Table 6.

Univariate and multivariate logistic regression of GPT-4’s responses to MIR examination questions: effects of number of words, characters, and tokens (all in hundreds).

	Univariate Logistic Regression		Multivariate Logistic Regression
Length of the Question	OR (95% CI)	p	OR (95% CI)	p
Number of Words *	1.82 (95% CI 0.51–6.47)	0.351	1.33 (95% CI 0.03–53.57)	0.880
Number of Characters *	1.09 (95% CI 0.90–1.34)	0.370	1.70 (95% CI 0.01–259.69)	0.835
Number of Tokens *	1.30 (95% CI 0.74–2.33)	0.361	0.92 (95% CI 0.24–3.45)	0.896

* In hundreds.