The Technical Expert Panel (TEP) was convened to evaluate the quality and utility of the Teacher Compensation Survey (TCS). The TCS is a research and development effort to see whether or not it is possible and realistic to collect and publish teacher-level data from the administrative records that reside in state departments of education. The TCS data files are flat files with one record per teacher assignment, with assignment being one teacher’s instructional activities at one school.
In particular, the panel was asked to:
- Provide an assessment of the data already collected and specific guidelines for its use by researchers, addressing guidelines on current data, to include:
- State-by-state notes of data issues and possible solutions.
- Possibilities and limitation of longitudinal analysis of teachers (with special focus on consistency of identification of individuals year-to-year).
- Possibilities and limitations of year-to-year comparisons given changes in data availability, data quality and changes in definitions of data elements.
- Make recommendations for changes to the TCS collection process that will improve data quality or increase the value of the overall collection.
Throughout the course of the panel’s discussions there was clear agreement among TEP members that there are three significant issues that should be addressed: missing data problems with the TCS, timeliness of TCS data that should and can be improved, and financial elements of the TCS that are problematic.
Further concerns arose including:
- Coverage of the TCS in three respects:
- Number of participating states - fewer than one-half of the states participate,
- Public school employees other than teachers are not included in the TCS, and
- Contract employees are not covered, which will cause a decline in coverage if the trend toward “privatization” of public schools continues.
- The support for longitudinal modeling. This is explicitly not a goal of the TCS as it has existed in the past. The panel recommends that NCES review its position on longitudinal modeling with the TCS. Year-to-year comparisons are feasible.
- Consistency of the TCS with the Schools and Staffing Survey (SASS), CCD and data released by the National Education Association (NEA).
- The defined scope of the TCS should be more precisely defined with the goal that it be sufficiently narrow to make the TCS financially and operationally viable.
Tied to the concerns over coverage is the potential use of the TCS as a frame for other NCES data collections. The panel pointed out that to be used as a frame, the TCS must contain data from all states, and it must contain design-level stratification variables.
The panel met via email with an in-person meeting in Washington DC on May 20-21, 2013.
Conversion of the TCS to Relational Form
The TCS should be “relationalized” - that is, converted to a relational database consisting of a teacher table, an assignment table, a payroll table and a school table. Conversion to relational form removes current ambiguities in the TCS, and supports its expansion and more efficient use.
- Use only actual expenditures taken from accounting databases for all financial elements in the TCS.
- Explore obtaining the financial information for the TCS from unemployment insurance (UI) or other records held by state workforce or tax agencies. Advantages include timeliness and higher quality.
- Investigate in detail issues associated with linking assignment, payroll and teacher data, which may require Social Security numbers.
- Explore adding variables, including dates of employment, standard occupation codes (SOCs), and year-of-degree or year-of-certification.
- Target releasing data for the school year ending 6/30/20xx by 7/1/20(xx+1).
- As part of the relationalization of the TCS, consider carefully whether it wishes the TCS at some point in the future to support linkage to performance data, and if so, ensure that the relationalized version contain the proper “hooks” for doing so.
Concomitant NISS Data Quality Analysis and Recommendations
NISS conducted a data quality analysis which supplements the report. Overall, the extent to which the TCS can enable insightful analyses not possible by other means remains an issue.
To provide an assessment of the data already collected and specific guidelines for its use by researchers, addressing guidelines on current data, and make recommendations for changes to the TCS collection process that will improve data quality or increase the value of the overall collection.
Technical Expert Panel Members
Frank Cernik, (North Carolina Department of Public Instruction)
Tom Gallagher, (Wyoming Department of Workforce Services)
Carrie Jones, (Bureau of Labor Statistics)
Steven Rivkin, (University of Illinois at Chicago)
Marguerite Roza, (Georgetown University)
Key NCES Liaisons
National Institute of Statistical Sciences
Alan F. Karr