The research constitutes a longitudinal study of why students underperform in the early years of tertiary engineering courses. It is claimed that, with the significant changes in engineering education (and in tertiary education in general) in Australia in the closing decades of the 20th century - changes that resulted in different curricula, and different staff and student profiles - university teaching styles often do not match students' learning styles. This mismatch is a likely contributing factor in the loss of some potentially good students. It is also claimed that it is feasible to develop a predictive model to identify students who are at risk of either not completing the undergraduate engineering program or performing below their innate capability. The thesis draws links between experiential learning, personality type and preferred learning styles, and explores the educational ramifications of these links for engineering.
From the beginning the study was seen as the first part of a much bigger picture; namely, the development of an instrument for screening students entering the undergraduate engineering program at tertiary level. The entire process was seen as comprising three stages:
(i) a feasibility stage to develop a mechanism to identify students who will be 'at risk' of not completing their undergraduate engineering studies;
(ii) a refinement stage where the mechanism developed is expanded by adding life experience factors such as cultural background and prior engineering knowledge, and where a system of weightings is devised to indicate the extent to which these factors affect performance and
(iii) the development, production and validation of the instrument constructed from the findings in (i) and (ii).
The feasibility stage, the subject of this research, tests three metrics as possible predictors of risk:
(i) OP (Overall Position) Indicator, a rank order score allocated to students completing Year 12 in Queensland schools;
(ii) LTM (Learning Type Measure), a self-reporting instrument in the form of a questionnaire, which claims to measure preferred ways of learning; and
(iii) MBTI (Myers-Briggs Type Indicator), another self-reporting instrument, which claims to measure psychological preferences.
Other data used are from information held in the University's student records database. Information including a student's psychological profiles and personal details is analysed in order to highlight factors that could be used to identify 'at risk' students. While these data were collected quantitatively, they were analysed primarily by qualitative methods.
Analysing the intake as a whole for the attributes that students bring with them when they enter university would not explain the reason for student failures, since factors such as the possession or not of an OP, the discipline the student becomes affiliated with and the course in which the student enrols could also have an effect on their performance. Consequently, a series of analyses was needed. The first step in the analysis was to examine the whole group in terms of their OP, LTM and MBTI, then by their respective disciplines (Civil, Electrical and Mechanical). Analysis at course level was not undertaken due to lack of numbers. Comparisons in terms of enrolment status, GPA and success rates were made between the OP and non-OP cohorts at cohort and discipline levels.
Findings from the analyses are that there is a link between success (GPA > 3.7) and having an OP in the range of 1 to 8, being LTM Types 2 and 3, and being MBTI Type ENTJ or ISTJ. Likewise, there is a link with being 'at risk' (GPA < 3.7) and having an OP of 9 and above, and being LTM Type 1 and MBTI Type ENTP.
Two predictive models are proposed - a generic, multi-dimensional model, which can be used for all students entering first year engineering at tertiary level; and a specialised model that can be used for students who have an OP score. The specialised model is useful because, even though OP students are subsumed in the generic model, the presence of an OP score, when combined with the LTM and MBTI, enables a reasonably accurate predictive model to be constructed without recourse to other life experience factors. While the models developed in this thesis have been designed for use with incoming engineering students, concepts embodied in them should find wider application over all courses of study.