Empirical assessment of UML static object diagrams

Author(s): M. Torchiano
Venue: Proceedings. 12th IEEE International Workshop on Program Comprehension
Date: June 2004

Sample Size: 17


This paper sought to show that using object diagrams in addition to class diagrams was
better than using only class diagrams with regards to student comprehension. They set up
an experiment which they did as a class exercise in a grad SE class of 17 students.

The students were given four exercises with UML diagrams representing a simple program and
a questionnaire of multiple choice questions for that exercise. Each exercise was on a different
program. On each exercise, half of
the students received UML diagrams with both class and object diagrams and half received
UML diagrams with only class diagrams. As such, each student did two of the exercises
with both object and class diagrams and two with only class diagrams. In addition,
an additional post-experiment questionnaire was given to measure the students' perceived
usefulness of object diagrams.

Out of the 4 exercises, 2 showed that the students were able to answer more questions
correctly when they had object diagrams in addition diagrams in addition to class diagrams.
In the other two exercises, there was no statistically significant difference between
having or not having object diagrams. In no case was it worse to have object diagrams.
The authors concluded that object diagrams have a medium effect in aiding program comprehension.

All in all, the study showed the object diagrams can increase program comprehension but
do not necessarily do so. Whether object diagrams will be of benefit probably depends on the
nature of the program. The results from this study indicate that at minimum object
diagrams are not a hindrance to comprehension and since they can be of benefit, it would
probably be a good idea to use them in general.