Syntactically correct code does not fall from the sky; the process that leads to a student’s first executable program is not well understood. At the University of Kent we have begun to explore the compilation behaviours of novice programmers, or the behaviours that students exhibit while authoring code; in our initial study, we have focused on when and what they choose to compile. By examining these behaviours, we have determined the most common errors encountered by students using BlueJ in our introductory course on object-oriented programming, how those students tend to program when in supervised laboratory sessions, and we have identified future directions of study driven by our initial observations. Our goal is to apply this research to the future development of BlueJ and instructional methodologies involving its use in the classroom. 1.

There is widespread concern about low pass rates on introductory programming courses. While considerable research has been carried out to elucidate the reasons for this situation, many of the parameters leading to success or failure in the subject remain unknown. This article describes the results of an experiment to test two cognitive characteristics that have been shown to be important in other conceptual areas: working memory space and field dependency. These are related to examination results of around 150 students on an introductory programming course at the University of Glasgow. The results show that whilst working memory space appears to have only a marginal influence on levels of achievement on the course, field dependency is an important factor in determining success. The implications of this on the teaching of the subject are discussed briefly.

Abstract not found

An ethnographic approach was applied to identify problem solving skills used in Software Design. The individual and groups of students were observed resolving software design problems. This was in the context of programming-in-the-small. A heuristic analysis was generated from these observations and the students' laboratory books. The analysis identified only 35% of the students formally created designs using abstraction and decomposition methods. 85% of the students completed their software design problems with no recourse to any conceptual design. These latter students could not formally verify their software. A further problem was recognized where 65% of the students were unable to differentiate between syntactic and semantic errors. The paper concludes with the importance of these findings f or the design of programming and software engineering curricula.