R. J. Manson and R. J. Olsen / Procedia Computer Science 00 (2010) 1–9 4
3.3. Tools for Assessment: A Pancurriculum Rubric
The aforementioned rubric was first used in PHYS 3220 in the Spring ’09 semester and contained six categories
(accuracy, learning and understanding, narrative, introduction, analysis and conclusion). Each category was assigned
a number from 0 to 3, giving a total score that ranged from 0 to 18. Despite providing a somewhat more objective way
of assessing computational science projects, the rubric was often found to be too coarse-grained. Since the second
author was assigning projects in CPLS 2110, the idea of developing a robust pancurriculum rubric arose.
The authors worked together to modify the original rubric, subdividing the categories to provide a more fine-
grained instrument. The modified rubric had 18 categories evaluated on a 0–10 scale, giving a maximum possible
score of 180. Table 3 contains a sample category. To test the modified rubric the authors created a pool of ten
exemplary projects, five each from the most recent instances of CPLS 2110 (Fall ’09) and PHYS 3220 (Spring ’09).
This pool was then graded by both authors.
Table 3: Example rubric category. The scale is labeled by the category name and followed by guidelines describing excellent (9–10, A range), good
(7–8, B range), satisfactory (5–6, C range), and poor (0–4, D and F range) work.
model
construction
10
9
8
7
6
5
4
3
2
1
0
The flow of information in the model is easily followed in an excellent project, is followed with some effort in a
good project, is followed only with significant effort in a satisfactory project, and is followed only with considerable
difficulty in a poor project.
4. Results and Discussion
4.1. Assessment Outcomes: CPLS 2110
14 of 17 students (82%) answered questions about slope and y-intercept of a line correctly. Given “noisy” data
following exponential, linear, and sinusoidal trends and equations for each function, 14 of 17 students (82%) correctly
matched the exponential data and function, 13 of 17 students (76%) correctly matched the linear data and function,
and 16 of 17 students (94%) correctly matched the sinusoidal data and function. Of the seven students who answered
one or more of these questions incorrectly, three withdrew from the course almost immediately, one withdrew early
in the semester, and the remaining three obtained the three lowest course grades.
On the question involving rates of change and slope at a point on a curve, 13 of 17 students (76%) identified the
minimum with a slope of smallest magnitude and 2 of 17 students (12%) identified the maximum with a slope of
smallest magnitude. Just 4 of the 13 students selecting the minimum realized that the maximum should be likewise
identified, and neither of the two students selecting the maximum realized that the minimum should be likewise iden-
tified. Although the question clearly stated that more than one point could be associated with a label, it is likely that
habit built from years of test taking caused students to answer this question quickly rather than carefully. Redesigning
the question to ask students to rank the rate of change at each point may provide better information. 10 of 17 stu-
dents (59%) correctly identified the function as having a positive rate of change at the point with positive slope and
a negative rate of change at the point with negative slope. Several incorrect responses appear to be due to differences
between the slopes at the labeled points being insufficiently pronounced; a graph with more distinct features will be
used in subsequent versions of the assessment.
Figure 1 demonstrates that course grade correlates well with the score on the diagnostic assessment. One student,
represented by the open circle, fared considerably worse in the course than performance on the diagnostic would
imply. The diagnostic provides relatively little resolution at the top end of the scale, as indicated by the clustering
of points at diagnostic scores above 90. The self-assessments revealed by the survey of Table 2 were not positively
correlated with course grade.