45
“Teacher evaluations are not supported by research to improve student performance. There is
no culturally responsiveness embedded in the framework and its evidence in teaching and
learning.”
In addition to the qualitative perceptions on equity outcomes, our team ran quantitative analyses,
analyzing how scores on IMPACT overall and on various components of IMPACT varied by
teacher and school characteristics, including teacher race and gender, school ward, Title I status,
grades served, and subject areas. Through these analyses, we sought to determine if there are
notable patterns of results that could indicate potential bias or unfairness in the system, either for
or against particular groups of teachers or rating teachers serving students with greater needs
more harshly, thus not accounting for differences in student needs and discouraging teachers
from serving schools with greater needs. Because, as noted above, teachers of certain identities
are differentially likely to also teach in particular schools (for instance, Black and
Hispanic/Latino teachers are significantly more likely to teach in Title I schools and Black
teachers are significantly more likely to teach in Wards 7 and 8), and teacher and school
characteristics could be confounded, we also ran regressions to determine how IMPACT scores
vary on these dimensions independently of (holding constant) the other dimensions, as well as
variability in components of the Essential Practices rubric.
Tables 6-10 present average score by IMPACT component for these groups. Female teachers
consistently score higher than male teachers, and White, Asian, and American Indian teachers
consistently score higher than Black, Hispanic or Latino, or Other/Unknown race teachers (there
were an insufficient number of Native Hawaiian or Pacific Islander teachers in the dataset to
report results without risking breach of confidentiality), although teachers in the latter groups do
receive higher average bonus amounts, likely because they are more likely to serve in high-needs
schools eligible for higher bonuses. Notably, the gaps were greatest in Essential Practices, Core
Professionalism, and Contributions to the School Community, and smaller in TAS and Value-
added, and student survey responses were higher for Black female, American Indian female, and
Other/Unknown race female teachers than other groups. We see similar patterns by ward, with
lower scores overall in wards 5, 7 and 8 driven by differences in EP, CP, and CSC. Teachers in
Title I schools generally scored lower than in non-Title I schools, and teachers in elementary
schools scored slightly higher overall and on EP and in high schools on other components, with
middle school teachers generally slightly lower than other levels in almost all categories. The
regression in Table 11 shows that, even holding constant other characteristics, Black,
Hispanic/Latino, and Other/Unknown race teachers, teachers in wards 5 and 8, and middle
school teachers scored lower overall, and female teachers, and teachers in Ward 2 scored higher
overall than other teachers. Finally, Table 12 shows differences in specific components of the
Essential Practices rubric by teacher and school characteristic. These differences could indicate
potential biased or coded language in the rubric or in observation protocols or in observers
themselves. We also examined difference in individuals EPs to provide insight to the possibility
of biased EP measures. The EP components are listed below for reference:
• EP 1: Cultivate a responsive learning community
• EP 2: Challenge students with rigorous content
• EP 3: Lead a well-planned, purposeful learning experience
• EP 4: Maximize student ownership of learning
• EP 5: Respond to evidence of student learning