Email Status

Volume 15, Number 2
March/April 1999

The Human Cost of Over-Reliance on Tests


When students maintaining a C or B average in school receive "failing" or "needs improvement" on standardized tests, many people assume they are victims of low expectations and "social promotion" in schools. By this logic, the test scores reveal truths about students that their schools could not or chose not to see. The ultimate result of ignoring these truths, according to supporters of testing "gates," is that students will graduate from high school unable to get a job or be productive members of society.

Before school systems rush to implement policies of widespread grade retention or skill remediation based on performing below a cut-off score on a standardized test, they should take a look at the research that points to the ineffectiveness and human cost of such policies. For example, in the early 1980s, New York City instituted a program that resulted in the retention of all 4th- and 7th-graders who failed to meet a cut-off score on a standardized test, even after participating in special summer classes. A study conducted for the mayor's office found that it resulted in no greater achievement gains for retained students than for their low-achieving counterparts in previous years. Worse, the dropout rate for those held back was much higher than that of similar students who had been promoted.

Emerging research also indicates that such policies jeopardize approaches that are working. For example, Protech, a long-standing and well-developed school-to-career program in Boston accepts students with a C or above average and an 85 percent attendance rate. Most students who entered the program in 1993-95, scored in the lowest 40th percentile on the MAT, the standardized reading and math test given in Boston at that time. Yet, these young people have done very well in making the transition to college and careers.

This is an excerpt from the Harvard Education Letter. Subscribers can click here to continue reading this article.