I sent the following to the California State Board of Education and the California Department of Education earlier today. About a year ago I read Tedra Osell’s blog entry that included information about the California STAR exam test booklet that was highly disturbing. I contacted a few education officials in California who gave disturbingly vague answers, and it took some while to track down confirmation. But I think the letter is self-explanatory:
January 16, 2013
Dr. Michael Kirst, Chair, California State Board of Education
Dr. Tom Torlakson, California State Superintendent of Public Instruction
Dear Dr. Kirst and Superintendent Torlakson,
I write you with concern after confirming that the California STAR test booklets include a structure that is likely to bias test results through the stereotype-threat effect. I urge you to take prompt action to change the test administration rules for the spring STAR exams and to change the test booklet format for 2014 and future years.
According to the test administrator’s manual for last spring’s test (at http://www.startest.org/pdfs/STAR.coord_man.2012.pdf) sections 6, 8, 9, and 10 require that either a test administrator or the student indicate the student’s gender, the student’s race/ethnicity, and the education of the student’s parents.
According to my understanding of the stereotype-threat literature, this is the type of test structure that can trigger the stereotype-threat effect. The research on stereotype threat is deep and extensive, represented well by Claude Steele’s Whistling Vivaldi (2010), which the current Stanford Graduate School of Education Dean wrote as a lay introduction to the research. When individuals are reminded even in subtle ways of stereotypes about underperformance of their own demographic group, that reminder by itself results in a lower performance on the task than if the person completed the task without that reminder or with more positive and constructive messages about effort.
There are several consequences of the demographic information questions and display on the STAR test booklet. One is the inaccuracy that flows from the biased test scores: some portion of California children have scores from prior years that are deflated by the stereotype-threat effect. The effect of that deflation is made arbitrary by the choice afforded schools to require that children mark their own gender, their own race/ethnicity, and the education of their parents — something that would very likely magnify the stereotype-threat effect.
Schools’ performance ratings are also biased by the stereotype-threat effect, likely moreso for schools with higher concentrations of racial minorities and children whose parents have lower educational attainment. And for those tests that are treated as promotional gateways or placement indicators by schools or the state, children’s educational opportunities are likely being restricted on a racially-disproportionate basis because of the stereotype-threat effect that the California STAR test booklets would induce.
I urge you in your capacity as chair of the State Board of Education1 to take immediate steps to eliminate the presence of these items on the test booklet. At the very least, the State Board of Education and State Superintendent should order that students are not to mark their own demographic information for the 2013 administration, and those sections should be removed from all future test booklets. There is absolutely no need to induce a stereotype threat on the highest-stakes tests that K-12 students face in California, and many reasons to avoid it.
I write as a proud graduate of a California high school from 1983, as a relative of many California residents, and someone deeply concerned that a fifth of America’s schoolchildren are asked to take annual tests under conditions that research clearly indicate provide unequal opportunity for demonstration of achievement.
Yours very truly,
Corona del Mar High School ‘83
- This is my error, forgetting to include Superintendent Torlakson here. [↩]