An investigation into professional programmers' mental representations of variables

Abstract

Very little is known about professional programmers' mental representations of variables, yet this information is vital in designing effective tools for program comprehension. In order to find out what types of information programmers have about variables and their relations, we conducted a knowledge elicitation study where professional programmers studied programs and the resulting mental representations were elicited using card sorting and interviews. The mental representations were based on fourteen principles that can be organized in four main categories: domain-based, technology-based, execution-based, and strategy-based. Most frequent information types dealt with two execution-based criteria: behavior and utilization.