# Validating the AIM–N: An AI-motivation and needs scale with multi-group invariance and MIMIC-DIF evidence in higher education

## TL;DR

This study introduces and validates a new scale to measure how AI affects student motivation and needs in higher education.

## Contribution

The paper presents the AIM-N scale, validated for multi-group invariance and DIF evidence in assessing AI's impact on student motivation.

## Key findings

- The AIM-N scale has a multi-factor structure with strong model fit and internal consistency across various student groups.
- Higher AI usage correlates with stronger redundancy beliefs and controlled motivation, particularly in STEM fields.
- Differential item functioning was identified, such as competitive field students endorsing AI pressure items more.

## Abstract

The rapid adoption of generative AI in higher education raises critical questions about its impact on student motivation and basic psychological needs. This study introduces and validates the AI-Motivation and Needs (AIM-N) scale, a new instrument assessing how AI integration influences students’ motivational orientations and need satisfaction in learning. Survey data were collected from N = 904 university students. A confirmatory factor analysis (CFA) supported a multi-factor structure for the AIM-N, comprising two subscales of AI-related redundancy beliefs (task-level and motivational-level) and three subscales of AI-related motivational orientations (intrinsic, identified, controlled), with acceptable model fit (CFI ≈ 0.96, TLI ≈ 0.95, RMSEA ≈ 0.05) and strong factor loadings. Internal consistency was good for most subscales (Cronbach’s α = 0.70–0.90; McDonald’s ω in similar range), except a single-item amotivation indicator. Multi-group CFA indicated that the AIM-N achieved configural, metric, and scalar invariance across gender, study level (Bachelor’s, Master’s, PhD), academic field, and frequency of AI use (ΔCFI < 0.01), after minor modifications for the AI-use groups. A MIMIC model (Multiple Indicators, Multiple Causes) revealed that higher AI tool usage was associated with stronger beliefs that AI renders learning tasks redundant and slightly more controlled motivation (β ≈ 0.30 and 0.21, p <.001), while gender showed no significant effects. Field of study had significant impacts: STEM students reported higher redundancy beliefs and controlled motivation than humanities students (p <.01). The MIMIC analysis also identified differential item functioning (DIF) for certain items; for example, students in competitive fields endorsed the “pressure to use AI” item more than expected from their latent trait levels. These results demonstrate that the AIM-N is a reliable and valid instrument for measuring the nuanced ways AI influences student motivation and needs. The discussion addresses theoretical implications for Self-Determination Theory in the age of AI, practical implications for educators, and recommendations for future research on sustaining meaningful student engagement when AI tools are pervasive.