Mobile agents hold significant benefits for the rapid expansion of Internet applications and current trends in computing. Despite continued interest, the promised deployment has not taken place, indicating a need for a programming model to introduce novice mobile agent programmers to this environment/paradigm. Accordingly the research question asked was, ”Since novice mobile agent programmers1 require a paradigm shift to construct successful systems, how can they be equipped to grasp the contextual issues and gain the necessary skills within reasonable time limits?” To answer the question, a complete reference providing contextual information and knowledge of mobile agent system development was compiled. Simultaneously novices are introduced to agent orientation. A generic mobile agent system architectural model, incorporating guidelines for programming mobile agents, further provides a framework that can be used to design a mobile agent system. These two structures are presented in a knowledge base that serves as a referencing tool to unlock concepts and knowledge units to novices while developing mobile agent systems. / Computing / (M.Sc. (Computer Science))

Mobile agents hold significant benefits for the rapid expansion of Internet applications and current trends in computing. Despite continued interest, the promised deployment has not taken place, indicating a need for a programming model to introduce novice mobile agent programmers to this environment/paradigm. Accordingly the research question asked was, ”Since novice mobile agent programmers1 require a paradigm shift to construct successful systems, how can they be equipped to grasp the contextual issues and gain the necessary skills within reasonable time limits?” To answer the question, a complete reference providing contextual information and knowledge of mobile agent system development was compiled. Simultaneously novices are introduced to agent orientation. A generic mobile agent system architectural model, incorporating guidelines for programming mobile agents, further provides a framework that can be used to design a mobile agent system. These two structures are presented in a knowledge base that serves as a referencing tool to unlock concepts and knowledge units to novices while developing mobile agent systems. / Computing / (M.Sc. (Computer Science))

The subject of iterative prototyping with data is still in its infancy and problems arise when people with low programming skills aim to build a full-stack interactive web application on their own. This research attempts to evaluate how providing a framework with predefined notations and data structures can improve this process for novice developers. The work consists of an initial reimplementation of a model website and the evaluation of study participants with little experience in the frontend development field through the application of the Cognitive Dimensions and the Think-aloud methods. We have found that by eliminating the requirement of dealing with back-end queries through the offering of such a framework, the process of designing an interactive web application can be facilitated and hastened. There are, nevertheless, some improvements that can be made to further refine their programming experience. / Ämnet kring att framställa iterativa prototyper med data är fortfarande i sin begynnelse och problem uppstår när personer med låg kunskap i programmering har som mål att bygga en full-stack interaktiv webbapplikation. Denna forskning utvärderar hur utvecklingsprocessen kan förbättras för nya utvecklare genom att erbjuda ett ramverk med fördefinierade beteckningar och datastrukturer. Denna studie består av en implementerad modell av en webbapplikation samt utvärdering av studiedeltagare med begränsad erfarenhet inom front-end-utveckling. Utvärderingen gjordes genom användning av ramverket “Cognitive Dimensions” med tillhörande “Think-aloud”-metod. Resultatet visar att genom att eliminera krav för att hantera back-end-anrop har processen att designa en interaktiv webbapplikation kunnat underlättats och påskyndats med hjälp av ett ramverk. Resultatet identifierade även förbättringspunkter som kan implementeras för att vidare underlätta ramverkets programmeringsupplevelse för användare.

This study was motivated by a desire to address the challenges of introductory programming courses. Ethiopian universities teach such courses in large classes (80+ students) and students complain about the difficulty of the courses and teaching variation of instructors. The study was set to explore optimum course and learning environment design approaches. The research question raised was: how can a blended learning approach be used to improve large class teaching of programming? In an action design research approach, the study was initiated with redesigning two consecutive courses and a supportive blended learning environment on the basis of existing learning theories and educational design frameworks. Two cycles of action research were conducted for a dual goal of refinement and evaluation of the intervention. The action research was conducted during the 2012/13 academic year with 240 students at the beginning. A predominantly quantitative first cycle of action research produced a mixed outcome. The students’ marks from assessment activities were fairly close to results from two other international universities. A pre- and post-implementation survey of students’ approach to learning showed a slight class level change towards the deep learning approach. Conversely, some students were found at-risk (not progressing well) and certain technologies, particularly program visualisation tools, were found underutilised. The second action research cycle aimed to explain the result from the first round. A grounded action research evaluation of data from focus group discussions, interviews and participants’ memos identified plausible factors for meaningful programming learning in a large class. These factors were use of collaborative and pair programming; alignment of learning and assignment activities; integrated use of e-learning; and use of large class strategies like student mentors and team teaching. A critical realist interpretation of the result of the action research suggested that students can learn programming in large classes, 200+ in this study, with a course and learning environment design that keeps them engaged in learning and assessment activities. The study concludes that improved learning of programming can be possible with the use of students as mentors and changed role-dynamics of instructors, which presupposes adaptation of suitable pedagogical approaches and use of technologies. / School of Computing / D. Litt. et Phil. (Information Systems)

This study was motivated by a desire to address the challenges of introductory programming courses. Ethiopian universities teach such courses in large classes (80+ students) and students complain about the difficulty of the courses and teaching variation of instructors. The study was set to explore optimum course and learning environment design approaches. The research question raised was: how can a blended learning approach be used to improve large class teaching of programming? In an action design research approach, the study was initiated with redesigning two consecutive courses and a supportive blended learning environment on the basis of existing learning theories and educational design frameworks. Two cycles of action research were conducted for a dual goal of refinement and evaluation of the intervention. The action research was conducted during the 2012/13 academic year with 240 students at the beginning. A predominantly quantitative first cycle of action research produced a mixed outcome. The students’ marks from assessment activities were fairly close to results from two other international universities. A pre- and post-implementation survey of students’ approach to learning showed a slight class level change towards the deep learning approach. Conversely, some students were found at-risk (not progressing well) and certain technologies, particularly program visualisation tools, were found underutilised. The second action research cycle aimed to explain the result from the first round. A grounded action research evaluation of data from focus group discussions, interviews and participants’ memos identified plausible factors for meaningful programming learning in a large class. These factors were use of collaborative and pair programming; alignment of learning and assignment activities; integrated use of e-learning; and use of large class strategies like student mentors and team teaching. A critical realist interpretation of the result of the action research suggested that students can learn programming in large classes, 200+ in this study, with a course and learning environment design that keeps them engaged in learning and assessment activities. The study concludes that improved learning of programming can be possible with the use of students as mentors and changed role-dynamics of instructors, which presupposes adaptation of suitable pedagogical approaches and use of technologies. / School of Computing / D. Litt. et Phil. (Information Systems)

This thesis describes a research project aimed at improving the tracing skills of first-year programming students enrolled for an introductory C++ course at an open distance learning institution by means of a tutorial in the form of a program visualization tool to teach the students to draw variable diagrams. The research was based on the findings from the BRACElet project (Clear, Whalley, Robbins, Philpott, Eckerdal, Laakso & Lister, 2011). A design-based research methodology was followed. To guide the process of developing the tutorial, a framework of 26 guidelines for developing and using visualization tools to teach programming was synthesized from the literature on computing education research CER, educational psychology and computer graphics. Guidelines were supplemented with reasons or explanations for their recommendation and considerations to be taken into account when using a guideline. The framework was enhanced by lessons learnt during the development and testing of the tutorial. The tutorial was tested and refined during two implementation cycles. Both cycles included quantitative and qualitative investigations. All students registered for the introductory module received the tool with their study material. For the quantitative investigations, students completed a questionnaire after using the tutorial. Through the questionnaire biographical data was acquired, the manner in which students used the tutorial and how they experienced using it. The responses to the questionnaires were statistically analysed in combination with respondents’ final marks. The statistical modelling indicated that the students’ biographical properties (a combination of level of programming experience, marks obtained for Mathematics and English in matric and first-time registration for COS1511 or not), had the biggest impact on their final marks by far. During the qualitative investigations students were eye tracked in a Human-Computer Interaction laboratory. The gaze replays in both cycles revealed that students’ reading skills impacted largely on their success, connecting with the findings from the quantitative investigations. Reflections on why the tutorial did not achieve its purpose; and why poor reading skills may have such a strong effect on learning to program, contribute some theoretical understanding as to how novices learn to program. / Computer Science / D. Phil. (Computer Science)

This study was motivated by the need to overcome the pedagogical hindrances experienced by introductory object-oriented programming students in order to address the high attrition rate evident among novice programmers in distance education. The initial phase of the research process involved exploring a variety of alternative visual programming environments for novices. Thereafter the selection process detailed several requirements that would define the ideal choice of the most appropriate tool. An educational tool Raptor was selected. Lastly, the core focus of this mixed method research was to evaluate undergraduate UNISA students’ perceptions of the Raptor e-learning tools with respect to the perceived effectiveness in enhancing novices’ learning experience, in an attempt to lower the barriers to object-oriented programming. Students’ perceptions collectively of the Raptor visual tool were positive and despite the fact that the sample size was too small to achieve statistical significance, these quantitative and qualitative results provide the practical basis for implementing Raptor in future. Thus providing learning opportunities suited to learner interests and needs, can lead to an enormous potential to stimulate individuals’ motivation and development in creating a more positive learning experience to overcome barriers in programming and enhance concept understanding to address the diverse needs of students in distance education that could lead to a reduced dropout rate. / Computing / M. Sc. (Computing)