 where comparison could be made

13. Mean Teacher Salaries from NEA and TCS, 2006-07

14. Limitations of TCS  Not all SEA’s collect administrative data on teachers compensation  Differences in how states interpret variable definitions  Unique ID’s not being reported on longitudinal basis  ID’s cannot be used to track teachers across state borders  TCS cannot meet all data needs-less comprehensive than SASS

15. Advantages of TCS  First individual level teacher data base in the country  Reliable database  TCS can be linked with the NCES School Universe- provides ability to analyze the association of teachers salaries with free and reduced lunch eligible students, ELL students, and geographic areas, etc.

16. Data Analysis  Descriptive statistics such as the median salaries of teachers and counts for different groupings by experience, education level, age, race, and gender; new teachers’ salaries (Research and Development Report: Evaluation of Data from Pilot TCS 2006-07)

17. Teachers’ Mean Base Salaries: School year 2006–07 SOURCE: U.S. Department of Education, National Center for Education Statistics, Common Core of Data (CCD), "Teacher Compensation Survey," school year 2006–07, Version 1a.

18. Median years of experience, school year 2006–07 SOURCE: U.S. Department of Education, National Center for Education Statistics, Common Core of Data (CCD), "Teacher Compensation Survey," school year 2006–07, Version 1a. Participating state

19. Median Base Salary by Years of Experience SY 2006-07

20. Teachers Level of Education and Base Salary, SY 2006-07 SOURCE: U.S. Department of Education, National Center for Education Statistics, Common Core of Data (CCD), "Teacher Compensation Survey," school year 2006–07, Version 1a.

21. New & Exp. Teachers’ Median Base Salaries, SY 2006–07 SOURCE: U.S. Department of Education, National Center for Education Statistics, Common Core of Data (CCD), "Teacher Compensation Survey," school year 2006–07, Version 1a.

22. Percentage distribution of teachers by age, SY 2006–07 SOURCE: U.S. Department of Education, National Center for Education Statistics, Common Core of Data (CCD), "Teacher Compensation Survey," school year 2006–07, Version 1a.

23. Examples of Research Questions  What is the association between the percentage of students eligible for free or reduced lunch and teachers’ base salaries?  What is the association between geographic location (urbanicity) and teachers’ base salaries?  What is the association between teaching in charter schools (compared to regular public schools) and teachers’ base salaries?

24. Regression Analysis  Merge the TCS file with the Public Elementary/ Secondary School Universe Survey of the Common Core of Data (CCD)  Remove outliers  Establish “cut points” after review of salary schedules  City, rural, suburb, and town variables created by collapsing categories from locale codes

25. TCS Data Files and Products  School Year 2005–06 (available now)  Research and Development Report: An Exploratory Evaluation of the Data from the Pilot Teacher Compensation Survey: School Year 2005–06 (http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2008440)http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2008440  Restricted-Use Data file (http://nces.ed.gov/ccd/rutcs.asp)http://nces.ed.gov/ccd/rutcs.asp  Public-Use Data files (http://nces.ed.gov/ccd/tcssurv.asp)http://nces.ed.gov/ccd/tcssurv.asp  School Year 2006–07 (available June 2010)  Research and Development Report: An Evaluation of the Data from the Teacher Compensation Survey: School Year 2006–07  Restricted-Use Data file  Public-Use Data file

26. Contact information  Stephen.Cornman@ed.gov Stephen.Cornman@ed.gov (202)502-7338  Frank.Johnson@ed.gov Frank.Johnson@ed.gov (202)503-7362  Elizabeth.e.Holland@census.gov Elizabeth.e.Holland@census.gov

27. Example of Regression Model Dependent Variables: teachers’ base salary Independent Variables (variables of interest)  Proportion of students eligible for free or reduced lunch  Rural, city, town (compared to suburbs) Controls for Teacher Characteristics  Experience (years teaching)  Education: MA, PhD, less than BA (compared to BA)  Teacher gender: Female (compared to Male)  Teacher race (American Indian, Asian, Hispanic, Black (compared to White) School Independent Variables  School level: middle school, HS (compared to elementary school)  Charter schools (compared to regular schools)  Type of school: voc., special education, other (compared to regular schools)