Portfolios, Assessment and Equity

K-12 Testing

Proponents of performance assessments have argued that such assessments could reduce bias in testing in several ways, such as by providing students from diverse backgrounds with multiple and flexible means to demonstrate their learning and by ending the cultural and cognitive biases built into multiple-choice tests. On the other side, concerns have been raised that biases of teachers and scorers could influence scores on performance assessments.


Additionally, if performance assessments measure student ability to think and solve problems in a subject area, students who have had less opportunity to do so in their schools (who are likely to be disproportionately from low-income families or minority groups) will score lower, countering any positive effects from their having varied means to show their learning. Of course, using high quality performance assessments could enhance instruction in problem-solving and higher order thinking, particularly benefitting those student groups who have been least likely to receive such instruction. Over time, that should then lead to improvement in student outcomes.


In "Mirror, Mirror on the Wall, Which Is the Fairest Test of All?," Jonathan Supovitz and Robert Brennan contribute some valuable research to this complex issue by investigating the relative contribution of student background characteristics to performance on standardized tests and portfolios, and the size of the score gaps between kinds of assessments for different student population groups. The research was conducted on over 5000 students, who completed both the portfolio and the test, from all 261 grade 1 and 2 classrooms in Rochester, NY. The classroom portfolio system for language arts was relatively new, having been in place for two years at grade 1 and one year at grade 2.


Background equity characteristics (including race and gender) explained about 10 percent of the overall variance in student performance. They explained more of the variation in portfolios than in test scores at grade 1, but the reverse at grade 2. At both grades, the performance gaps between Black and White students remained significant, but were reduced by about half in the portfolio as compared with the standardized tests, a statistically significant finding. The gender gap, however, increased in favor of girls. (On the National Assessment of Educational Progress, the female over male score gap is wider on writing than on reading, and the portfolios may emphasize writing more than do the tests.) How Black boys fared as a result of these contradictory tendencies was not discussed. Disparities associated with limited English proficiency status and free-or-reduced lunch were similar for both assessments. (Because 80 percent of the students receive lunch subsidies, the study might not have been sensitive to income differences.)


Hopefully, similar large and careful studies will be done with older students and students and teachers who have participated in portfolio assessments for longer than one or two years.


Harvard Educational Review, V. 67, N. 3, Fall 1997.