EUVIMEDEUVIMED entdecken
Open Access · CC BY-SA 4.0

Exploring ChatGPT’s Proficiency in Nonparametric Statistics: An Initial Review and Benchmark Assessment

Joel Lagundi De Castro

International Journal in Information Technology in Governance, Education and Business · 18.04.2025 · 7 (1) · 1-23

DOI: 10.32664/ijitgeb.v7i1.174 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal in Information Technology in Governance, Education and Business, online erschienen am 18.04.2025. Verfasst von Joel Lagundi De Castro. Die Arbeit wurde bisher 1-mal zitiert (Crossref). Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Artificial Intelligence (AI) is transforming education, particularly in teaching statistics, by enhancing personalized learning and feedback through tools like ChatGPT (Tulsiani, 2024). ChatGPT is an advanced artificial intelligence chatbot developed by OpenAI that uses deep learning to understand and generate human-like text. It is based on the GPT (Generative Pre-trained Transformer) model, trained on vast amounts of text data to assist with answering questions, generating content, and engaging in natural conversations. This study evaluates ChatGPT version 3.5 performance in nonparametric statistical analysis by assessing its ability to generate solutions for seven tests, including the Test of Randomness, ANOVA, Chi-Square Goodness-of-Fit Test, Median Test, Cochran’s Q Test, Wilcoxon-Mann-Whitney Test, and Binomial Probability Test. Using three prompt engineering strategies—Basic Prompt (BP), Structured Prompt (SP), and Error-Awareness Prompt (EAP)—ChatGPT's outputs are compared against manual calculations and statistical software (Jeffreys’s Amazing Statistics Program(JASP) and Excel) for accuracy, consistency, and clarity. Results show significant discrepancies in Basic Prompt outputs between November 2023 and 2024, with sum of squares values of 6421.82 and 6928.00, and an F-value of 0.93 (p = 0.53), indicating no significant difference. Similarly, the effect of prompt type is statistically insignificant (F = 1.43, p = 0.26), as is the absolute error analysis (F = 0.59, p = 0.57). However, differences in statistical test approaches are significant (F = 3.10, p = 0.04), suggesting that method selection impacts accuracy. Findings emphasize the role of structured and error-aware prompts in improving ChatGPT’s performance, highlighting the importance of effective prompt engineering in nonparametric statistics. These insights contribute to improving AI-assisted learning in statistical education and research, ensuring more reliable computational outputs. Lastly, guidelines for effective prompt engineering in Nonparametric Statistics were formulated.Received Date: February 2, 22025Revised Date: March 18, 2025Accepted Date: March 30, 2025 Click to Access and Download the Article:

Zitieren

APA

Castro, J. L. D. (2025). Exploring ChatGPT’s Proficiency in Nonparametric Statistics: An Initial Review and Benchmark Assessment. International Journal in Information Technology in Governance, Education and Business, 7(1), 1-23. https://doi.org/10.32664/ijitgeb.v7i1.174

Vancouver

Castro JLD. Exploring ChatGPT’s Proficiency in Nonparametric Statistics: An Initial Review and Benchmark Assessment. International Journal in Information Technology in Governance, Education and Business. 2025;7(1):1-23. doi:10.32664/ijitgeb.v7i1.174

BibTeX
@article{Castro2025,
  author = {Castro, Joel Lagundi De},
  title = {Exploring ChatGPT’s Proficiency in Nonparametric Statistics: An Initial Review and Benchmark Assessment},
  journal = {International Journal in Information Technology in Governance, Education and Business},
  year = {2025},
  volume = {7},
  number = {1},
  pages = {1--23},
  doi = {10.32664/ijitgeb.v7i1.174}
}
Kostenlos starten →