Does making a test high stakes for students artificially inflate acheivement gap by race & gender? Evidence from the California High School Exit Exam
Abstract/Contents
- Abstract
- In a prior study of high school exit exams, my co-authors and I found that socio-economically disadvantaged students -- female, low income, English Language Learners (ELLs) and students of color -- are less likely to pass a high school exit exam than their more socio-economically advantaged peers, even when controlling for prior and concurrent achievement on a similar exam with no individual stakes for students (Reardon, Atteberry, & Arshan, 2011). These findings raise significant concerns about the accuracy of standardized test scores and the role they may play in reproducing societal and educational inequality. In this dissertation I use the same data to more thoroughly explore the achievement gaps on the high stakes California High School Exit Exam (CAHSEE) and low stakes California Standards Test (CST). I posit that the reasons we may observe a CAHSEE/ CST gap fall into three broad categories. The first category consists of statistical artifacts resulting from measurement error and violations of statistical assumptions. The second category of explanations consists of unintended consequences to policy decisions. These decisions may exacerbate achievement differences between groups but do not stem from performance differences between different groups introduced along with the high stakes of the test. If there is no evidence that the CAHSEE/ CST gaps are driven by true differences in ability -- either due to differences in student knowledge (including test prep) or differences in rigor of content or preparation—then these gaps are most likely the direct result of the "high stakes" nature of the exam. I use student level administrative data from three large California districts to examine these possibilities. I demonstrate that these measures are not driven by school segregation or measurement error using school fixed effects, multiple measures of ability, test scores shrunken to the group mean and instrumental variables. I find that women overperform on the English Language Arts (ELA) exam and underperform on the mathematics exam. Other groups underperform on both exams. Breaking down the exams by the content substrands, I find that the writing sample, which is included in the high stakes CAHSEE, but not the low stakes CST, drives the female overperformance on the ELA section. I conclude that the patterns of these gaps, including dramatic differences by school attended, are consistent with a stereotype threat explanation. In an era of increasing accountability for students, these performance differences on meaningful tests need to be considered both by policymakers as a possible negative unintended consequence to student level accountability and by educators as an impediment to the current of future success of their students.
Description
| Type of resource | text |
|---|---|
| Form | electronic; electronic resource; remote |
| Extent | 1 online resource. |
| Publication date | 2012 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Associated with | Arshan, Nicole Leigh |
|---|---|
| Associated with | Stanford University, School of Education. |
| Primary advisor | Reardon, Sean F |
| Thesis advisor | Reardon, Sean F |
| Thesis advisor | Cohen, Geoffrey |
| Thesis advisor | Loeb, Susanna |
| Advisor | Cohen, Geoffrey |
| Advisor | Loeb, Susanna |
Subjects
| Genre | Theses |
|---|
Bibliographic information
| Statement of responsibility | Nicole Leigh Arshan. |
|---|---|
| Note | Submitted to the School of Education. |
| Thesis | Thesis (Ph.D.)--Stanford University, 2012. |
| Location | electronic resource |
Access conditions
- Copyright
- © 2012 by Nicole Leigh Arshan
- 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...