Learning Behaviors and Background Characteristics that Promote Retention of Women and Minorities in Undergraduate Computer Science

Funded by NSF EIA, 09/15/00 - 08/30/03 (Co-PI with Sandra Katz)

In this research project, we focus of how students' learning strategies and behaviors affect their performance in undergraduate computer science programs. Since programming is one of the first skills that computer students learn, and a stumbling block for many, we will focus on students' programming learning strategies. The specific issues we will address are

Do "unsuccessful" and "successful" students differ with respect t o how they prepare to accomplish a task, how they generate a program, and the extent to which they evaluate whether the program solves the specified problem?
How do the stable characteristics of race and gender relate to learning behaviors, achievement, and persistence?

""Success" will be defined by task specific measure such as completion time and number of errors, and more general measures such as course grades. We hypothesize that withdrawal from computer science programs can be explained, in part, by ineffective learning and programming behaviors. Furthermore, because women and minority students often enter computer science programs with less experience that white male students, we predict that women and minorities will be less likely to have well developed strategies for learning how to program.

To address these questions, will will first observe a small sample of computer sincere students as they carry out programming tasks in a laboratory setting. Our sample will consist of students taking computer science courses at the University of Pittsburgh, as well as students particiatping in a special program to promote the enrollment and retention of Pitt female and minority student called "link-to-Learn." We will have students "think aloud" during these programing sessions, and videotape them. We will then analyze the transcripts in order to differentiate "successful" from "unsuccessful" students in terms of how they went about studying relevant material, and how they carried out the task -e.g., did they elaborate on their lecture notes? Did they experiment with examples, testing and modifying them using various inputs? Did they evaluate their plans and the final program. We will determine if male and female students, white and minority students, differ with respect to these learning and programming behaviors. In order to investigate the relationships between learning behaviors, achievements and background (experiential) characteristics, we will issue a background survey to students who participate in our study. To test our finds about which characteristics correlate with effective learning behaviors and achievement. we will issue the same survey to computer professionals.

After we identify strategies and behaviors that characterize "successful" CS students, we will develop an intervention to train students to execute effective learning strategies. We will implement the intervention within the Link-to_learn program and test it in a controlled study. The evaluation will consist of a larger sample of female and minority students than the exploratory, think aloud protocol phase, and will be longitudinal.. We will compare students who receive the learning strategy training with a control group, over the course of their undergraduate CS training, with respect ot learning and programming behaviors, performance, feels of self efficacy and retention rates. Our finds will provide guidance for computer science instruction in general, and for meeting
the particular needs of female and minority students