The role of students’ characteristics, self-regulated learning and cognitive abilities on prediction of students’ academic outcomes in blended learning.

Introduction Recent technical advances in information and communication technologies (ICT), and other societal transformations have changed the way people communicate, interact, and are taught. The advent and exponential growth of the internet resulted in new trends and uses of media in the creation of learning environments in educational settings. Consequently, the learning environments are altered dramatically in terms of new opportunities and challenges to design, develop, and implement effective instruction. This trend affected even developing countries, including Albania. One of the biggest problems that we face nowadays is that students’ lack of competences, such as critical thinking, self-management and problem solving which are needed in our society (Melo, 2013), which are reflected in their academic outcomes. According to a World Bank Rapport (2017), higher education faces many perennial challenges, including expanding and promoting equitable access, improving learning achievement, fostering educational quality and relevance, strengthening knowledge and technology transfer, and encouraging desired values, behaviors, and attitudes. From an external perspective, quality is associated with higher education’s contributions to society, including economic and social benefits. From a student-centric approach, quality in learning focuses on student prior experience (Tam, 2001). The main characteristics of a student-centered approach are: students’ self-reliance, the role of the teacher as a facilitator and previous knowledge, and the main example of it is problem-solving and project-based learning knowledge (Dochy, Segers, Gijbels & Van den Bossche, 2002). According to Dochy ,Janssens and Gisele (2006) a student-centered approach aims to provide a deeper approach to students’ learning. Also, as Thorne (2014) stated, collaboration and interaction help in developing students’ intelligence: mobilizing attention, improving perception and memory. Based on the previous statements, blended learning helps students to overcome these difficulties and facilitate and support students’ learning processes by enhancing content learning, increasing accessibility for students in remote and regional areas, facilitating deeper scholarly enquiry, and encouraging staff to develop innovative, collaborative, and flexible teaching and learning practices (Graham, 2006; Milthorpe, Clarke, & Fletcher, 2017). Pelleth Yohannan (2010) found that online learning could provide students with opportunity for increased interactive engagement (more than that is normally possible in 'face-to-face-only' or 'online-only' environments), flexibility and cognitive scaffolding that enhanced their learning experience. Based on these researches, the aim of this study is to investigate the role of students’ characteristics and background (e.g., self-regulation, perception (P) motivation (M), perceived flexibility (PF), engagement (E) and cognitive skills (WM/attention and/or general intelligence), as well as learning strategies (LS) on students’ performance in blended learning (BL). Research objectives and hypothesis: The following objectives were formulated for the research: o To examine learner characteristics and background in a blended learning environment compared to those in a traditional learning environment. o To study the mediating role of general intelligence, WM, executive attention, learning strategies and SRL on learning outcomes in a blended learning environment compared to those in a traditional learning environment. o To determine the participation of students’ cognitive skills (WM, attention,), LS, self-regulation variables and background characteristics in predicting students’ learning outcomes in a blended learning environment compared to a traditional learning environment. This study is guided by four main specific hypotheses: Past researches have contributed to the understanding of the relationship between students’ characteristics background (self-regulation, perception (P) motivation (M), perceived flexibility (PF) and engagement (E), learning strategies (LS)) and cognitive skills (WM, attention skills ) and they found a positive relationship between them, but there are limited in terms of investigating their relationship in blended learning context in developed countries. (Musso, Boekaerts & Cascallar, 2019; Laer & Elen, 2016; Cascallar & Boekaerts 2006). According to Manwaring, Larsen, Graham, Brigham & Halverson, (2017); Laer & Elen, (2016); Dembo, (2004) students show higher levels of SRL factors, perception (P) motivation (M), perceived learning (PL) and engagement (E) in blended learning, and learner background characteristics influence on students’ academic outcomes ( Cascallar & Boekaerts, 2006; Musso, Boekaerts, Segers & Cascallar, 2019). Based on these previous researches the following hypotheses are formulated: o Students show higher levels of SRL factors, perception (P) motivation (M), perceived learning (PL) and engagement (E) in blended learning when compared to studying in a traditional learning (TL) setting. o Learner background characteristics will show a higher contribution to the prediction of students learning outcomes in blended learning rather than traditional learning Musso, Boekaerts, Segers & Cascallar (2019) pointed out that WM and attention influenced academic outcomes. Cobanoglu and Yurdakul (2014); San, Lim & Morris (2015; 2009); Ramirez‐Arellano, Acosta‐Gonzaga, Bory‐Reyes & Hernández‐Simón (2017; 2018) stressed the influence of WM and attention in students’ thinking, exploring the subject matter, sharing their opinions, discussing and appraising others’ opinions in blended context. . It was also shown that students gained different perspectives and were able to think more deeply and critically, and that students were able to transfer those skills to real-life education, when BL was implemented. Based in this research the following hypotheses can be formulated: o WM will show higher contribution than Executive Attention in predicting students’ academic outcomes. o Cognitive skills (WM, attention and/or General intelligence) will show a higher contribution to the prediction of students learning outcomes in blended learning rather than in traditional learning settings. Methodology: Methods and design A mixed methodological approach will be followed in this study. Therefore, quantitative and qualitative research methods will be applied, and a triangulation research design will be used. Therefore, qualitative results will be used to validate and to support quantitative results providing evidence in support of the validity and reliability of the research. Participants Students of the first and the second year of English instruction at the university level will be selected for the study. A non-random sample of students will be selected to participate in a survey. Two experimental and two control groups of freshman and sophomore students as a non-random sample (full class sections) will be selected to participate in this research. Instruments A structured questionnaire will be used to survey a non-random sample of students. A focus group-interview will be used to interview five focused groups of students, and a semi-structured interview will be used to interview the sample of lecturers. A survey instrument will be used to gather the data during the English language classes. Automated Operation Span (AOSPAN) and Attention Network Test (ANT) could be used to measure the WM and attention. LASSI could be used to measure learning strategies. (Kwong, Wong & Downing, 2009). Procedure and Data Analysis The results obtained from the administration of the various instruments will be summarized in a synthetic way to use as the basis for the analysis of the findings. A General Linear Model (GLM)approach and Structural Equation Modeling (SEM) will be used to study differences between groups, as well as the structural pattern of the results. ANOVAs analyses will be used to assess the differences in perceived learning, posttest, and learning outcomes means between the control and experimental groups based on learner variables. Repeated measures ANOVA will be used to test if differences exist for the perceived learning and actual learning outcomes over the time period between the ‘before’ and ‘after’ conditions in each semester based on the learner variables. The hypothesis that investigates the relationship between the blended learning approach, self-regulations, perception (P) motivation (M), perceived flexibility (PF), engagement, working memory and attention of students in English teaching and academic achievements of students will be tested using structural equation models. In addition, predictive models will be explored, using both traditional regression techniques (linear and non-linear logistic regression) as well as several machine-learning methods, with multilayer perceptron (MLP) neural networks with back propagation algorithms among them, to explore the prediction of learning outcomes in both modalities of teaching with the available information from the students, and determine the contribution of each set of variables to the correct predictive classification of the performance level. Other methods that will be explored include: decision trees, Naïve Bayes, Random Forest, and Gradient Boosted Trees. The study will compare results from these various methods and will discuss the optimization of the predictive classification by each approach, compared to classical generalized linear model approaches and logistic regression.