SELF-ASSESSED EXPERIENCE AND GRADES
To give context for effects of pre-college CS experience on students’ grades in their courses, we first present the proportion
of students reporting different levels of experience via survey
question Q1. Figure 1 shows the breakdown of these experience levels by course. Overall, 18% of study participants report
having a fair amount of pre-college CS experience, 43% report
having a little experience, and 39% report having absolutely no
We studied the relationship between self-assessed pre-college experience level and final course grades, using a 4. 33 point
grading scale (A+= 4. 33, A= 4, A-= 3. 67, etc) 1 in each of the
courses in our study. The results are shown in Figure 2. We analyzed the differences between the mean grades for students
with the different experience levels using an ANOVA.
Figure 2 shows clear differences in average grades between
the students in the three experience level groups for several
courses, with more pre-college experience generally leading to
a higher average grade. Differences are more pronounced in the
for those with prior experience (n=411). We note that the
two-course introductory sequence employs many educational
best-practices and has been shown in prior work to improve
student outcomes and retention [ 18, 19]. Also in the lower-di-
vision, we studied the basic data structures (CS2: n=1117) and
computer systems/low-level programming (n=913) courses.
The two required upper division courses we included were ad-
vanced data structures (n=915) and programming languages
(n=381). The last course in the study was computer networks,
an advanced upper-division elective (n=132).
To measure pre-college experience, we administered a be-
ginning of term survey in each course. The survey was compre-
hensive, but for the purpose of this study we focus on the two
questions pertaining to students’ prior CS experience:
• Q1: How much experience had you had with
programming/computer science before taking your first CS
course at [this university]?
A1: Select one: Absolutely none / A little / A fair amount
• Q2: Please rate your level of experience with CS/
programming before taking your first CS course at [this
A2: Select all that apply: Took AP CS / Took another (non-AP)
CS course in high school or middle school / Took one or more
CS courses at a college or university / Self-taught one or more
programming languages (not including HTML) / Participated in a CS club or extracurricular activity / Learned HTML
(either in a class or on your own) / Participated in a summer
CS experience (e.g. summer academic program or camp) / No
experience / Other (please specify)
“AP CS” refers only to the AP CS A course, which is similar to our introductory CS course(s), as the AP CS Principles
course was launched two years after our data collection.
The survey also asked students whether or not they were
currently a declared CS major, and whether they entered our
university directly from high school or as a transfer student. In
our current study we included only students who entered our
university directly from high school because community college preparation varies so widely. Examining the performance
of transfer students is an important topic for future work.
To quantify student performance, we used final course
grades, persistence in the CS program (measured by whether
participants in the lower division courses went on to complete
Advanced Data Structures, the first required upper-division CS
course), and pre-test scores from two courses. The pre-tests,
given online as part of the first homework assignment in two
terms each of CS2 and Advanced Data Structures, were designed to test the prerequisite knowledge for each course. The
instructors of these courses modeled the pretest questions on
quiz questions that would have been given toward the end of
the preceding courses. Students were given full credit for attempting the quiz, regardless of their score.
Students’ final course grades, pre-test scores, and whether
or not they took Advanced Data Structures by spring 2017 were
associated with their survey results using de-identified coding.
1 Although our institution awards 4 grade points for both an A and an A+, we wished to
distinguish the higher level of performance afforded by an A+ in our study.
Figure 1: The distribution of self-reported pre-college experience levels
(Q1) by course in our study. Columns are normalized to 100%. Numbers
at the top give total n for each course across all terms in our study.
Figure 2: Mean grades for each course by pre-college experience level.
Error bars show standard error. Asterisks by the course name indicate
significant difference between experience levels’ means:
∗ = p < 0.05; ∗∗ = p < 0.01.