PDF & Related Info
ContactGiven their coverage and content, the SEA data systems can potentially be used to provide useful insights on teacher labor markets utilizing both cross-sectional and longitudinal designs. We consider examples of each in turn.
It is possible to compare experience and education-adjusted teacher pay across states. This could be done for base or total pay, taking into account the definitional issues discussed in the previous section. One option would be to compare relative pay at different points in teachers' careers (e.g., bachelor's degree, master's degree plus 10 years' experience, master's degree plus 20 years' experience). One would want to do this for several years to assess trends. It is also possible to factor in district or school characteristics, such as urban areas/rural areas, high poverty/low poverty, and other covariates. Such descriptive exercises would permit analysis of how relative salaries change when the concept of pay varies. For example, one could examine whether the association between teacher career trajectory and pay differs when considering base and total salary. Researchers would also be able to determine how much bias arises when one does not control for covariates. Finally, for states with linkable records, researchers could compare career growth in pay across states and districts.
As with studies that use SASS's Teacher Follow-Up Survey (TFS), researchers would be able to use SEA teacher compensation data to track mobility by experience levels. However, the SEA data would allow for a much more detailed analysis. TFS, for example, does not allow an analysis of teacher "stop out" behavior (i.e., temporary exits). As a result, teacher replacement needs are often overestimated. Multiyear studies using state data systems would permit analyses of the return of teachers after temporary exits.
Researchers could also disaggregate these turnover and mobility data by race, gender, and other district or market characteristics (e.g., they could report values for specific large urban districts), which cannot be done with data from the TFS because of the study's small sample size. In addition, these multistate data would permit interstate comparisons of teacher retirement behavior.28 Other studies could shed light on the possible effects of policy interventions. For example, a large, multistate longitudinal teacher data file would make it possible to examine the effects of statewide or districtwide policies on teacher retention and mobility.
It would be possible to examine interstate teacher mobility by linking records across the state systems.29 For example, if all states using Social Security numbers as a teacher identifier agreed to use a standard encryption algorithm, it would be possible to track teacher mobility across those states.
While there are many possible uses for these teacher-level compensation data, there are several limitations. For example, only 34 states report that they collect these data, and only 25 of these states indicated that they are currently willing to make the data available to NCES and/or researchers. Additionally, given that there are no requirements that SEAs collect or report these data, participation by states may fluctuate over time, making longitudinal analyses difficult. Similarly, the fact that not all states would be participating would result in a coverage issue that could limit or bias conclusions based on an analysis of the available data. One of the most powerful potential uses of these data is for longitudinal analyses across states that would allow researchers to investigate the flow of teachers back into teaching after a temporary exit. However, this analysis requires that states uniquely identify teachers, and do so in a uniform manner across all states. Many states already identify teachers using Social Security numbers, which (as previously noted) could be encrypted. This issue is likely the most difficult to resolve, as it would require that all states participate,30 and that they agree on a common encryption system.
Even with these limitations, the results of this research suggest that there is a wealth of teacher compensation data that states currently collect and that a number of states are willing to make these data available for research. Additionally, a majority of states collect salary data in a form that allows for comparability across states, and these data include critical variables that would be required for meaningful analyses. SEA teacher compensation data are a promising source of information that would allow researchers to more fully research a crucial aspect of education policy.
Recent collaboration between states and NCES has confirmed the feasibility of developing this data resource. In 2007, seven states31 participated in a pilot data collection, providing teacher compensation data to NCES. In addition to salary information, data submitted by states include many of the critical variables discussed in this report, including school ID, and teacher characteristics such as years of experience, highest degree earned, full-/part-time status, year of birth, and race/ethnicity and gender. States assign a unique identifier to each teacher, allowing for longitudinal tracking of teachers within a state. The 2008 collection is expected to include up to 20 states (including the seven states that participated in 2007), and NCES expects to expand this collection over time.
