“Are we reading too much into reading comprehension tests?” asks Robert J. Sternberg, IBM Professor of Psychology and Edcuatin at Yale University. Increasingly, test scores, including reading comprehension test scores, are being used by educational administrators in decision-making. Increasingly, also, schools and teachers are being held accountable for the results of the tests they are required to administer. Such use of test scores, Sternberg believes, would be appropriate if the tests accurately reflect the skills and abilities we wish to measure.
However, Sternberg reminds educators that most tests measure only a narrow range of student aptitudes and achievements and, as a result, test scores can only provide an incomplete view of student performance. While many tests, including reading comprehension tests, have merit, Sternberg warns that we must be considerably more careful in our interpretation of test scores.
Tests measure only narrow range of reading comprehension
In Sternberg’s opinion, currently available reading comprehension tests are not as valid a measure of real world reading behavior as most people believe. He writes that comprehension tests measure not only a very limited range of reading skills, but that these are not the kind of skills used most frequently in school or daily life. Sternberg believes that we must develop tests that stimulate more accurately the kinds of abilities and skills students need to function well in daily life.
One problem, according to Sternberg, is that most reading comprehension tests use a multiple choice format. Because this type of question requires only a certain kind of reasoning, the test can measure only a narrow range of a student’s abilities. Current reading comprehension tests require that students learn information in a short space of time and recall that information immediately afterward. In these tests, students are asked to assimilate only short passages of very well-organized but often uninteresting material. The testing situation itself is an artifically controlled environment that limits distractions and focuses the student’s attention on the task. Such an atmosphere is unlike that in which students normally are expected to read, learn and recall information, either within classrooms or outside school. Reading comprehension tests, Sternberg believes, simply do not measure the same kind of reading that occurs in the everyday world.
Most learning, he states, is incidental and carried on over a long period of time. It requires students to read and integrate information from much longer texts. The kind of cognitive processing required is, therefore, considerably different than that which comprehension tests measure. The short-term recall of information measured by these tests is not predictive of the long-term recall needed for successful learning in or out of school and Sternberg speculates that the type of comprehension and memory storage required of each is probably very different. “Test-wiseness” is also a very important factor in students’ test performance.
Sternberg concludes that multiple choice reading comprehension tests have limited predictive value. The test situation differs so greatly from the way reading and learning occur outside the test environment, that it is difficult to generalize performance from one situation to another. Sternberg maintains that if we want to predict real-world performance and not just test performance, then we must measure a broader range of reading behavior. We need to understand and assess reading as it occurs naturally and not just in artificially contrived testing situations.
“Are We Reading Too Much Into Reading Comprehension Tests?” Journal of Reading Volume 37, No. 7, p. 540-545.
Published in ERN May/June 1991 Volume 4 Number 3