Table 4.

Within-participant comparisons for advice evaluations (Studies 4-5). Table 4 presents the mean evaluations for self (vs. AI) generated advice in Study 4 (top) and Study 5 (bottom). The columns represent the mean evaluations for ChatGPT and human-generated advice (“M”) and their corresponding standard deviations (“SD”). The right-hand side shaded portion of the table presents the results of within-participant t-tests with the mean difference between the variables, with associated t-statistic, p-value, 95% confidence interval (“Lower CI” and “Upper CI”), and Cohen’s d effect sizes. Recall that in Study 4, participants were aware they were evaluating ChatGPT-generated advice (and self-generated advice). In Study 5, they were told they were evaluating human-generated advice (and self-generated advice).

Variable	TARGET				STUDY 4
	ChatGPT		Self
	M	SD	M	SD	Mean diff.	t	p-value	Cohen’s d	Lower CI	Upper CI
Effectiveness	5.62	1.19	5.65	0.95	−0.03	−0.49	.624	−0.03	-0.15	0.09
Quality	5.60	1.28	5.47	1.20	0.13	1.70	.091	0.09	-0.02	0.28
Authenticity	5.25	1.44	6.31	0.76	−1.06	−13.66	< .001	−0.72	−1.22	−0.91

Variable	TARGET				STUDY 5
	ChatGPT		Self
	M	SD	M	SD	Mean diff.	t	p-value	Cohen’s d	Lower CI	Upper CI
Effectiveness	5.84	1.05	5.55	0.96	0.29	4.86	< .001	0.25	0.17	0.41
Quality	6.01	1.09	5.34	1.12	0.67	9.20	< .001	0.47	0.52	0.81
Authenticity	5.94	1.15	6.33	0.70	−0.39	−6.32	< .001	−0.32	−0.51	−0.27