Machine learning (ML) has become integral in educational decision-making through technologies such as learning analytics and educational data mining. However, the adoption of machine learning-driven tools without scrutiny risks perpetuating biases. Despite ongoing efforts to tackle fairness issues, their application to educational datasets remains limited. To address the mentioned gap in the literature, this research evaluates the effectiveness of four bias mitigation techniques in an educational dataset aiming at predicting students’ dropout rate. The overarching research question is: “How effective are the techniques of reweighting, resampling, and Reject Option-based Classification (ROC) pivoting in mitigating the predictive bias associated with high school dropout rates in the HSLS:09 dataset?" The effectiveness of these techniques was assessed based on performance metrics including false positive rate (FPR), accuracy, and F1 score. The study focused on the biological sex of students as the protected attribute. The reweighting technique was found to be ineffective, showing results identical to the baseline condition. Both uniform and preferential resampling techniques significantly reduced predictive bias, especially in the FPR metric but at the cost of reduced accuracy and F1 scores. The ROC pivot technique marginally reduced predictive bias while maintaining the original performance of the classifier, emerging as the optimal method for the HSLS:09 dataset. This research extends the understanding of bias mitigation in educational contexts, demonstrating practical applications of various techniques and providing insights for educators and policymakers. By focusing on an educational dataset, it contributes novel insights beyond the commonly studied datasets, highlighting the importance of context-specific approaches in bias mitigation.

Position effects occur when changes in item positions on a test impact the test outcomes (e.g., item parameters or test scores). Previous studies found that position effects may vary by the testing context and conditions and thus affect each test-taker differently. With the increasing adoption of digital assessments involving innovative item types that are focused on problem-solving skills, it is also essential to study position effects within this context. This study incorporates item-level scores and screen-level response time data from the Trends in International Mathematics and Science Study (TIMSS) 2019 problem-solving and inquiry tasks for Grade 4 students to examine position effects on students’ ability and test-taking speed. This dataset included responses from 27,682 students from 36 countries. A structural equation modeling approach was employed to model ability and test-taking speed within the same model. The results indicated a small but statistically significant booklet effect on students’ overall mathematics and science ability. The impact of block position changes within the booklets seemed to be greater than the impact of a reordering of subjects tested in the two sessions. The results also showed that when an item block was placed earlier in a test session, students spent more time on the items and performed better. The implications of these findings are discussed.

This chapter explores how AI-powered tools—including intelligent tutoring systems (ITSs), adaptive learning platforms (ALPs), and AI-driven chatbots—can personalize instruction and transform K–12 education. Drawing on recent research, theoretical frameworks, and practical case studies (e.g., ALEKS and i-Ready), the chapter examines how these technologies dynamically adapt to individual learners' needs, support instructional decision-making, and enhance student engagement and outcomes. It highlights the pedagogical architectures and functionalities of each tool type, assesses their benefits in promoting equity and data-driven instruction, and discusses key challenges related to integration, teacher preparedness, and algorithmic fairness. The chapter concludes with practical implications for educators and school systems, and offers future research directions focused on AI-enhanced personalization, generative AI integration, and teacher–AI collaboration.

Low-stakes assessments in K-12 education play a crucial role in monitoring student progress, yet their validity is often compromised by disengaged test-taking behaviors, such as rapid guessing and idling. This study introduces a novel application of the Nested Logit Model (NLM) to integrate test-taking engagement, as indicated by response times (RTs), with item responses to improve ability estimation. Using data from a low-stakes reading assessment in the United States (n = 27,556 students in grades 5 to 8), we compared six scoring approaches, including traditional dichotomous scoring, effort-moderated scoring, and nominal scoring that categorized responses based on RT-informed engagement. Our results demonstrated that nominal scoring approaches, particularly those distinguishing rapid guesses and idle responses, yielded superior model fit, increased measurement precision, and provided nuanced insights into examinee behaviors compared to dichotomous scoring methods. Latent class analysis further identified three distinct engagement profiles—effortful responders, rapid guessers, and idle responders—highlighting the need to address both rapid and idle behaviors in modeling. This study emphasizes the value of leveraging RT data to enhance the accuracy of low-stakes assessments while preserving response information. Findings also suggest the NLM framework as a practical and accessible tool for researchers and practitioners seeking to address disengaged behaviors and ensure the reliability of low-stakes assessments.