Students’ engagement in a learning environment has been a crucial factor in evaluating the quality of learning and predicting their academic success. With technologies being more applied in science education for enhancing learning engagement, it is important to study how learning environments with different levels of technological engagement (LTEs) affect students’ learning. Data were collected from 168 fourth-grade students, ages 9–10, divided into 3 groups of different learning environments: (1) game-based instruction (high LTE) and (2) video-based instruction (medium LTE) as the two experimental groups, and (3) traditional instruction (low LTE) as the control group. Students’ learning was monitored with two tests: scientific concept test and the scientific argumentation skills test. Descriptive and inferential statistics including t-test, ANCOVA, and the Johnson-Neyman procedure were used to examine how different LTEs affect students’ learning as a group and by level (high-, medium-, and low-achieving). Findings demonstrated that (a) three forms of technology-supported learning significantly improve students’ scientific knowledge and argumentation skills; (b) higher LTEs lead to significantly better learning of scientific knowledge, except for some high-achieving students who can learn equally well in traditional instruction; and (c) higher LTEs lead to significantly better learning of argumentation skills across all students.

As game-based learning continues to draw attention, students’ perceptions toward classroom activities are vital in influencing the quality of learning. This study used the social cognitive theory to show the perception traits of learning and playing in game-based environments and for students to identify their self-efficacy toward game-based learning by different trait groups. The game Formosa Hope was used in an experiment with 362 fifth- and sixth-grade students at ages 11 to 12 years as participants. Three perception traits were identified through a two-step cluster analysis: I—strong perceptions of learning and playing, II—moderate perceptions of learning and playing, and III—strong perception of playing but weak perception of learning. This study showed that regardless of trait type, students demonstrated positive self-efficacy, with those with Trait I having significantly higher self-efficacy than those with Traits II and III, indicating that students’ positive perceptions of learning and playing are essential in prompting self-efficacy in game-based learning.

Predicting students' line of actions helps educators give adequate guidance to students, but this remains a challenge in science, technology, engineering, and mathematics (STEM) education. Given this, there is a scarcity of related research that will help improve teachers' prediction capabilities on students' line of actions when tackling ill-defined problems (IDPs), as well as how emerging data mining techniques could contribute to such prediction. The present study aims to fill the gap by measuring the quality of teachers' predictions (labeled expert prediction), where 43 elementary teachers predict students' step-by-step actions when solving an IDP through the light path task (LPT), and then comparing its quality with that of machine prediction, executed via sequential pattern mining techniques. Data on students' lines of action were collected from 501 5th- and 6th-grade students, aged 11-12. The results showed the significantly lower accuracy of expert prediction compared to machine prediction, which highlights the advantages of using data mining in predicting students' actions and shows its possible application as a recommendation system to provide adaptive guidance in future STEM education.

Previous studies measured flow states using students’ self-reported experiences, resulting in issues regarding nonobjective and nonreal-time data. Thus, this study used an electroencephalogram (EEG) to measure the EEG-detected real-time flow states (EEG-Fs) of 30 students from the 4th and 5th grades. Their EEG measurements, self-reported reflective flow experiences (SR-Fs), grade levels (GLs), balance of challenge and skill (BCS), and sense of control, represented by their overall test performance (OA-tp) and momentary test performance (MOM-tp), were analyzed to establish their EEG-F’s construct. Based on the results of a chi-square test, the EEG-F correlates significantly with SR-F, BCS, OA-tp, and MOM-tp. A J48 decision tree analysis and logistic regression further revealed that in-flow experiences (in-EEG-F) were detected when students had high SR-Fs, where the BCS contributed to flow states. In particular, students with a low-challenge/high-skill BCS demonstrated an in-EEG-F state upon having a high OA-tp. For high-challenge/high-skill, the in-EEG-F state was determined through their MOM-tp. Through the EEG and flow state construct, this study revealed a whole-part association between students’ momentary and overall reflective flow experiences and identified viable paths for inducing students’ EEG-Fs, which can contribute to future e-learning development when integrated with a brain-computer interface for e-learning or e-evaluation systems.

Studies have shown that although traditional instruction is effective in developing students’ conceptual understanding, it lacks in strengthening their thinking skills. To enhance traditional instruction and establish an approach using game-based instruction, this study developed Multi-genre digital Game-based Instruction (MGI), which integrates various game genres—situated and competitive—with traditional instruction. A total of 115 fourth-graders (9–10 years old) of 4 classes participated in the study, of which 2 classes made up the MGI (experimental) group and the other 2, the traditional instruction (control) group. Quantitative and qualitative data were collected to answer research questions by examining the comparative effectiveness of the two groups separately as well as the effectiveness by achievement level, and exploring students’ learning experiences. Findings showed that using MGI resulted in a significant improvement in conceptual understanding and argumentation skills compared to those who were taught only with traditional instruction. Furthermore, it showed that MGI was significantly more effective than traditional instruction in improving low-achieving students’ conceptual understanding, as well as improving argumentation skills for all achievement levels. Students’ learning experiences with MGI showed how the proposed instruction is superior compared to existing traditional instruction.