Introduction

- This project seeks to improve pass rates in core second year units involving programming skill. We wish to identify and examine factors that correlate with poor performance, in order to identify students at risk early in the year. We aim to improve the overall learning and conceptual understanding of programming as a whole in the degree program.

- Students who choose to take the information technology stream study SOFT1001 (Software Development 1) and SOFT1002 (Software Development 2). These cover the basics of programming. In second year the students do SOFT2004 which is used as a gatekeeper unit for senior software and networks units. This is the focus of the current project as it is a prerequisite for most third year software development and network topics. We aim to analyse the aspects of first year computer science that are essential for the success of second year studies and beyond.

- There is evidence that some first year university students function at Piagetian cognitive levels described as concrete level rather than the formal level1. Notably students often revert to such pre-formal operational levels when confronted with a problem outside their scope of experience2. The challenge is helping students move to the formal level required for programming competence. Computer science can be quite a large paradigm shift for students and often lacks any concrete operational level examples or instruction. If this is taken into account when developing remedial resources then the resources developed can be designed to focus on raising students understanding from a concrete level to a formal level.

Learning Outcomes

1. Ability to read and write correct, clean code in C that allocates, deallocates and manages memory.

2. Understanding of common memory-related errors (such as memory leaks and dangling pointers) and how to avoid these. [Higher performance could involve detecting errors in examples code, and fixing them using debuggers (see below)]

3. Ability to read and write code that correctly uses the main standard library functions, especially for I/O, file handling and string handling. [Higher performance could involve elegant use of these functions, particularly avoiding idioms which are extremely inefficient.]

4. Ability to correctly implement standard linked list data structures [Higher performance could involve slightly more complicated structures such as binary search trees]

5. Ability to use code quality strategies appropriate for C, including preprocessor techniques, and use of common idioms.

6. Experience of following a thorough automated testing regime using tools such as make, diff, scripts to present the outcomes, and a tool to manage regression testing. [Higher performance could involve ability to construct such a regime]

7. Experience in using debugging tools.

8. Understanding of the approach and concepts of Unix, including its tools, philosophy, processes (including pipes and redirection), the file system and the shell.

9. Ability to learn to use Unix commands and system calls (including usage of flags etc) from online manual system

Risk Factors

- One of the factors that has been identified as an ongoing risk for students passing SOFT2004 is their performance in the exam of the prerequisite course COMP1002.

- Students who scored in the forty to forty-four band had a three times higher failure rate, for SOFT2004 than those students who scored forty-five to forty-nine in the COMP1002 exam (see Fig. 1). Even more marked is the difference between students who scored in the thirty-five to thirty-nine and the later group who had over eleven times better chance of passing.

- The COMP1002 exam is designed so that students who fail to perform adequately in the first question (often referred to the barrier question) are not normally permitted to continue onto SOFT2004. The barrier question tests the students ability to iterate through a data structure and count the occurrences of a value. However when students pass rate in SOFT2004 is plotted against their mark in the barrier question for COMP1002 it can be seen that there is no correlation between the barrier question mark and the pass rate in SOFT2004 (see Fig. 2).

- This suggests that the barrier question is not serving its function as a gatekeeper question. The intention is to discriminate against students who show no programming ability, but even students who do very well (> 7) have a 15 % chance of failing.

COMP1002 Exam Mark vs SOFT2004 Pass Rate (with mark granularity of 5)

Figure 1 – COMP1002 exam mark vs. SOFT2004 Grade as a percentage with a mark granularity of 5.