ABSTRACT
This paper provides detailed description of students’ access to one critical educational resource, teachers that effectively promote learning. Using large-scale administrative data from North Carolina in grades 3–8 and value-added measures of effectiveness, I find disadvantages for poor, American Indian, African American, and Hispanic students, but disparities represent less than 2% of observed achievement gaps. Gaps are driven by differential risks of exposure to especially ineffective teachers, which occur between and within schools. The distribution of teacher-related learning opportunities therefore highlights White and higher SES students’ advantaged access to important educational resources as well as apparent limits to those advantages.
Supplemental data
Supplemental data for this article can be accessed here.
Acknowledgments
I am grateful for generous input on this research from many individuals, including specific comments related to this manuscript from Adam Gamoran, Eric Grodsky, Geoffrey Borman, Felix Elwert, Harry Brighouse, Jeffrey Grigg, Anna Haskins, Elizabeth Wrigley-Field, George Farkas, Andrew Penner, seminar participants at the University of Wisconsin-Madison, and attendees at the 2015 Annual Meeting of the American Sociological Association.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1. I use the terms “Black” and “African American” interchangeably. Results for multiracial students are not reported due to inconsistency in this category’s meaning over time.
2. Author’s calculation based on data collected by the National Center for Education Statistics for 2010–2011.
3. Value-added estimates for grade 3 are only possible for 2007–2010, when pre-tests were administered at the start of grade 3. In those cases, I treat pre-tests the same as lagged test scores in other grades (including allowing for interactions between prior achievement and grade level in the full model).
4. In the current data, I find that test proctors accurately reflect classroom teachers for more than 80% of students in elementary grades, but less than 30% in middle school.
5. This problem is expected even if student demographic characteristics are omitted from the estimating equation, given the association between these characteristics and prior achievement, which cannot be plausibly omitted. That is, if more effective teachers are assigned to non-poor students, then true effectiveness would be (positively) correlated with prior achievement.
6. Some research reports adjusted teacher effects estimates, in which raw estimates are “shrunken” toward the mean as a function of estimated reliability. Such adjustments are inappropriate in this context because they would attenuate estimated relationships between social background characteristics and teacher effectiveness (Chetty, Friedman, and Rockoff Citation2014a, see Appendix D).
7. Several additional critiques for policy purposes are less relevant here. Most notably, imprecision in effectiveness estimates – while critical to individual teachers – do not threaten the aggregate conclusions here. It may also be unfair to attribute effectiveness solely to the teacher, since they may reflect resources outside her control, such as available curricular resources (see Raudenbush Citation2004). Since these resources reflect relative learning opportunities, they are valuable components of relative disparities in opportunities to learn. My precise argument is that value-added estimates measure teacher-related opportunities, not solely teacher-caused opportunities.
8. The metric of standard deviations of teacher effectiveness corresponds roughly to that of an effect size for the disparity. Weeks of learning are calculated using the benchmarks provided by Hill et al. (Citation2008) based on typical growth in standardized achievement scores.
9. All corresponding estimates and those for alternate specifications are available in supplementary materials.
10. This calculation is based on a simple combination of the weeks of learning estimates in from Kindergarten to eighth grade (six elementary, three middle). The cumulative effects of teacher assignments on achievement at the end of grade 8 may be less, given decay in observed teacher effects over time. Although outside the scope of this paper, one important area for future research is how teacher-related learning opportunities impact students over time, especially for schooling influences on the development of within-group inequality.
11. Manfield’s (Citation2015) detailed analysis of disparities in teacher effectiveness in high schools illustrates a trade-off between leverage for identifying teacher effectiveness and how wide a population can be considered. A strength of the study is that it capitalizes on teacher transfers to identify parameter estimates, but as a result, analyses of inequality are based on 386 of 1000 total schools. It is notable that the general conclusion of that paper, small disparities in the expected directions, are similar to the current design, which includes a larger proportion of students.
Additional information
Funding
Notes on contributors
Paul Hanselman
Paul Hanselman is an Assistant Professor in the Department of Sociology at the University of California, Irvine. His research focuses on how schools contribute to social stratification and the impacts of interventions to mitigate educational inequality