Academic Procrastination in Higher Education: A Review from Theory to Gamified Analytics

Academic procrastination is defined as the intentional and unnecessary delay of academic tasks despite knowing it will lead to negative consequences. It is a common issue among students and is consistently related to poorer academic performance, increased stress, and lower psychological well-being. Psychological and emotional factors such as fear of failure, perfectionism, preference to avoid mental fatigue, and low esteem, as well as behavioural and motivational factors such as lack of proper time management, poor study habits, lack of interest and task avoidance can contribute to procrastination [1]. External factors such as academic environments can further increase students’ vulnerability to procrastination by draining their coping resources and reducing their ability to regulate negative emotions [2].
Various approaches that have been introduced to address academic procrastination, such as time management techniques, self-regulated learning strategies, and behavioural interventions. However, the need for more individualized support is highlighted when different psychological and behavioural determinants of procrastination are targeted by intervention approaches. Further evaluation is required as well to determine the effectiveness of various intervention designs and its suitability for individuals with different needs [3], [4]. The study has linked intrinsic motivation and academic self-efficacy to academic procrastination [5].
Recent advances in machine learning-based predictive models can identify students at-risk by analysing educational data, thereby, offering potential opportunities for early intervention and personalized support [6]. Meanwhile, gamification elements such as rewards, challenges, and progress tracking may support task engagement. Nevertheless, as predictive approaches and gamification interventions are often studied separately, their integration into personalized support systems is limited.
Therefore, this review synthesizes theoretical and empirical research on academic procrastination in higher education by focusing on psychological, behavioural and contextual factors as well as mitigation strategies that are technology-driven. It identifies the limitations and opportunities as well to develop interventions that are more personalized and adaptive in higher education. Finally, this review proposes an integrated conceptual framework where theoretical risk factors, behavioural signals, predictive analytics, and personalized gamified interventions are linked to guide research in the future.

A narrative approach was adopted in this review to examine literature on academic procrastination in higher education, its contributing factors, and technology-driven mitigation strategies. Literature was identified across IEEE, Xplore, ScienceDirect, Scopus, SpringerLink, PubMed, and Google Scholar (2020 to 2026) by using the search terms “academic procrastination”, “higher education”, “university students”, “gamification”, and “machine learning”. The studies were considered relevant if they addressed academic procrastination, its cognitive, emotional or behavioural factors, mitigation approaches, gamification or predictive analytics in educational settings. Research unrelated to these topics was excluded. Of 84 records identified, 68 abstracts were screened, and 31 primary studies were retained for the narrative synthesis. The studies selected were then organized in themes so that the theoretical explanations, mitigation approaches, predictive models and research gaps can be compared.

From a cognitive perspective, academic procrastination is understood as the product of how students evaluate their capabilities and regulate their learning. Across the reviewed studies, academic self-efficacy emerges as a central mechanism. Huang, et al. [7] found that self-efficacy partially mediated the relationship between adaptive perfectionism and academic procrastination whereas resilience moderated this relationship among Chinese nursing students, while Jing, et al. [8] reported that academic self-efficacy partially mediated the relationship between negative perfectionism and academic procrastination among Chinese university students suggesting that the relationship may change depending on the type of perfectionism and the group of individuals involved. Extending this view to self-regulation, García-Ros, et al. [9] showed that self-regulated learning, including metacognitive strategies and self-efficacy, shapes procrastination, which in turn predicts lower achievement and well-being and higher stress. Taken together, these findings position self-efficacy as a key cognitive pathway, although reliance on cross-sectional designs and differing educational contexts limits causal and cross-cultural generalization.

From an emotional perspective, academic procrastination is associated with regulating negative emotions, including anxiety, depression and fear of failure. The procrastination-health model provides understanding of the relationship between procrastination and health-related outcomes, while recent evidence shows that procrastination is closely tied to higher levels of depression, anxiety, and stress [10]. Moreover, research has also identified impulsivity and depression-related difficulties, notably self-hatred and poor concentration are closely relevant to procrastination [11]. Komal, et al. [12] found that guilt mediates the relationship between academic procrastination, academic burnout and sleep quality, emphasizing how emotions can link procrastination to the well-being of a student. However, it remains unclear whether negative emotions are the cause of procrastination or the result of it. Future studies could clarify this relationship and design interventions to support both student emotions and time management.

From a behavioural perspective, academic procrastination is associated with habitual delay, poor time management and ineffective study practices. Kurniawan [13] identified several factors linked with academic procrastination such as poor time management, fatigue, and trouble understanding assignments, which suggests that by delaying the completion of tasks, it may reflect more than laziness. Additionally, research on college students highlighted the relation between academic procrastination and difficulty in self-regulation and bad habits, emphasizing the link among procrastination, daily behaviours of the students, and well-being [14]. Similarly, Fentaw, et al. [15] identified procrastination is associated with factors such as poor scheduling and inefficient use of time, including time spent on social media, leisure, and social events. Together, these findings indicate that daily routines and competing demands can reinforce procrastination although varying study groups and self-reports limits the certainty. These behavioural patterns are important as they provide visible signals to track student engagement and inform timely, personalized support.

Existing efforts to reduce academic procrastination fall into three broad approaches: traditional classroom interventions, gamification, and predictive analytics. Each targets a different aspect of the problem, and Table 1 summarizes their key mechanisms, strengths, and limitations.
Traditional approaches focus on helping students regulate their cognition and emotion. Drawing on a self-control model, Miyake and Kane [16] proposed a holistic classroom framework that views procrastination as a failure of goal management, in which students prioritize short-term mood repair over long-term academic success. The framework combines strategies that address task-related negative emotions and planning difficulties with student reflection and community building. Although it is low cost and designed to fit into everyday classroom activities, it depends on instructors to select and apply suitable techniques, which limits personalization and long-term engagement.
Gamification takes a more engagement-driven route. A systematic review of 24 empirical studies [17] identified points, badges, leaderboards, challenges, and progress tracking as elements used to address procrastination. However, outcomes depend on system design, user characteristics, and implementation context, so it remains unclear whether increased engagement translates into sustained reductions in procrastination.
In contrast, predictive analytics aims at early detection rather than direct intervention. Sackey et al. [18] combined autoencoders and logistic regression to detect procrastination patterns from clickstream data, demographics, grades, and admission records. While such models can effectively identify at-risk students, prediction alone does not resolve procrastination, and the findings need to be validated across different student groups and linked to meaningful interventions.
Taken together, these approaches are complementary but incomplete: traditional methods are scalable yet instructor-dependent, gamification may boost engagement without proven lasting effect, and predictive models identify who is at risk without addressing what to do about it.

Gamification implements game-design features into non-gaming environments that helps boost student motivation and engagement via tools like rewards, real-time feedback, and progress tracking [19]. Platforms like ClassDojo and Khan Academy use virtual characters, badges, and progress tracking to improve student involvement and track their performance while ProScore and Forest are designed to support focused work and manage procrastination.
ProScore combines both Pomodoro techniques and gamification features including point system, leaderboards, and tracking of productive session, to encourage productive work and reduce procrastination among the employees [20]. Similarly, a study involving surveys and interviews with five active users of the Forest app examined how rewards, punishment, customization, and collaboration affect students’ focus and their ability to manage impulsive procrastination [21]. Although both platforms illustrate how gamification can be utilized to support task engagement, they both use different mechanisms, where ProScore highlights structured work sessions and productivity feedback whereas Forest explores features that are motivational and social to support focused behaviour.
However, one main limitation across these studies is the evidence. Findings from small sample in the Forest study should not be generalized as conclusive evidence of long-term behavioural change. Further investigation is needed to identify the extent of gamified interventions maintain their effects beyond initial use. These limitations highlight the need for personalized interventions and evaluations of behavioural outcomes over the longer-term.

Machine learning (ML) is a subgroup of artificial intelligence (AI) that enables computer systems to learn patterns from data and make predictions or decisions without needing explicit programming [22]. In educational settings, these techniques have been incorporated to identify behavioural patterns of the students, classify their learning profiles, and predict academic outcomes [23-26]. While these studies demonstrate the potential of machine learning to understand student behaviour, models that examine procrastination-related indicators are required specifically to predict academic procrastination.

The use of learning management system (LMS) activity data have been explored in studies to identify the tendencies of procrastinating. Öztaş and Akçapınar [27] analysed 14 weeks of online learning behaviours among 51 university students and reported that Logistic Regression achieved 90% accuracy using 10-fold cross validation, where early-week engagement features were the most significant predictors. Similarly, Sartana, et al. [28] predicted procrastination behaviour and academic performance using Moodle log data of 377 students and reported that Random Forest performed the best with an R2 of 0.812, while revealing a strong negative correlation of 0.7777 between delayed tasks and academic outcomes. Other studies have examined ensemble models and larger virtual learning, showing promising results when using Gradient Boosting, Logistic Regression, and artificial neural networks [29, 30].
While behavioural data helps identify procrastination risks early, model performance must be interpreted within the unique context of each study’s dataset, sample size, and validation method. Direct comparisons can be difficult when using varied behavioural markers and modelling techniques. Predictive insights must be linked to practical and effective support strategies as only flagging students who are at-risk will not solve the problem.

While using machine learning for education purposes, ethical considerations must be addressed especially regarding data privacy, algorithmic bias, and model transparency [31]. Unrepresentative datasets can result in inaccurate risk identification and uneven support whereas tracking continuous LMS interactions requires strict data governance, informed consent, and anonymisation protocols. Crucially, predictive models should serve solely as formative decision-support systems for educators.

Three key gaps from the reviewed literature. First, findings from small sample sizes or datasets from specific platforms, or single institution can limit the applicability across different educational contexts [27, 28, 30]. Future studies should therefore examine larger and more diverse student populations to improve generalisability of findings. Second, many studies rely heavily on digital footprint such as LMS activity logs without sufficiently documenting emotion regulation, mental health, and motivation [28, 29]. This may oversimplify procrastination as a behavioural issue instead of complex interaction of cognitive, emotional, and contextual influences, highlighting the need to incorporate these factors into predictive models while maintaining interpretability and protecting student privacy. Third is the lack of integration between predictive analytics and intervention mechanisms. Although machine learning models may identify at-risk students, they are often not connected to actionable interventions. Concurrently, the gamification strategies are not aware of the real-time behavioural data or the predictive insights even though they are effective in enhancing motivation and engagement. This gap highlights the need to develop a framework that unifies gamification, machine learning and predictive modelling so that procrastination risks can be detected and personalized gamified interventions will be triggered.

Figure 1 below presents an integrated conceptual framework by linking the four components including theoretical risk factors, behavioural signals, predictive models, and personalized gamified interventions. This framework does require empirical evaluation to determine whether the student outcomes can be improved with this integration.

This review summarizes both theoretical and practical perspectives for academic procrastination by highlighting its complex, multifaceted nature shaped by cognitive, emotional and behavioural factors. Although traditional psychological approaches can provide foundational support, they often lack scalability and personalization. Emerging approaches like gamification together with machine learning and predictive analytics have shown potential to reduce academic procrastination. Gamification elements can boost the motivation of the students whereas machine learning and predictive modelling enables early identification of at-risk students. A powerful framework for adaptive student support can be provided by combining theoretical risk factors, behavioural tracking, and predictive analytics with personalized gamification. Further research is required to test its effectiveness in the real-world, ensure the model operates across student groups with diverse backgrounds, and address critical concerns around privacy and long-term behavioural change.