The fourth-grade sample assessment test question on the Feb. 26 front page is a horrific example of a badly constructed test question. It not only requires a sophistication in English grammar beyond many 9-year-olds and thus misses a chance to truly test their mathematical understanding (the point of the question), but it is biased toward children growing up in sophisticated English-speaking, middle- or upper-middle-income homes.The question was: "By how much would the value of 5,647 be decreased if the 5 were replaced by a 2?"
This could easily be stated more plainly and measure arithmetic skills more accurately. "In the number 5,647, if you took out the 5 and replaced it with a 2, how much less would the new number be worth?"
Jerome Kagan, a Princeton psychologist, deconstructed many such test questions in his 1970s research to identify cultural bias and show that white and financially well-off students usually perform better than others on standardized tests because of the unconscious biases of test constructors.
On the basis of one test item, I do not want to assume bias in an entire test. Yet it is certainly worth a look by educators trained in avoiding cultural, racial, and socioeconomic bias. We owe it to our society to help all children reach their fullest potential.
—Sherry Zitter
Newton
[Ed.: A test question is relatively difficult, therefore unfair to those minorities who do not do as well on tests. For the test to be made "fair" by this measure, we must thus make it easier to pass. Note that standardized tests were originally set up as a progressive, meritocratic device designed to objectively evaluate the performance of immigrant children in an era when WASPs kept elite academies ethnically exclusive.]
