Author:

Keywords:

Academic Performance, Cognitive Processes, Slef-regulated Learning, Task Characteristics

Abstract:

Research from cognitive and self-regulated learning approaches has contributed to the understanding of how students regulate and achieve certain performance outcomes in educational settings. During the last decade, more comprehensive models emerged supporting the integration of findings from different perspectives. The important role of working memory and attentional systems for the general processing of information, and specifically for math performance, is well known. Likewise, motivational beliefs such as subjective competence, cognitive and affective self-regulation, influence general academic performance and also performance in specific domains. Yet, we have little understanding about how these cognitive and non-cognitive variables interact with each other. Moreover, there are controversial results about how much they can predict academic performance from an integrated SRL model. The majority of previous studies have not considered cognitive individual differences at the information processing level. The goal of this research was to analyse the relationships between cognitive and motivational underlying factors that could explain two educational outcomes in the higher education context, focusing on general academic performance (GAP), on the one hand, and on math performance (MP), on the other. Since we understand that the performance level depends on the fit of an individual’s cognitive profile with the demands of the task, another aim of this research was to analyse these cognitive and motivational effects taking into consideration two characteristics, complexity and difficulty of the items of a math test. This doctoral dissertation involves 6 articles: a literature review and five studies based on a large sample of first-year university students. The first section introduces the relevance of the problem and the main research questions. The literature review identified the most important cognitive and non-cognitive factors impacting academic performance. Moreover, this study allowed to set up a conceptual map of the relationships between these factors, involving the main selected variables for this research (working memory, executive attention, self-regulation factors, learning strategies, characteristics of items, individual background variables). Chapter 2 analysed the interactions of cognitive group (defined by the combination of different levels of Working Memory and Executive Attention) with different components of Self-Regulated Learning (SRL) in their impact on MP. This study showed that cognitive group (CG) has the strongest effect on MP, but subjective competence and appraisals interact with cognitive group in their impact on MP, depending in part, on the complexity and difficulty of the items. Chapter 3 analysed the relationships between CG, self-regulated learning factors and learning strategies, focusing now on GAP represented by the Grade Point Average at the end of the first year. The results showed that students with high working memory (WM) performed significantly better than students with low WM, independently of the Executive Attention level. Several main effects were found for various learning strategies and for achievement motivation, but only one learning strategy interacts with cognitive group on GAP. Both chapter 2 and 3 showed that, WM directly influences academic performance, and EA compensates under some conditions, but it does not substitute WM, supporting the relative functional independence hypothesis regarding both cognitive mechanisms, in spite of their high correlation. Chapter 4 and 5 introduced Artificial Neural Networks (ANN) as a relatively new methodological approach to develop predictive and architectural models for the field of learning and education. Both studies used cognitive and non-cognitive measures of students, together with background information in order to design predictive models for different levels of MP and GAP, respectively, using ANN. Results demonstrated the greater accuracy of the ANN methodology compared to traditional methods such as discriminant analyses. In addition, the ANN provided information on those factors that contributed the most to the prediction of the different levels of expected performance. Chapter 6 tested several mediation models for both math and general academic performance using Structural Equation Modelling (SEM). The tested models obtained a satisfactory data fit, especially in MP. The MP and GAP models were analysed and contrasted in terms of their direct and indirect effects of the variables included in the models. Finally, a last chapter provides a general discussion of the findings, with conclusions and implications for education and further research.