Scholarlink

Background <p>This study aimed to develop and validate the AI-Supported Self-Regulated Science and Mathematics Learning Scale (AI-SSRSML Scale) for technology-enhanced educational settings in Tabuk Province, Saudi Arabia.</p> Methods <p> This study used an exploratory sequential mixed-methods design. In 2025, data were collected from students in Tabuk Province, Saudi Arabia. The scale was developed using a systematic literature review, 18 semi-structured interviews with secondary students, and iterative psychometric testing. Exploratory factor analysis (EFA), confirmatory factor analysis (CFA), exploratory graph analysis (EGA), convergent and discriminant validity, measurement invariance by gender, and an exploratory Random Forest regression model were used to analyze data from a pilot study (n = 60) and a large-scale survey (n = 920) to establish the construct validity and predictive utility of the scale. Reliability was investigated using Cronbach’s alpha, McDonald’s omega, composite reliability (CR), and intraclass correlation coefficients (ICC).</p> Results <p>EFA and CFA. results supported a six-factor solution: Meta-Cognitive Planning, Growth Mindset Endorsement, Self-Monitoring and Reflection, Strategic AI Utilization, Motivational Resilience, and Adaptive Help-Seeking, which accounted for 62.3% of the variance. The final AI-SSRSML Scale included 48 items (eight items per factor). The final model had acceptable fit indices (RMSEA = 0.075, CFI = 0.928, TLI = 0.911), and reliability indices showed excellent internal consistency and temporal stability (Cronbach’s α = 0.914–0.959; McDonald’s ω = 0.914–0.959; CR = 0.903–0.959; ICC = 0.758–0.869). The measurement invariance was supported by gender. The Random Forest analysis. results showed that items from the Meta-Cognitive Planning factor were the most important in predicting the overall AI-SSRSML score. </p> Conclusion <p>The AI-SSRSML Scale is a valid, reliable, and gender-invariant instrument for measuring AI-supported self-regulated learning in secondary science and mathematics education. The scale has the potential to provide a psychometrically sound foundation for research, intervention design, and educational policy to support students’ cognitive, motivational, and metacognitive engagement in technology-enhanced, growth-oriented educational settings.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

The development and validation of the AI-supported self-regulated science and mathematics learning scale (AI-SSRSML) among secondary school students in Saudi Arabia

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月26日

查看全文
  • Ataallh Aodh Alatoai,
  • Ali Saleh Alshahri

摘要

Background

This study aimed to develop and validate the AI-Supported Self-Regulated Science and Mathematics Learning Scale (AI-SSRSML Scale) for technology-enhanced educational settings in Tabuk Province, Saudi Arabia.

Methods

This study used an exploratory sequential mixed-methods design. In 2025, data were collected from students in Tabuk Province, Saudi Arabia. The scale was developed using a systematic literature review, 18 semi-structured interviews with secondary students, and iterative psychometric testing. Exploratory factor analysis (EFA), confirmatory factor analysis (CFA), exploratory graph analysis (EGA), convergent and discriminant validity, measurement invariance by gender, and an exploratory Random Forest regression model were used to analyze data from a pilot study (n = 60) and a large-scale survey (n = 920) to establish the construct validity and predictive utility of the scale. Reliability was investigated using Cronbach’s alpha, McDonald’s omega, composite reliability (CR), and intraclass correlation coefficients (ICC).

Results

EFA and CFA. results supported a six-factor solution: Meta-Cognitive Planning, Growth Mindset Endorsement, Self-Monitoring and Reflection, Strategic AI Utilization, Motivational Resilience, and Adaptive Help-Seeking, which accounted for 62.3% of the variance. The final AI-SSRSML Scale included 48 items (eight items per factor). The final model had acceptable fit indices (RMSEA = 0.075, CFI = 0.928, TLI = 0.911), and reliability indices showed excellent internal consistency and temporal stability (Cronbach’s α = 0.914–0.959; McDonald’s ω = 0.914–0.959; CR = 0.903–0.959; ICC = 0.758–0.869). The measurement invariance was supported by gender. The Random Forest analysis. results showed that items from the Meta-Cognitive Planning factor were the most important in predicting the overall AI-SSRSML score.

Conclusion

The AI-SSRSML Scale is a valid, reliable, and gender-invariant instrument for measuring AI-supported self-regulated learning in secondary science and mathematics education. The scale has the potential to provide a psychometrically sound foundation for research, intervention design, and educational policy to support students’ cognitive, motivational, and metacognitive engagement in technology-enhanced, growth-oriented educational settings.