Cognitive, metacognitive and motivational perspectives on learning analytics : Synthesizing self-regulated learning, assessment, and feedback with Learning Analytics

Increasingly more students with growing heterogenous background are enrolling in higher education. Due to limited resources individual support is marginal. Furthermore, learning in higher education is evermore facilitated through technology. However, both higher education and digital learning environments are considered to be less structured. Hence, learners need to use strategies to self- regulate their learning processes. Such activities include cognitive, metacognitive, and motivational components, and are considered to take place in three cyclical phases, the forethought, performance, and self-reflection phase. However, learners often do not show such strategies spontaneously. Hence, timely support meeting learners’ needs is required. When learners are using digital learning environments they produce trace data. Learning analytics enable to analyze learning behavior and learning environments which can be used for understanding and optimizing learning processes and environments and to support educational decision making. To receive valid insights and derive appropriate interventions learning analytics need to be grounded in theory on learning, including motivation, assessment and feedback. However, currently learning analytics are lacking this theoretical foundation and empirical evidence. Thus, the overall research question of this thesis is how cognitive, metacognitive and motivational components of learning and theory on assessment inform learning analytics and vice versa. The thesis includes three quantitative studies (studies 2, 3, and 4), and one qualitative study (study 1). To enhance the theoretical foundation of learning analytics the thesis comprises one integrative review (paper 4) focusing on the link of learning analytics to theory on assessment and feedback with regard to self-regulated learning. As learning analytics should support learning processes and the learners have a central role the first two studies investigate students’ expectations towards features of learning analytics. With regard to self-regulated learning potential features of learning analytics were assigned to the three phases forethought, performance and self-reflection. Learning analytics mostly use dashboards to provide feedback to learners. However, how learners interpret and react to feedback depends besides the quality and level of the feedback also on their individual characteristics such as prior knowledge, attributions or motivational dispositions. Hence, in the third study learners’ motivational dispositions such as their goal orientations and their academic self- concept were investigated in relation to their expected support through learning analytics. Furthermore, potential support of learning analytics with focus on enhancing motivation was assigned to the three phases of self-regulated learning. Many learners have difficulties in self-regulating their learning especially in not particularly structured environments such as higher education or digital learning environments. Hence, instructional means such as prompts are considered to provide additional support. In the fourth study, using a quasi-experimental design, learners were confronted with prompts based on theory of self-regulated learning to investigate how they impact learners’ declarative and transfer knowledge and their digital learning behavior plus if trace data can inform learning achievement. As the collected data need to be interpreted based on theoretical foundation in the fourth paper of the thesis the aim is to synthesize theory on assessment and learning analytics. By integrating current theory on assessment, assessment design, feedback and learning analytics an integrative framework was developed. Learning analytics might offer additional guidance for increasingly heterogenous learners and support teachers to adjust their instruction to learners’ needs and reduce their workload. However, learning analytics are still at an initial level where this thesis adds additional empirical evidence and theoretical contribution to promote learning analytics further. But learning analytics face several limitations and further research especially using an experimental approach is needed which will be discussed further in the concluding section