The relationship of problem-solving ability and course performance among novice programmers

Reviewer: David C Haddad

The purpose of the paper is an attempt to identify problem-solving skills and other factors that predict success for a college student enrolled in computer science courses. Approximately 300 students enrolled in introductory programming, elementary COBOL, or advanced level computer science courses were given a test at the beginning of the course. The test contained seven problems to be solved, and requested information on the student's previous academic performance in mathematics and language courses, previous computer programming experience, attitude toward computer science, and personal locus of control. (The term “personal locus of control” refers to the view that either one is in control of his or her actions or is responding to primarily outside influences.) The problems tested the student's ability to understand and represent the problem in a solvable form (e.g., algebraic form) and his or her ability to approach this representation with a solution procedure. The problems were of several types: representation of word problems in algebraic form, identification of problems that had similar solution strategies, state-space problems (e.g., missionary-cannibal problem or Tower of Hanoi problem), and problems involving inductive and deductive logic. The paper uses statistical analyses to conclude that performance on the test is related to the performance in the course indicated by the final grade in the course. The study found significant differences in problem solving performance between students in elementary classes and those in advanced classes. Significant differences in problem solving performance were also found between students receiving a grade of “A” and those receiving a grade of “D” or “F” in the elementary courses. No such differences were found for students in the advanced courses. Some problems were better at predicting course grade than others. Other factors that were related to differences in grades include previous computer programming experience and previous academic performance in math and English courses. The paper extends the findings of Cheney [1] and those of Kurtz [2]. Although the paper presents no startling conclusions, it is interesting and informative. The author does provide a nice list of references for the reader interested in this or related subjects.