The effects of comments and identifier names on program comprehensibility: An experimental investigation

The effects of comments and identifier names on program comprehension are examined. A number of studies have been undertaken to investigate the independent impact of comments or identifier names on comprehension but here the combined effect of comments and identifier names on program comprehension are also considered using controlled experimentation Based on existing program comprehension theories, three hypotheses were formulated: (i) commented programs are more understandable than non-commented programs; (ii) programs that contain 'full' identifier names are more understandable than those with abbreviated identifier names; and (iii) the combined effect of comments and identifier names tend to enhance the understandability of a program more than the independent effect of comments or identifier names. Eighty-nine undergraduate computer science students, divided into four groups, were presented with four different versions of a Modula-2 program and two methods were used to measure each participant's level of understanding of the program - an objective multiple-choice test and a subjective scoring procedure. A two-way analysis of variance of the results revealed that only hypothesis (i) was supported in the light of the objective test scores but only hypothesis (ii) was supported when the subjective scores were analysed. The discrepancy in these findings raises questions about the reliability of using just a single method to measure program comprehension and the use of statistical analysis on program comprehension data. Problems inherent in experiments such as those reported in this study are raised, and warnings against a simple reliance on their findings are given.