The purpose of this research is to understand why so many students find learning to program difficult. I studied novice computer programmers in (close to) their natural environment at university. This research provides evidence that the development of programming skills can be described using the neo-Piagetian cognitive development framework. While others have used classical Piagetian theory to talk about learning to program, the neo-Piagetian framework has not previously been used empirically to study programming behaviour. Understanding how novices develop programming skills will enable educators to better assist those who struggle to learn programming. As neo-Piagetian theory is not age-specific, this research informs not only tertiary pedagogy, but teaching and learning of computer programming in any setting.
A longitudinal study which forms part of this thesis shows that the development of programming skills is sequential and cumulative. Novice programmers pass through at least three stages: sensorimotor, preoperational and concrete operational. An overlapping waves model of development was adopted to explain concurrent manifestation of two of those stages. The most mature stage, formal operational, was not considered in depth in this thesis.
I completed both qualitative and quantitative data analysis. The qualitative data was from novice programmers thinking aloud as they completed programming exercises. The quantitative data was collected from in-class testing of entire cohorts of introductory programming students. An analysis of this triangulated data has identified the behaviours of sensorimotor and preoperational novices in contrast to the more mature concrete operational novices.
It was found that sensorimotor and preoperational reasoning behaviours are evident in novices very early in semester. Further, even by the end of that first semester, many students had not yet reached the concrete operational stage.
This research explains why some students have a hard time learning to program. It is not because they are not smart enough; not because they do not have a "geek gene"; and not because programming is an intrinsically difficult subject. Programming competence requires abstract reasoning skills, and learning to program is about the sequential and cumulative development of those abstract reasoning skills in an unfamiliar domain. This thesis identifies the manifestations of each of the early stages of development in the programming domain. That information will enable educators to (a) identify the developmental stage of their students, (b) provide stage-appropriate learning resources and (c) assist students in transitioning to the next more mature stage of reasoning.