Development and Validation of ChatGPT Reliance Scale

Abstract

As ChatGPT becomes more common in universities, many people are concerned about its effects on students’ learning, research quality, originality, and even procrastination habits. In response, this study developed and validated a detailed ChatGPT Reliance Scale for university students. The goal was to provide a reliable and accurate tool for capturing the complex ways students use ChatGPT in higher education. To this end, we conducted a cross-sectional survey of 1,000 students from various universities and postgraduate colleges in Abbottabad and Peshawar. An exploratory factor analysis (EFA) was performed on the first segment of the dataset, comprising 409 participants. The EFA indicated a three-factor structure that explained 61.99% of the total variance. A confirmatory factor analysis (CFA) was then conducted on the remaining 400 participants, supporting the three-factor model and yielding acceptable fit indices (χ² (235) = 1074, p < .001; CMIN/DF = 4.57; CFI = 0.91; GFI = 0.91; RMSEA = 0.06). All factor loadings were statistically significant (p < .001) and ranged from 0.54 to 0.85. Overall, the scale demonstrated strong internal consistency (Cronbach’s α = 0.92, McDonald’s ω = 0.93), suggesting it is a reliable instrument for assessing how and to what extent university students rely on ChatGPT.