What difference does a variable make? : the robustness of teacher value-added estimates when including additional variables
Abstract/Contents
- Abstract
- Using an eleven-year longitudinal database of over 210,000 grade 3-11 student scores and 7,500 teachers ratings, this study examined whether teacher VA ratings remained consistent when seldom-used variables were added to a Base Model. Ten separate alternative models were run in English Language Arts (ELA) and in Math. These alternative models added student gifted status (GATE), language development level for English learners, engagement measured by attendance and suspension rate, ACT Explore performance, PSAT performance, class-size, variability in achievement, school fixed effects, and combined models that included (a) all variables but ACT Explore and PSAT, and (b) all variables but ACT Explore, PSAT, and school fixed effects. Correlations between Base Model VA estimates and alternative model VA estimates ranged from 0.93 (ACT Explore) to 1.00 (Engagement, Class-Size, and Achievement Heterogeneity) in ELA. In Math, correlations between alternative models and the Base Model ranged from 0.92 (PSAT) to 1.00. The effects of these alternative variables on teacher VA were grouped into three tiers, with school effects (estimated by school fixed effects) making a difference in VA ratings for a substantial majority of teachers (on average, 65.9% changed deciles compared with the Base Model), measures related to academic preparation and/or student ability altering the decile rankings for, on average, 41.1% of teachers, and other types of variables having affecting fewer teacher ratings, with an average of 19.7% of teacher decile ranks changing for these models compared with the Base Model. However, even in this last category, a given variable change in the model may have a substantial effect on the ratings of a small subset of teachers. Including school fixed effects led to 30% of teachers moving out of the lowest decile in ELA, and almost 40% doing so in Math. Including ACT and PSAT led 19.0% and 27.3% of teachers to move out of bottom decile in ELA and 17.4% and 10.3% to do so in Math. Including GATE moved just over 10% of teachers to move out of the bottom decile in ELA and in Math. Other variables affected the ratings of less than 10% of teachers in the bottom decile. Implications for teacher evaluation as formative feedback and summative decision-making were discussed.
Description
| Type of resource | text |
|---|---|
| Form | electronic; electronic resource; remote |
| Extent | 1 online resource. |
| Publication date | 2015 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Associated with | Newton, Stephen P |
|---|---|
| Associated with | Stanford University, Graduate School of Education. |
| Primary advisor | Darling-Hammond, Linda, 1951- |
| Thesis advisor | Darling-Hammond, Linda, 1951- |
| Thesis advisor | Haertel, Edward |
| Thesis advisor | Stipek, Deborah J, 1950- |
| Advisor | Haertel, Edward |
| Advisor | Stipek, Deborah J, 1950- |
Subjects
| Genre | Theses |
|---|
Bibliographic information
| Statement of responsibility | Stephen P. Newton. |
|---|---|
| Note | Submitted to the Graduate School of Education. |
| Thesis | Thesis (Ph.D.)--Stanford University, 2015. |
| Location | electronic resource |
Access conditions
- Copyright
- © 2015 by Stephen Paul Newton
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...