Exploring the impact of intelligent learning tools on students' independent learning abilities: a PLS-SEM analysis of grade 6 students in China

The paper examines how intelligent learning tools—systems leveraging technologies such as big data, AI, learning analytics, and multimodal access—affect Grade 6 students’ learning, particularly independent learning abilities. Prior work highlights the importance of digital proficiency and self-directed learning in online environments and identifies potential determinants of tool use (e.g., self-efficacy, satisfaction, infrastructure). However, limited research clarifies how interaction quality (learner-learner, learner-instructor, learner-content) and information quality (accuracy, understandability, relevance, and richness of resources) influence satisfaction, intention to use, and ultimately independent learning abilities. The study addresses two research questions: (1) whether interaction quality and information quality affect students' satisfaction and intention to use intelligent learning tools; and (2) whether intention to use impacts students' independent learning abilities. The goal is to provide recommendations for optimizing tool design to strengthen independent learning.

The literature review frames interaction quality using Moore’s three interaction types (learner-learner, learner-instructor, learner-content), noting mixed findings on which most strongly predicts satisfaction, and evidence that interaction can increase intention to use. Information quality, situated in the DeLone and McLean (D&M) model, encompasses accuracy, understandability, relevance, and abundance of resources; prior studies generally link information quality to satisfaction, with mixed evidence on its effect on intention. Satisfaction is positioned as a key determinant of intention (and continuance) in ECM and D&M. Intention to use is central in TAM/UTAUT and linked to downstream outcomes, but its relation to independent learning ability is underexplored. Independent learning ability involves autonomy, self-regulation, and proactive strategy use; intelligent learning tools may both require and foster such abilities. Hypotheses: H1: Interaction quality positively influences satisfaction; H2: Interaction quality positively influences intention to use; H3: Information quality positively influences satisfaction; H4: Information quality positively influences intention to use; H5: Satisfaction positively influences intention to use; H6: Intention to use positively influences independent learning ability.

Design and participants: A cross-sectional survey using convenience sampling was administered online via Wenjuanxing in 2022 to Grade 6 students at a public primary school in Hangzhou, Zhejiang Province, China. The sixth grade was chosen due to its critical role in developing independent learning abilities. A total of 384 valid responses were obtained from students who had previously used intelligent learning tools. Demographic summaries indicated varied academic levels, internet experience, and usage frequency of intelligent tools. Instrument: The questionnaire comprised two parts: demographics and 19 items across five constructs measured on a 5-point Likert scale (1=strongly disagree to 5=strongly agree). Constructs and items: Information quality (4 items: accuracy, comprehensibility, relevance, resource richness), Interaction quality (4 items: learning reports, personalized recommendations, teacher interaction, peer interaction), Independent learning ability (4 items: planning, time management, method adaptation, reflection), Satisfaction with tools (3 items), Intention to use (4 items). Items were adapted from established sources (e.g., DeLone & McLean, Pituch & Lee, Zimmerman, Bhattacherjee, Roca et al.). Analysis: Partial Least Squares Structural Equation Modeling (PLS-SEM) using Smart-PLS 3 was employed. A two-step approach assessed the measurement model (reliability, convergent and discriminant validity) and then the structural model (collinearity via VIF, path coefficients with bootstrapping of 5,000 samples, R², Q²). Common method bias was checked via Harman’s single-factor test.

Common method bias: Largest single factor explained 46.127% (<50%), indicating no serious CMB. Model fit: SRMR=0.066 (<0.08), d_ULS=0.825, d_G=0.345 (<0.95), NFI=0.902; good fit. Measurement model: Cronbach’s alpha >0.810; rho_A ≥0.840; CR=0.888–0.953; AVE=0.670–0.834; all loadings >0.70; Fornell-Larcker and HTMT (0.672–0.804) supported discriminant validity. Collinearity: Inner VIFs 1.000–2.199, indicating no collinearity issues. Structural paths (bootstrapped; Table 8): - H1 ITQ→ST: β=0.519, t=9.451, p<0.001 (Supported) - H2 ITQ→IU: β=0.097, t=1.550, n.s. (Unsupported) - H3 IFQ→ST: β=0.282, t=5.107, p<0.001 (Supported) - H4 IFQ→IU: β=0.165, t=3.705, p<0.001 (Supported) - H5 ST→IU: β=0.617, t=12.497, p<0.001 (Supported) - H6 IU→ILA: β=0.727, t=25.051, p<0.001 (Supported) Explained variance (Table 9): ST R²=0.531; IU R²=0.644; ILA R²=0.528. Predictive relevance: Q² ST=0.376; IU=0.426; ILA=0.436. Mediation (Table 10): Satisfaction fully mediated ITQ→IU (indirect=0.320, p<0.001; direct n.s.) and showed complementary partial mediation for IFQ→IU (indirect=0.174, p<0.001; direct=0.165, p<0.001). Overall, intention to use strongly and directly predicted independent learning ability; information quality and satisfaction directly influenced intention, while interaction quality influenced intention only indirectly via satisfaction.

The findings address the research questions by showing that information quality and satisfaction significantly shape students’ intention to use intelligent learning tools, and that intention to use is a strong predictor of independent learning ability. Interaction quality elevates satisfaction but does not directly increase intention; rather, its contribution to intention is fully mediated through satisfaction. This implies that enhancing users’ perceived value and contentment with the tools is pivotal to fostering use intentions, which in turn supports self-regulatory and autonomous learning behaviors. The salience of information quality highlights the central role of accurate, understandable, relevant, and resource-rich content in primary students’ acceptance and use. The strong path from intention to independent learning ability suggests that cultivating willingness to use such tools can translate into more effective planning, time management, adaptive strategies, and reflective learning—key elements of independent learning.

This study contributes evidence that, among Grade 6 students in China, intelligent learning tools’ information and interaction qualities increase satisfaction; satisfaction and information quality, in turn, drive intention to use; and intention robustly predicts independent learning ability. Interaction quality does not directly affect intention but operates via satisfaction. Practical recommendations include prioritizing user-friendly, multi-modal interaction features (learner-learner, learner-instructor, learner-content), timely feedback, personalized recommendations, and high-quality, comprehensible learning content with abundant resources. Future research should broaden sampling beyond a single school, differentiate types of interaction to test distinct effects, examine the roles of teachers and parents in mediating tool use and independent learning, and explore guidance strategies that help students leverage intelligent tools to build independent learning capacities.

The sample was limited to one primary school in Hangzhou, which may restrict generalizability. The study did not disaggregate interaction types beyond the broad construct when testing impacts on intention; different interaction modes may have varied effects. The cross-sectional design limits causal inference, and self-reported measures may introduce response bias.