The Exploratory Factor Analysis (EFA) produced a three-factor solution based on eigenvalues greater than 1 and inspection of the scree plot. These three factors together ...
The Exploratory Factor Analysis (EFA) produced a three-factor solution based on eigenvalues greater than 1 and inspection of the scree plot. These three factors together explain a substantial proportion of the total variance, suggesting a reasonably strong underlying structure in the dataset.
Based on the Rotated Component Matrix (Varimax rotation), the items clustered into the following factors:
- Factor 1: Perceived Usefulness of AI
(e.g., items related to improving writing quality, saving time, enhancing ideas) - Factor 2: Ethical Concerns
(e.g., items addressing plagiarism, fairness, and academic integrity) - Factor 3: Writing Confidence
(e.g., items reflecting self-efficacy and comfort in writing tasks)
Overall, the item groupings largely align with theoretical expectations. The constructs were designed to capture distinct dimensions of students’ perceptions of AI in writing, and the EFA results support this conceptual framework. The clear separation between usefulness, ethics, and confidence suggests good construct validity.
However, there were some unexpected loadings and minor cross-loadings. For example, one item related to “AI helps me express ideas more clearly” loaded moderately on both Factor 1 (usefulness) and Factor 3 (confidence). This may indicate that students interpret AI support not only as a functional tool but also as something that enhances their self-confidence. Additionally, a few items under ethical concerns showed slightly lower loadings (< .50), suggesting they may not strongly represent the intended construct.
These cross-loadings and weaker items highlight areas for improvement. In future revisions, it would be beneficial to refine or remove ambiguous items, ensure clearer wording, and possibly conduct a follow-up Confirmatory Factor Analysis (CFA) to validate the factor structure.
In conclusion, the EFA results reveal a theoretically meaningful and interpretable three-factor structure, with minor issues that can be addressed to strengthen the scale’s construct validity.