Abstract:
Previous research has proved that motivation is influential to students’ performance at study, yet the intercorrelations are usually moderate. One underlying reason could be that in previous variable-centred analysis, participants with various statuses of motivation variables impact the association of variables. Hence, recent research has started to employ less common person-centred approaches to compare students with mixed motivational variables on their study and social-economic status. The current thesis adopted Programme for International Students Assessment (PISA) 2015 science database for seven jurisdictions (i.e., Australia, Chinese Taipei, Hongkong, Japan, Korea, New Zealand and Singapore). It analysed the data with two different person-centred approaches, i.e., two-step cluster analysis and latent profile analysis (LPA). The present study aimed to examine whether the results from two different analytic strategies align with each other and whether the findings are consistent with previous relevant person-centred research. Also, by comparing two different approaches, a better analytic method with consistently robust results was selected. Effect size (Cohen’s d) was computed to decide whether a cluster/profile was mixed or consistent. The significant difference in students’ characteristics was identified by pairwise comparison through ANOVA. Findings of research revealed that there was inconsistency in research results of two-step cluster and LPA. Cluster analysis showed that self-efficacy was more related to science achievement, while LPA showed enjoyment was more strongly related. Also, the similarity between the current research and previous studies was generally modest, as differences in analytic approach and decision-making about the best solution tend to prevent consistency in results. Results also indicated that among these seven cases, LPA was a slightly superior solution to cluster analysis, but consistency was not found.