Tertiary Education Choices in New Zealand: A Pluralistic Investigation

Most policy regarding tertiary training choice is based on the presumption that human capital theory is the most appropriate model and the cost-benefit analysis implicit in that model provides a reasonable approximation of the choice parameters included in any decision to study. This has certainly been the approach that underlies the significant changes in the New Zealand tertiary sector in the last twenty years. Unfortunately it has failed to explain the observed enrolment trends. This study starts from that point and discusses the various shortcomings in this approach and relates these to solutions provided in the behavioural literature found in other areas of economics or in other behavioural sciences. Through this a pluralistic methodological approach is chosen that can be used to supplement the more traditional human capital model by the inclusion of behavioural variables. It does so through the collection of primary data that specifically addresses human capital, behavioural and expectation variables. Firstly it finds variables commonly used in human capital models are do explain most of the variance between the various costs and benefit variables considered and this is consistent with the human capital literature. However, it is also found that this only holds when these variables are considered in isolation of behavioural variables. The addition of behavioural and expectation variables used in this study finds the variance explained by these HCT variables falls off dramatically and instead the behavioural variables explain far greater variance. These findings then give an indication as to the types of variables that need to be included in extended models if a better explanation and predictor of the tertiary choices made by individuals is to be derived. It is found that behavioural models such as the Fishbein model of behavioural intentions and Kelly’s personal construct theory both provide useful predictors of the different choice possibilities in terms of the decision to study, the institution type chosen and course level. It finds that the behavioural variables, in particular the expectation variables, give rise the far greater predictive discrimination than the HCT variables do. The models used in this work enable discrimination at the institutional level to produce very good results, significantly better than by chance or by human capital theory. This is even when the participants have very similar demographic characteristics that do not allow for any separation. Secondly the predictions of the Fishbein behavioural intentions model were supported in that the more defined and specific the intention or action the stronger the views toward it and the higher likelihood of action following. This again has policy implications that are reflected in the growing popularity and number of specialised degrees that deliver access to very specific parts of the labour market even though this could have conflicting labour market outcomes in different economic climes. This direct labour market link was found to be important even when the participants held little knowledge of the realities of the labour market. Lastly it was found the key constructs were those around self-esteem and expectations of tertiary training. This has important policy implications if the goal is to improve enrolment rates from currently under-represented groups. Potential students need to view themselves as above average in order to consider tertiary training. The corollary of this is that there will always be a percentage of the population that will be potentially disenfranchised through tertiary training. Clearly not everyone is able to view themselves above as average. While the study provides some very useful discriminant factor analysis, shortcomings in the sampling frame used in this study restrict the application of this to course level choice though the expectation is that this could be achieved with improved data collection strategies. The sampling difficulties also restrict the application of these results too widely among the tertiary population. However, the factors identified are very robust and do provide a way forward for further, more widely applicable study that has a greater focus on dependence methods rather than the interdependence methods used in this work.