The reasoning ability of 20 ninth grade, 20 eleventh grade, and 15 college freshman boys was explored. Three tasks in reasoning were given each student. In the "turtles task" the student was asked to test a hypothesis; from his responses the investigator determined whether the student tested the hypothesis by trying to find confirming instance or, logically, by trying to find the disconfirming instance. The "eye color-hair color problem," presented to the student as a game, asked the student to identify a relationship between two variables. The similar "bats task" asked the student to make a clear decision from the data. Results indicate that logical reasoning abilities developed with age, but it was clear that high school teachers should not assume that their students have these abilities. Even in the "bats task," the students indicated inability to deal with correlated data. The value in using the clinical method and interacting with the students was realized in that most students found ways to deal with the data, often coming up with the right answer but for the wrong reasons. (CS)