Table 2.
Model consistency.
| Model | Consistency, mean (SD) | t test resultsa | |||||||
| GPT 3.5 | GPT 4 | GPT 4o | |||||||
|
|
|
t test (df) | P value | t test (df) | P value | t test (df) | P value | ||
| GPT-3.5 | 0.082 (0.0549) | —b | — | — | — | — | — | ||
| GPT-4 | 0.182 (0.0394) | 18.0 (298) | <.001 | — | — | — | — | ||
| GPT-4o | 0.186 (0.0324) | 19.6 (298) | <.001 | 0.917 (298) | .360 | — | — | ||
at tests were used to compare average response consistency between models with α=.01 after Bonferroni correction.
bNot applicable.