Subjects and procedures
The participants in this study were self-selected volunteers in Year3, Year 4 and Year 5 of a six year undergraduate program at the University of Auckland, New Zealand. The first part of this study was carried out over a two-week period between June and July 2014. At the end of a lecture, students were given a self-report Likert-type questionnaire composed of validated measures of academic motivation, burnout, and quality of life. Contained within the questionnaire was the Academic Motivation Scale, which measured intrinsic motivation, extrinsic motivation and amotivation, and subscales of the Motivated Strategies for Learning Questionnaire (MSLQ), which measured self-efficacy and test anxiety [14, 15]. The Academic Motivation Scale and MSLQ have been previously used in medical education research, both internationally and in New Zealand, and among a University of Auckland medical student population [5, 16, 17].
Consistent with studies by Kusurkar et al. , the author modified the Academic Motivation Scale, which was originally designed for college and university students, so that it could be applied to medical students and further checked the reliability of each scale. Intrinsic motivation scores were calculated from the Academic Motivation Scale as an average of the intrinsic motivation scores on the three subscales . Extrinsic motivation scores were calculated by taking an average of introjected regulation and external regulation scores . The identified regulation subscale was not included within calculations and subsequent data analysis as the items on this subscale are such that most students in professional education would answer positively . Therefore, this subscale is not likely to discriminate within a medical student population .
The questionnaire also contained a World Health Organization quality of life questionnaire (WHOQOL-BREF), and the ‘personal burnout’ scale from the Copenhagen Burnout Inventory to measure quality of life and burnout respectively [19, 20]. Both the WHOQOL-BREF and the Copenhagen Burnout Inventory have been used in a number of quality of life and burnout studies during medical training, and the WHOQOL-BREF has been previously used among the University of Auckland medical student population [5, 21, 22]. The study questionnaire also contained a survey of student characteristics (including age, gender, admission scheme into medical school, and year level of the curriculum).
The WHOQOL-BREF questionnaire included 24 items that encompass four quality of life domains (physical health, psychological health, social relationships and environmental conditions). The scores for each domain were calculated using a well-recognized WHOQOL-BREF syntax . The personal burnout subscale of the Copenhagen Burnout Inventory contains six items with scoring from 0–100 for each item. The total score on the scale was the mean of the scores on the items .
The second part of this study was carried out in October 2014, which included collecting student academic performance data based on progress tests completed in April, July and October 2014 respectively.
Ethics approval and statistical analyses
Written informed consent to participate was obtained from all students, and ethics approval was obtained from the University of Auckland Human Participants Ethics Committee (UAHPEC Ref 8467).
All statistical analyses were performed using IBM SPSS 22.0 for Mac. Internal reliability measures, Cronbach’s alpha coefficients, for each section of the questionnaires were determined.
Participants were classified to different profiles based on WHOQOL-BREF and Copenhagen Burnout Inventory scores using a two-step cluster analysis . The two-step clustering method within the auto-cluster modelling node was chosen because of its ability to handle both continuous and categorical variables. The two-step cluster analysis method operates through firstly scanning the data in a pre-classificatory stage and identifying ‘dense’ regions of data that share similar values across a range of variables . An algorithm similar to an agglomerative hierarchical clustering method is then used to classify the data . The algorithm used the log-likelihood distance measure and Schwarz’s Bayesian Criterion to derive the cluster solutions (burnout and quality of life profiles) by maximizing between-group heterogeneity and within-group homogeneity, thereby capturing the interactions between dimensions of quality of life and burnout.
Once the burnout and quality of life profiles were derived, Chi-square analyses were conducted to determine any significant differences in characteristics between profiles. Any significant differences were then included in a multivariate analysis of covariance (MANCOVA) model as covariates.
A MANCOVA model was used to determine differences between profiles in relation to academic motivation, self-efficacy and academic performance. Profile membership was included in the model as the independent variable, and the dependent variables were intrinsic motivation, extrinsic motivation, amotivation, test anxiety and self-efficacy scores A correlation matrix that presents the correlations between the dependent variables in the MANCOVA model is included in Table 1. Post hoc multiple group comparisons were adjusted using the Bonferroni correction for controlling Type 1 error. The effect size was calculated from partial eta squared: small = 0.0–0.06, medium = 0.06–0.138, large >0.138 .
A separate repeated measures analysis of covariance (ANCOVA) method was also used to compare changes in academic achievement over time. Profile membership was included in the model as the independent variable, and the three progress test scores were designated as dependent variables. Year level and gender were included in both models as potential confounding variables.
Pearson’s correlations were also used to determine any associations between quality of life domain scores, and academic motivation scores.