Are We There Yet? AI for Statistical Analysis
Author: Dr. Tal Slemrod and Dr. Eleazar “Trey” Vasquez; info@ciddl.org
Statistical Analysis and AI
As we start 2024, one of the newest (if not the newest) push in technology is the introduction and use of Artificial Intelligence (AI). Where and how AI will be used is one of the ongoing questions in both K12 and higher education. More specifically, one question that the CIDDL team is exploring is how AI can and should be used in data analysis.
AI vs Traditional Analyses
To explore this question, we decided to run statistical analyses using OpenAI’s ChatGPT 4.0 Data Analysis and Wolfram Alpha and compare the same analysis results with traditional statistical analysis software (SPSS).
We ran various statistical analyses using large data sets, including descriptive statistics, t-tests, ANOVAs, and chi-square. ChatGPT could understand all the variables and suggest types of statistical analysis when asked about the best ways of comparing variables. Similarly, we ran various statistical tests using Wolfram Alpha.
Was AI Accurate?
After running statistical analyses in all three platforms, similar results were found when conducting descriptive statistics, t-tests, and ANOVA. Chi-square analysis, however, was only sometimes accurate. While there were consistencies in the descriptive statistics results, there were differences in p-values when other analyses were calculated. When p-values were found significant in SPSS, results remained significant when using AI. However, p-values varied in how close they were to SPSS. Open AI’s ChatGPT 4.0 also suggested the correct type of statistical tests when asked. While Wolfram Alpha does not suggest types of analysis, descriptive statistics were found to be accurate. Interestingly, when asked, ChatGPT provided a step-by-step guide on running the analysis in SPSS, which can provide a worthwhile implementation tool to guide researchers conducting quantitative analysis.
Are We There Yet?
There is no doubt that the use of AI is and will continue to be an essential part of society and education. When it comes to statistical analysis, there is also little doubt that AI has the potential to bridge a significant gap for those who conduct research in higher education, K-12 schools, and the private sector. Additionally, it can potentially revolutionize how educators, researchers, and students determine best practices. Even more so, it expands the potential to shorten and simplify research and to shorten the research-to-practice gap. That being said, are we there yet? We don’t think we are quite there, but we see the signs ahead and are pretty sure we’ll get there soon.
What Do You Think?
Share your thoughts with us in our community.