Analysis of the Unit Testing Component of the Personalized System of Instruction

Abstract

The present study compared examination scores of 173 undergraduate students in a course taught by the Personalized System of Instruction under three conditions: required unit testing, optional unit testing, and no unit testing. Each of the six course sections was randomly assigned a different testing sequence in order to determine whether the success of PSI had been due to unit testing procedures or to the study questions. The dependent variable was a 65-item multiple choice examination administered at the end of each three-week phase of unit testing. The results of a Lindquist Type I ANO VA indicated a significant sequence by conditions interaction (p < .05). Tests of simple main effects showed that when required unit testing came first in the test sequence, the scores on the subsequent exam were significantly higher than the scores on the same exam following the other two unit test conditions. A Lindquist Type III ANO VA revealed a significant interaction between GPA and unit test conditions. Tests of simple main effects indicated no significant difference among the three unit test conditions on exams of students with GPAs in the upper 25 % of the class. Students with GPAs in the lowest 25%, however, attained their highest scores on the exam which followed the required unit testing, and scored higher on the exam following no unit testing than on the exam following optional testing.