Publications
This study investigates secondary education students' self-regulated learning (SRL) processes with digital trace data, particularly whether SRL processes found in secondary education are comparable to those observed in higher education. We therefore adapted a digital learning environment and rule-based AI algorithm originally designed to measure SRL in higher education and collected multi-trace data from 13-year-old students ( N = 179) across three European countries during an essay-writing task. Hidden Markov modeling was employed to capture latent SRL processes. Four latent SRL processes emerged: orientation, first-reading, writing, and re-reading combined with monitoring. By clustering sequences of these latent SRL processes, we identified four sequential patterns of SRL processes at the task level: writing with metacognitive monitoring, writing intensively, reading first, writing next, and reading and writing simultaneously. Our findings highlight how AI and multi-trace data can be used to measure SRL during learning, providing a basis for enhancing personalized support. Self-regulated learning (SRL) is vital in the digital world. In this study, we investigated secondary education students' SRL processes with digital trace data. We also demonstrated that the instrumentation of a digital learning environment and rule-based AI algorithm originally designed to measure SRL processes in higher education can be leveraged to measure SRL processes in secondary education students. The real-time measurement of covert SRL processes is important as this information can 1) raise students' awareness on their learning, 2) help teachers to support students' learning, and 3) form the basis for providing personalized support for these processes with the help of AI-enhanced learning technologies. • Collected multi-trace data from 13-year-old students during an essay-writing task. • Found four SRL processes: orientation, reading, writing, re-reading with monitoring. • Identified four sequential patterns of SRL processes at the task level. • Students' task knowledge was associated with SRL at the task level.
ABSTRACT Elementary science education, particularly in the 4th and 5th grades, is essential for setting the foundation for lifelong science learning, fostering critical thinking, and preparing students for success in science, technology, engineering, and mathematics (STEM) fields. This stage is especially critical for students with disabilities, as achievement gaps between them and their peers emerge during elementary school. Despite this importance, little is known about how science is taught in elementary classrooms during these critical years, particularly for students with disabilities. To address this gap, we surveyed teachers from a nationally representative sample of U.S. schools to examine elementary science education, including instructional practices, allocation of time, and the inclusion and support of students with disabilities. Our findings reveal that limited instructional time is allocated to science, with significant variability across classrooms. The amount of time dedicated to science instruction was significantly influenced by external factors, such as whether science was a tested subject. Students with disabilities often face additional barriers, including being pulled out of science instruction for special education services, resulting in missed opportunities to engage in science. These findings highlight the need to address opportunity gaps in science instruction to ensure all students have meaningful access to quality science education.
Digital traces have been used to measure self-regulated learning (SRL), yet the validity of inferences made about these traces has often been questioned. Recently, researchers have used multiple channels of data — including digital traces, verbalizations, and self-reports — to validate inferences about individual SRL events. Research on the validation of inferences about sequences of multiple SRL events remains limited; however, investigating these sequences has the potential to refine SRL theories. To study the validation of sequences of SRL events, we collected multimodal data from 49 undergraduates completing a math task in a lab setting. Participants were asked to think aloud while interacting with different digital platforms. Then, we used sequence pattern mining to identify the digital events most predictive of post-test scores. Next, we used student verbalizations during the learning process to validate the inferences about what those predictive sequences reflected. Sequences representing learner conscientiousness predicted better performance; sequences that included pausing and rewinding videos predicted poorer performance. Some learner verbalizations co-occurred with digital events and consistently aligned with SRL processes, providing validity evidence for SRL sequences. Heterogeneity in verbal-to-digital trace alignment emerged and will require methodological advances to validate the sequences specific to individuals and task conditions.
Inequalities by disability status in college outcomes contribute to inequalities throughout adulthood in multiple domains. As the share of students with learning disabilities and/or ADHD increases on college campuses, research is needed on how best to support these undergraduates, both through formal and informal means. We integrate the social and medical models of disability with data on nearly 17,000 young adults in the US, who typically finished high school in 2013, and find that formal-disability-programming relates negatively to college enrollment but has no apparent bearing on college persistence. Among youth with similar sociodemographic backgrounds and comparable end of high school achievement levels, youth with a learning disability or ADHD who participate in special education during high school are less likely to enroll in college than youth with a learning disability or ADHD who do not. Multiple informal-disability-supports retain a significant relationship with college enrollment and persistence, even after accounting for important potential confounders. These supports may improve college outcomes because they demonstrate the relevance of education for careers (e.g., internships), build dominant social and cultural capital (e.g., high school counselors, career services), build human capital (e.g., college exam prep, academic support for college courses), and increase accessibility (e.g., online programs).
Many students in the United States experience educational disruptions when changing schools for reasons other than grade promotion. These nonstructural moves can detrimentally affect students' mathematics learning. While gender significantly shapes individual learning experiences, how student mobility plays out in mathematics learning based on gender has remained largely unexplored. Analysing six California school districts' panel data using school-fixed effect models, the findings in this study showed that the negative association of nonstructural mobility with mathematics performance is more detrimental for males than females and that self-management competency levels fully account for the differential effect. The findings suggest gender should be considered in understanding students' mobility experiences and underscore the need for incorporating self-management into educational interventions to mitigate the negative effects of nonstructural school transitions on mathematics performance.
Peer critique is essential for collaborative learning of science but presents challenges for students.Drawing on distributed and synergistic scaffolding, we designed multiple supports within a CSCL modeling environment, including integrated evidence and peer-feedback mechanisms, to help 6 th -grade students develop peer critique practices.Through analysis of students' comments and engagement patterns during critique activities, we found that students demonstrated progressive uptake of peer critique practices that the scaffolds were designed to promote, evident in more balanced, specific critiques over time.These findings inform the design of distributed scaffolds for supporting peer critique in science classrooms.
The importance of early numerical and executive function (EF) skills is well-established, with each skill set positively and specifically predicting later mathematics achievement, income, postsecondary education, and more. Less is known, however, about the relations between EF and numerical skills. Therefore, we examined the concurrent and predictive relations between EF and numerical skills in preschoolers to third graders (N = 205; 4.67-8.75 years of age; 43.9% female; 51.2% White non-Hispanic, 18% multiracial, 6.3% Hispanic, 12.2% Black, 2% American Indian/Alaska Native, 4.9% Asian, 1% not otherwise listed). We found positive concurrent relations between EF and all six numerical skills examined: nonsymbolic magnitude comparison, verbal counting, numerical literacy, count on, non-rote counting, and numerical problem solving. There were unidirectional predictive relations between EF and four of the six numerical skills after controlling for covariates and prior performance on the skill of interest. Bidirectional relations were found only for EF and nonsymbolic magnitude comparison. We also found that the concurrent relation between EF and count on was higher for children with typical versus persistently low mathematics achievement. All other concurrent and predictive relations were similar for children with typical and persistently low mathematics achievement. Overall, these findings show that the relations between EF and numerical skills are both pervasive and nuanced, such that they vary by timing of assessments (i.e., concurrent or predictive) and numerical skill. These results can inform future theoretical models on the role of EF in numerical development , have practical implications for designing interventions targeting these skill sets in early childhood. (c) 2024 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training , similar technologies.
We develop design and analysis methods for the multisite replication of partially nested experiments. A novel feature in this design is that replicating partially nested designs across multiple sites can create site-level random effects that are correlated yet distinct across treatment conditions. Such site- and condition-specific effects can arise because, for example, the treatment condition introduces an intermediate layer of clustering—such as shared provider influences—that introduces dependencies among individuals within the treatment condition in ways that are absent in the control group. Although a review of prior substantive literature suggests that multisite partially nested designs are embedded in both theory and practice, the deliberate design and proper analysis of these structures have largely been overlooked. In this study, we outline the motivation for this design, map out its use in practice, and develop methods to estimate treatment effects, quantify cross-site heterogeneity, and assess statistical power to detect both the treatment effect and the cross-site heterogeneity. Collectively, the results provide the tools to effectively and efficiently design and analyze studies drawing on multisite partially nested experiments.
Error detection and inhibition are deemed central to STEM learning, although there is little research on the neural bases of these processes and their role in the developing expertise of life sciences students. METHOD: Fifty undergraduate life sciences students performed a biology model evaluation task and a general go/no-go inhibition task during functional MRI. A year later, they completed a biology model evaluation task outside the scanner. RESULTS: Students showed increased ACC and bilateral inferior frontal/insula activity when viewing models with, vs. without, conceptual errors. Multivariate activation patterns showed little correspondence across the fMRI model and go/no-go tasks. Functional connectivity between the ACC and lateral PFC correlated with course grades, while activity in the right inferior frontal gyrus/insula predicted later model evaluation performance. CONCLUSIONS: Students who activate lateral PFC regions during scientific error detection may show better concept retention, although this neural activity may not reflect inhibitory control.


