The first theme from the literature reviewed addresses the use of learning analytics to facilitate student success prediction, timely intervention, and ultimately improve students' learning outcomes. Students are the key stakeholder in this theme.

Firstly, many studies in the literature reviewed have focused on student success and performance prediction. Dhankhar and Solanki (2020) provided a bird's eye view of learning analytics usage in Higher Education worldwide by reviewing leading publications and research evidence. The learning analytics implementation in different countries and regions is summarized in this research, such as the United States, United Kingdom, Austria, and Europe. It identified one of the main functions of learning analytics, which is discovering hidden patterns in the educational process, assessing student learning, and making predictions, which in turn provide a better understanding of teaching, learning, and interpretation of student data. Furthermore, in another systemic review conducted by Iffenthaler and Yau (2020), learning analytics can identify potential students at risk of dropping out by analyzing students' online behavior (login frequency, online forum posts, and assignment submission) data combined with their grades.

Furthermore, students' demographics (academic self-concept, academic history, and financial condition) can also predict student success. Unlike predictions based on students' demographic information, Foster and Siddle (2020) examined non-engagement student data (non-attendance or non-submission of coursework) by comparing disadvantaged groups and their control group. The non-engagement student data can also indicate at-risk students. They proved that non-engagement alerts are more efficient at identifying at-risk students than demographic data. It provided a more neutral framework to allocate resources to students who need support without stigmatizing students' backgrounds. Their findings demonstrate the potential of using automatic non-engagement alerts to address the attainment disparity from an equity, diversity, and inclusion lens. Learning analytics could predict students' performance equitably so students will get immediate remediation from the instructors.

Secondly, some research has shed light on the Instructors' Intervention. Liu et al. (2017) analyzed data from a large-scale online course and identified different learning behaviors and performance patterns. They found significant differences in the learning behaviors of students who earned high grades versus those who earned low grades. Based on these findings, the authors suggested several design principles for adaptive learning systems, such as providing personalized feedback and resources to students based on their learning behaviors, using data to predict which students may be at risk of falling behind, providing personalized interventions to help them catch up, and designing course content and activities to promote more active and engaged learning behaviors. Similarly, several intervention practices to support study success are summarized in this review by Ifenthaler and Yau (2020), such as visual signals and dashboard features for instructors, peer interactions, adaptive learning materials, prior knowledge building, reduction of test anxiety, and student-teacher perceptions. The review emphasized that personalized learning paths and interactions can be effective intervention strategies to mediate the students who are at risk. However, Sonderlund et al. (2019) conducted a literature review of 11 studies on the efficacy of learning analytics interventions in Higher Education. They summarized that while there are plenty of studies on predicting student performance and retention, the research on the effectiveness of learning analytics intervention is limited because of lacking synthesis of effective practices from different regions and conditions. This review advocated for more research to investigate the total value of learning analytics intervention. Thus, the above research examined how instructors' intervention related to learning analytics supports student success.

Lastly, Several studies in the literature reviewed investigated how Learning Analytics can develop students on different learning outcomes, such as cognitive gains, self-regulation skills, and student agency. Viberg et al. (2018) reviewed 252 studies investigating the effects of using learning analytics dashboards. Viberg et al. summarized that cognitive gains refer to intellectual abilities such as self-reflecting, metacognition, and analyzing skills. Sonnenberg and Bannert. (2015) conducted a study in a Higher Education context, specifically in a programming course. The authors found that using dashboards was positively associated with cognitive gains among the students. Students who used the dashboards improved their understanding of the course material and ability to apply the learned concepts to practical programming tasks. Overall, the study suggested that using learning analytics dashboards can effectively enhance students' cognitive gains in Higher Education settings.

In Viberg et al.'s review, studies examined that learning analytics dashboards can improve students' self-regulated learning, presentation, and problem-solving skills in Higher Education settings. Similarly, Jääskelä et al. (2021) defined student agency means "a student's experience of access to/having (and using of) personal, relational (i.e., interactional), and context-specific participatory resources to engage in intentional and meaningful action and learning." The same authors presented a student agency learning analytic service architecture that can visualize the student agency profiles. It also summarized the definition and framework of agency in Higher Education. Furthermore, it proved that learning analytics could present student agency profiles to optimize learning and teaching through personalized academic advising, self-reflection, and self-regulation.

Jivet et al. (2020) also researched the relationship between learning analytics dashboard design and students' self-regulated skills. The authors conducted a literature review and a quantitative study of 176 students to explore the impact of learning analytics dashboards on self-regulated learning. They identified several studies that suggested a positive relationship between learning analytics dashboards and developing self-regulated learning skills. In addition, Jivet et al. provided some recommendations for designing the learning analytics dashboard to provide immediate feedback to students on their learning performance and help students reflect on their learning behaviors and strategies, then build and refine the learner behavior. Lastly, Niet et al. (2016) conducted a study of 364 undergraduate students with a mixed method to explore the impact of an academic performance module (APM) on students' academic performance, attitudes, and behaviors. The APM is a learning analytics tool designed to provide students with personalized feedback on their academic performance and promote self-regulated learning. The authors found that the APM positively impacted students' attitudes toward learning, such as their motivation and confidence. Many studies in the literature reviewed shed light on how learning analytics can impact learners' cognitive gains, self-regulation, and affection.

This theme focuses on how learning analytics can help instructors with teaching skills and practices. The teaching aspect of learning analytics has gained popularity in Higher Education. Viberg et al. (2018) conducted a review and found that 62% of research focuses on improving learning support and teaching, which suggests we should consider how to transfer research into practice. There is more evidence supporting institutions and teachers rather than students themselves. Numerous studies in the literature reviewed investigated the following areas: teacher feedback and intervention, online teaching practices, and pedagogical knowledge.

As addressed in the student success section, teachers can use learning analytics to identify students' problems and provide personalized support. Several studies in the literature reviewed examined teachers' feedback and intervention via learning analytics. Viberg et al. (2018) conducted a systematic review of learning analytics for teacher intervention. They identified 17 studies that met the inclusion criteria, and the results indicated that interventions that provide personalized feedback and early warning systems improved student learning outcomes. The authors also found that the design and implementation of the interventions and the level of engagement and support from the instructors were critical factors in their effectiveness. While Viberg et al. specifically focused on using learning analytics in interventions, Sonderlund et al. (2019) provided a broader perspective on the effectiveness of educational interventions in Higher Education. The authors conducted a meta-analysis of 67 randomized controlled trials that evaluate the effectiveness of educational interventions in Higher Education. The results indicated that active learning interventions (e.g., flipped classroom, problem-based learning) moderately positively affected student learning outcomes. The authors also found that interventions that provided immediate feedback and targeted study skills improved student learning outcomes.

To better use learning analytics for teaching excellence, it is vital to understand the definitions of excellent teaching competencies and practices in the online learning environment. Oliva-Córdova et al. (2021) built a literature review of 50 studies on learning analytics and teaching skills. It summarized the benefits of the learning analytics application for teaching competencies and teaching practices. Oliva-Córdova et al. addressed that the learning and application of technologies should be an integral part of teachers’ continuing education throughout their careers. The literature reviewed suggested that the learning analytics tool can help instructors improve their teaching practices in several areas, such as learning design, learning management, and pedagogical mediation. The table (See Figure 1.5 below) summarizes the benefits of using learning analytics in teaching practices.

Figure 1.5: Benefits of using Learning Analytics in teaching practice. (Oliva-Córdova et al., 2021)

In Oliva-Córdova’s review of the benefits of using learning analytics in teaching practice, Learning Management System (LMS) is a significant component. Some examples are making decisions related to redesigning learning experiences, identifying effective teaching methods, and measuring what is happening in the LMS. Using learning analytics requires faculty to gain digital skills, computer literacy, and new pedagogies to use these digital tools for teaching.

The following three studies in the literature reviewed focused on the relationship between learning analytics and Teaching Practices. Cope and Kalantzis (2016) argued that the LMS could embed formative assessments and present structured data to show evidence of learning. This evidence will alert faculty for instructional recalibration. The authors argue that learning analytics can provide teachers with real-time data on their students’ learning progress, enabling them to make informed decisions about adjusting their teaching to meet their students’ needs. Similarly, Dhankhar and Solanki (2020) discussed that a ‘Loop’ project in some Australian universities targeted the practical problem. They explored how to support effective teaching better online and understand the needs and perceptions of teachers in Higher Education to ensure that learning analytics can be genuinely helpful in teaching and learning practice. In addition, Patil and Gupta (2019) focused more on using learning analytics to support collaborative learning. The authors argue that learning analytics can provide teachers and students with real-time data on their collaborative learning activities, enabling them to make informed decisions about adjusting their learning strategies to support collaboration better. The authors addressed the importance of using learning analytics to support collaborative learning practices prioritizing active participation, equal contribution, and feedback.

A few studies in the literature reviewed focused on the relationship between learning analytics and pedagogical knowledge. Bronnimann et al. (2018) conducted a case study and suggested that academic staff can articulate pedagogical questions around learning analytics based on the six stages of Scholarship of Teaching and Learning. The six stages provide a structure and process to explore the relationship between learning analytics and instruction. They found that learning analytics can enhance pedagogical knowledge by providing insights into student learning and engagement, facilitating formative assessment, and informing teaching practices. Likewise, Viberg et al. (2018) explored teachers' role in using learning analytics. They argued that teachers must be involved in designing and implementing learning analytics systems to ensure that they align with their pedagogical goals and practices. They found that teachers were more likely to adopt learning analytics if they perceived it valuable and relevant to their teaching practices. They suggest that developing learning analytics systems should be a collaborative process between teachers and learning analytics designers to ensure that the systems meet both needs. Lastly, Cerro Martínez et al. (2020) agreed that using learning analytics could support pedagogical knowledge development by providing teachers with real-time data about their student's learning progress. The authors conducted a case study in a university setting. They found that teachers who used learning analytics could better understand their students' learning needs and tailor their teaching approaches accordingly. They suggested that learning analytics can contribute to creating a data-driven culture in education, where teachers use data to make informed decisions about their teaching practices. Overall, all three studies in the literature reviewed suggested that learning analytics has the potential to enhance pedagogical knowledge by providing teachers with real-time data about their student's learning progress. These studies addressed that teachers must be involved in designing and implementing learning analytics systems to ensure that they align with their pedagogical goals and practices.

This section discusses on how learning analytics can help designers and instructors continuously improve learning design for online courses and programs. Rienties et al. (2015) defined learning design's objective as "establish the objectives and pedagogical plans which can be evaluated against the outcomes captured through learning analytics" (p. 315). They emphasized that learning design usually influences learners' engagement in the online learning environment. Viberg et al.'s review echoed that learning design is a popular research area in learning analytics. However, Rienties et al. disputed that Learning design, often ignored by the learning analytics knowledge community, is a process in which educators "make informed design decisions with pedagogical focus and communicate these to colleagues and students" (p. 315). They also valued that learning analytics can evaluate the objectives and teaching plans established by learning design. This theme centers on three major topics under learning design: enhance curriculum alignment, guide curriculum redesign, and promote learner engagement.

Firstly, in the literature reviewed examined how learning analytics can support curriculum alignment in online learning. Liu et al. (2017) organized a study using learning analytics to improve adaptive learning design. In their study, Liu et al. analyzed data from 215 students who completed an online adaptive learning course. They used data mining techniques to identify patterns in the data and then used these patterns to develop recommendations for improving the design of adaptive learning experiences. This study explored the potential of data to inform the design of practical adaptive learning experiences. These findings align with the broader literature on learning design, emphasizing the importance of alignment with learning outcomes, collaboration and interaction, and feedback provision. Likewise, Rienties et al. (2015) agreed the Liu et al.'s finding about curriculum alignment. Using a cluster and correlational study, compared 87 learning modules, Rienties et al. found that academics did not consider blueprints when designing them. They also discovered that it is difficult to interpret the learning analytics data without associating the learning design activities with the Learning Management System usage.

Secondly, in Oliva-Córdova et al. (2021)'s systematic review of 50 studies, learning analytics applications are related to curriculum review. Learning analytics can provide invaluable insight into how students react to different learning designs. The data-driven approach can also identify what has worked and has not worked in course design. Here are some examples of how learning analytics can support course design:

• Identify which learning activities are practical for students.
• Improve competencies and the mapping of the curriculum.
• Make decisions related to redesigning learning experiences.

However, Jayashanka et al. (2019) emphasized that creating synergy between learning analytics and learning design is essential to improve learners' performance, engagement, and satisfaction. Jayashanka et al. proposed a framework and a tool to create a dashboard for faculty and students to enhance blended learning in Higher Education. The tool will provide faculty information to allow course revision while teaching the course. The authors found that this tool improved students' engagement and learning outcomes, examining the importance of incorporating learning analytics into effective learning design. Moreover, Nguyen et al. (2018) discovered a mismatch between the faculty's estimated learning time and students' actual learning time. They conducted a study that analyzed the timing of students' engagement with online course content using learning analytics and found that students who engaged earlier and more frequently tended to perform better academically. This research allows faculty to take action on adjusting the course materials. Nguyen's study also emphasized the pedagogical context so faculty can interpret learning analytics into actionable insights for continuous improvement in course design.

Lastly, Rienties et al. (2015) researched the relationship between learning design and learner engagement and performance in the Learning Management System. The author conducted a study that involved 10,975 undergraduate students across 36 modules at a UK university. One of the findings is that learning design activities impact students' engagement in the online learning environment. Another finding is that the learning design activities can influence learner performance, motivation, and engagement. Similarly, Liu et al. (2017) emphasized students' online engagement in the adaptive learning system by studying 128 first year pharmacy students. Their finding demonstrated that cognitive ability and affective characteristics (e.g., motivation, mastery goal orientation) play a role in students' engagement and performance. The finding prompted the agenda to develop learner profiling of different characteristics throughout time to generate more personalized learning content for students. Klein et al. (2019) agreed that learner experience and engagement are associated with the quality of the learning design. The study used a mixed-methods approach, including surveys and interviews, to collect data from 147 participants from 14 Higher Education institutions in the United States. The participants included faculty, administrators, and technical support staff. Learning analytics could enhance curriculum alignment, course redesign, and learner engagement.

Using learning analytics for institutional goals has been widely studied in the review. This theme centers on how administrators can implement Learning analytics for staff performance, workload, program review, institutional effectiveness, and institutional adoption of learning analytics.

Firstly, a few researchers investigated learning analytics to measure the performance of instructors, support staff, and student services employees for institutional-level goal alignment. Jonathan et al. (2018) presented a theoretical model consisting of two stages to measure key performance indicators of staff. However, no researchers have yet to prove the model. This study demonstrated learning analytics' potential in institutional-level goal alignment and performance measurement besides predicting students' performance and intervention. However, little research has been done on this staff performance measurement, Dhankhar and Solanki (2020) conducted a case study that Strayer University utilized learning analytics to increase faculty engagement and improve their behavioral mindset. They concluded that learning analytics is not only a research issue but also an organizational issue. Thus, they suggested that institutions address faculty engagement issues according to institutional and cultural context. Likewise, Vanessa Niet et al. (2016) echoed that assigning appropriate workloads to faculty and staff has challenged administrators in Higher Education institutions. To meet the administrators' needs, Vanessa Niet's team proposed a Decision-Making Support System Model and developed a software called UDLearn to support administrators' decision-making process. This software provides a tool for assigning new supervisors or examiners to graduation projects, considering the number of documents that each teacher workload. It also helps with the business logic report generation, and the scalability of the learning analytics application is feasible after a pilot test carried out in a public university in Latin America.

Secondly, like how learning analytics can be utilized to inform course review and continuous improvement explored in the earlier section, it can also be used to guide program review and evaluation. Bronnimann et al. (2018) conducted a case study by an Australian university to use learning analytics to inform program review. It used the Scholarship of Teaching and Learning approach and discovered pulling data from Learning Management System and Student Information System to conduct program review and program redesign. It is also helpful to provide ongoing student support in a new program. Thus, they recommended that administrators include institutional-level data about the student experience in the program review process, generating new insight. Ultimately this new usage of learning analytics leads to data-driven continuous improvement for better student retention and success, which coincided with a section on student success.

Similarly, a research project was outlined by Jayashanka et al. (2019) on using learning analytics for program research and development. The authors also proposed a framework called Intelligent Interactive Visualizer, which can provide real-time feedback on students' progress and examine program areas that require improvement. Further, Horn et al. (2019) examined the effectiveness of community colleges via learning analytics. After examining the effectiveness scores of over 800 community colleges in the United States, they found that the effectiveness measure can indicate whether closer scrutiny of institutional conditions is needed. The result also supported to use of the effectiveness scores as a holistic indicator for community colleges.

The last common from the literature reviewed is the adoption process of learning analytics at an institutional level. Several studies in the literature reviewed focused on inventing models and frameworks for successful implementation. Arnold et al. (2014) proposed an instrument survey called Learning Analytics Readiness Instrument (LARI) to measure the readiness of institutions to embark on the implementation of learning analytics. This instrument centered on five primary factors: ability, culture and process, data, governance and infrastructure, and overall readiness perception. The utility of this tool can identify deficiencies in the institution and provide actions to remediate these areas for a higher chance of successful implementation. Following the invention of the LARI survey, Oster et al. (2016) presented an analysis of the LARI survey pilot administration from a larger sample. Their finding enhances the process of learning analytics implementation, and it is critical to consider institutional and role characteristics across different factors before initiating the action plan. Aiming to implement learning analytics in Higher Education, Tsai et al. (2018) addressed a framework that can generate strategies and policies around learning analytics. This study examined three case studies using the SHEILA (Supporting Higher Education to Integrate Learning Analytics) framework. It demonstrates the potential to utilize this framework to support institutions' strategic planning and policy-making process. Further, it provides three areas, including actions, challenges, and policies, that should be considered in the systematic adoption of learning analytics. Thus, the learning analytics adoption frameworks have been a critical research emphasis under the institutional goals section.

A few studies in the literature reviewed examined a few challenges of implementing learning analytics internationally. Tsai et al. (2020) pointed out that the imbalance of stakeholder engagement and resistance to the change is considered the most common barrier to learning analytics implications in European Higher Education. In the United States, institutions share a similar challenge as Tsai's team. A case study by Klein et al. (2019) has discovered a similar barrier of untrusting technological infrastructure. One faculty responded at a focus group, "So, you have an institutional learning analytics tool, a degree progress tool, a student information management tool, and then your e-mail and whatever. So, you have to be relatively savvy with all those programs to use them and know which one provides you with which information. Yeah, you have got to have a lot of stuff" (Klein, 2019, p. 614). Faculty and advisors feeling frustrated, would not use learning analytics, leading to no buy-in, uneven use, and alternative practices. Klein's team also identified some barriers regarding learning analytics adoption from the lens of faculty and advisors. They conducted a focus group and found that the first barrier is a lack of trust in the technology infrastructure from faculty and advisors. The second barrier is the cumbersome infrastructure, making the tools difficult to use. The third barrier is lacking accurate and timely data and lacking visualized dashboard. The last barrier is the conflict between actionable data and teaching pedagogy.

Several studies in the literature reviewed focused on the barriers encountered by instructional designers in learning analytics implementation. For example, Muljana and Luo (2021) outlined that Instructional Designers serve as support staff to provide helpful information generated from learning analytics tools to facilitate the course design process in most Higher Education institutions. The study used semi-structured interviews with 14 instructional designers from various Higher Education institutions in the United States. The findings demonstrated that several factors influence instructional designers' adoption of learning analytics integration in course design practice: individual differences (prior exposure and pre-perception, pedagogical belief), systematic characteristics, social influences, and facilitating conditions. To clear these barriers, the same authors suggested some practical implications, for example, conducting a needs analysis before jumping into the learning analytics adoption plan would be wise. The authors recommended seeking synergy among different stakeholders on campus to ease the integration and implementation process. To summarize, disengagement of stakeholders brought barriers to applying learning analytics in Higher Education institutions.

The literature reviewed revealed that learning analytics can only function effectively when this tool is based on solid educational foundations and theories. Cerro Martínez et al. (2020) found evidence of this challenge to optimize the pedagogical opportunities of online learning. However, they posit that learning analytics can support instructors in finding new ways to develop pedagogical dynamics. Tsai et al. (2020) indicated a lack of alignment between pedagogical theories to meet the needs of students and faculty in European Higher Education. Identifying associated learning science theories is essential, especially when collecting suitable data sets and choosing indicators of learning progression.

Similarly, Nguyen et al. (2018) emphasized that pedagogical context must be provided to users to interpret and translate the learning analytics into actional items. The authors noted that the timing and format of students' engagement can significantly impact their learning outcomes. Designing learning activities and assessments that align with the intended learning outcomes is essential. Finally, Nguyen's team emphasized the need for a pedagogical approach that integrates learning analytics with learning design to ensure that the data generated by learning analytics is meaningful and actionable.

The integration of learning analytics relies on strategic investment in human, financial, and technical resources (Tsai et al., 2020). Multiple studies in the literature reviewed demonstrated a lack of flexible components to support varying user needs when applying learning analytics. Oliva-Córdova et al. (2021) suggest that institutions must provide ongoing faculty training. The life-long professional development opportunities will be the central component to addressing these challenges, such as resistance to change, lack of knowledge, and ethical concerns. Besides the demand for human resources, Klein et al. (2019) indicated that technical support is needed due to a lack of integrated, accurate, and timely information and a helpful visualization dashboard. The authors conducted a qualitative study to identify the need for user-friendly interfaces that are easy to navigate and use. Users reported that complex and difficult-to-use interfaces were barriers to adoption, which hindered their ability to use the tools and interpret the data generated effectively. Al-Tameemi et al. (2020) agreed that learning analytics could be complicated. The authors systematically reviewed the literature to identify challenges associated with predictive learning analytics. For example, the five-step process includes capturing, reporting, predicting, acting, and refining. The authors found that the accuracy and effectiveness of predictive models heavily depended on the quality and availability of data. Poor data quality, incomplete data, and data silos were identified as barriers to successfully implementing predictive learning analytics. The research cited above echoed Tsai's summary of the tension between institutions' need to innovate and their existing capacity to tackle priorities.

Lastly, Tsai et al. (2020) stressed that, in European Higher Education, data collection for learning analytics had raised concerns about intruding on the learner's privacy and various ethical implications. In the United States, Cope and Kalantzis (2016) stated some similar challenges around data privacy. The first concern is how big data may negatively impact students and faculty by predetermining their academic performance and professional destinies. The second concern is the privacy protection for users when using extensive data methods. The last concern is how we can prevent significant data from becoming 'Big Brother' and recruit users as co-collectors and co-researchers so they are at the center of the data-driven decision process.

Besides data privacy, there is also a concern about increased bias against students when using predictive data. For example, Klein et al. (2019) discussed that learning analytics data, including prior academic experiences, could engender bias. Thus, the existence of learning analytics data, whether used to predict, surveil, or inform, needed a clear purpose for its use. Apart from the bias issue, learning analytics can also bring surveillance issues to faculty and students. Though faculty may provide quick responses to students, students may feel their every action in the learning management system is monitored, which will cause discomfort on the student end. Though the ethical and privacy issues in learning analytics concern students and instructors, Viberg et al. (2018) conducted a literature review of the themes in learning analytics research and found that only 18% of the studies indicated ethics and privacy issues. Their result also indicates that only a few studies approach this issue systematically among some empirical research. Examples of studies focus on institutional ethical considerations, moral and policy issues, and students' vulnerabilities and autonomy.