CoChat

The DOI matches its Crossref record; 71 of the 74 references it cites were found in a registry or an index.

Development and validation of generative artificial intelligence attitude scale for students

Open access
Marengo, Agostino, Karaoglan-Yilmaz, Fatma Gizem, Yılmaz, Ramazan, Ceylan, Mehmet · 2025 · Frontiers in Computer Science

Abstract

Introduction Generative artificial intelligence (AI) tools, such as ChatGPT, have gained significant traction in educational settings, offering novel opportunities for enhanced learning experiences. However, limited research has investigated how students perceive and accept these emerging technologies. This study addresses this gap by developing a scale to assess university students’ attitudes toward generative AI tools in education. Methods A three-stage process was employed to develop and validate the Generative AI Attitude Scale. Data were collected from 664 students from various faculties during the 2022–2023 academic year. Expert evaluations were conducted to establish face and content validity. An exploratory factor analysis (EFA) was performed on a subset of 400 participants, revealing a two-factor, 14-item structure that explained 78.440% of the variance. A subsequent confirmatory factor analysis (CFA) was conducted on a separate sample of 264 students to validate this structure, resulting in the removal of one item and a final 13-item scale. Results The 13-item scale demonstrated strong reliability, evidenced by a Cronbach’s alpha of 0.84 and a test–retest reliability of 0.90. Discriminative power was confirmed through corrected item-total correlations between lower and upper percentile groups. These findings indicate that the scale effectively differentiates student attitudes toward generative AI tools in educational contexts. Discussion The newly developed Generative AI Attitude Scale offers a valid and reliable instrument for measuring university students’ perspectives on integrating generative AI tools, such as ChatGPT, into educational environments. These results highlight the potential for more targeted research and informed implementation strategies to enhance learning outcomes through generative AI.

Overview

CoChat Summary

TL;DR

This study developed and validated a 13-item Generative AI Attitude Scale for university students, yielding a two-factor structure of positive and negative attitudes toward generative AI in education. The scale showed strong reliability and construct validity in a Turkish university sample.

Key findings

  • EFA on 400 students produced a two-factor, 14-item structure explaining 78.440% of total variance; CFA on 264 students led to removal of one item, resulting in a final 13-item scale.
  • Reliability was strong: overall Cronbach’s alpha = 0.84 and test–retest reliability = 0.90; positive attitude subscale α = 0.97, test–retest = 0.98; negative attitude subscale α = 0.94, test–retest = 0.96.
  • CFA fit indices for the two-factor model were χ²/df = 2.126, CFI = 0.98, GFI = 0.93, IFI = 0.98, TLI = 0.98, RMSEA = 0.065, SRMR = 0.0367.
  • Corrected item–total correlations after item removal ranged from 0.576 to 0.684 for positive attitude and 0.344 to 0.440 for negative attitude; all items showed significant discriminative power between upper and lower 27% groups.

Methods

  • Sequential exploratory mixed-method scale development: literature-based item generation (21 items), review by 15 students with generative AI experience, expert review by eight experts, pilot testing with 14 students, and final 14-item 5-point Likert scale.
  • Participants were 664 students from nine faculties, three vocational schools, and a graduate school at one Turkish state university during 2022–2023; EFA used n = 400 (217 female, 183 male), CFA used n = 264 (137 female, 127 male), and test–retest used n = 25 (13 female, 12 male) after excluding two of 27 initial participants.
  • Analyses included EFA with principal component analysis and varimax rotation, CFA in AMOS, Cronbach’s alpha, test–retest reliability, corrected item–total correlations, and comparisons between lower and upper 27% groups.

Limitations

  • Stated: the sample came from a single state university across various faculties and grade levels, limiting generalizability; many participants reported using ChatGPT specifically; the study focused on university students with generative AI experience and on short-term attitudes.
  • Stated: future research should include diverse universities, different generative AI applications, broader student populations, and longitudinal designs to examine long-term effects on learning outcomes, critical thinking, and motivation.
  • Evident from the text: the test–retest sample was small (n = 25), and the scale relies on self-reported Likert responses, which may not capture behavioral or contextual factors.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.3389/fcomp.2025.1528455

Cites

71 of 74 references verified0 citation errors · 3 not checkable

Statistics

Verification

Claims the paper makes

0 Valid · 1 Not valid · 2 Not checked · 80 Own results and interpretations

27 more claims

Create a free account to see all 83 claims and every source verdict.

See all claims — free account
  • It may state two separate claims as one. 16
  • Not yet classified. 21
  • The quote only partly states this claim. 3
  • Probably not a substantial claim. 13

Create a free account to see these claims.

Statistics recomputed

Every reported test result, recomputed from its test statistic and degrees of freedom. A result shows as not adding up only once an independent check has confirmed it.

  • Consistentχ²(91) = 5788.732reported p = 0 · recomputed p < .001

    “χ2 (91) = 5788.732, p = .000”

1 of 1

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

4 of 4

Citations

Cites

Research works printed but not linked: 8

  1. Akbulut, 2010, Sosyal bilimlerde SPSS uygulamaları [SPSS applications in social sciences]
  2. Buyukozturk, 2010, Sosyal bilimler için veri analizi el kitabı [the data analysis handbook for social sciences]
  3. Eagly, 1993, The psychology of attitudes
  4. Hooper, 2008, Structural equation modelling: guidelines for determining modelfit, Electron. J. Bus. Res. Methods, 6, 53
  5. Kline, 2005, Principle and practice of structural equation modelling
  6. Rogers, 2003, Diffusion of Innovations
  7. Tabachnick, 2007, Using multivariate statistics
  8. Yavuz, 2006, Developing a technology attitude scale for pre-service, Chemistry Teachers, 4

Cited by

See all 54

Metadata

Authors
Marengo, A., Karaoglan-Yilmaz, F. G., Yılmaz, R., & Ceylan, M.
Affiliations
  • University of Foggia (IT)
  • Bartin University (TR)
Corresponding author
Marengo, Agostino
Venue
Frontiers in Computer Science
Published
2025-02-25
Volume
7
ISSN
2624-9898 (ISSN-L)
OpenAlex
W4407929143
Cited by
54
References
73
Citations by year
2026: 33 · 2025: 21
FWCI
55.66
Type
journal article
Language
en
Open access
Locations
  • Frontiers in Computer Science (publishedVersion, cc-by)
  • SSRN Electronic Journal (acceptedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Topics
  • Social Sciences › Education › Artificial Intelligence in Education
  • Social Sciences › Gender Studies › Gender and Technology in Education
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
Keywords
generative AI, student attitudes, scale development, ChatGPT, university students, artificial intelligence in education, technology acceptance, psychometric validation

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-09-28, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-09-28 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.