Table 2 Regression analyses used to test...

Table 2

Regression analyses used to test our hypotheses

	Panel A: All cases			Panel B: Human		Panel C: Algorithm
	Coeff.	p value		Coef.	p value	Coef.	p value
Advice rationale	1.198	0.002 a	Advice rationale	1.198	0.003 a	1.852	0.000 a
Accountability	0.063	0.790	Accountability	0.063	0.790	0.467	0.243
Type of adviser	0.091	0.763
Advice rationale × Accountability	1.025	0.039 a	Advice rationale × Accountability	1.025	0.039 a	−0.227	0.738
Advice rationale × Type of Adviser	0.653	0.291	Constant	5.576	0.000	5.667	0.000
Accountability × Type of adviser	0.404	0.384
			Simple Effects
			Advice rationale @ Accountability = 1	2.223	0.000 a	1.625	0.001 a
Advice rationale × Accountability × Type of adviser	−1.251	0.081 a
			Accountability @ Advice rationale = 1	1.088	0.021 a	0.240	0.662
Constant	5.576	0.000
N	248		N	129		119
R_Sq	0.236		R_Sq	0.263		0.202

	Panel A: All cases			Panel B: Human		Panel C: Algorithm
	Coeff.	p value		Coef.	p value	Coef.	p value
Advice rationale	1.198	0.002 a	Advice rationale	1.198	0.003 a	1.852	0.000 a
Accountability	0.063	0.790	Accountability	0.063	0.790	0.467	0.243
Type of adviser	0.091	0.763
Advice rationale × Accountability	1.025	0.039 a	Advice rationale × Accountability	1.025	0.039 a	−0.227	0.738
Advice rationale × Type of Adviser	0.653	0.291	Constant	5.576	0.000	5.667	0.000
Accountability × Type of adviser	0.404	0.384
			Simple Effects
			Advice rationale @ Accountability = 1	2.223	0.000 a	1.625	0.001 a
Advice rationale × Accountability × Type of adviser	−1.251	0.081 a
			Accountability @ Advice rationale = 1	1.088	0.021 a	0.240	0.662
Constant	5.576	0.000
N	248		N	129		119
R_Sq	0.236		R_Sq	0.263		0.202

Note(s): Results are obtained using regression analysis with robust standard errors. p-values are two tailed, except where used for hypothesis testing, marked with an a

The dependent variable, advice-taking is measured on an 11-point Likert scale; higher values on the variable represent more advice-taking from the human/algorithmic adviser (depending on the type of adviser condition)

Type of adviser is an indicator variable equal to 1 (0) when the adviser was an algorithm (when the adviser was a human)

Advice rationale is an indicator variable equal to 1 (0) when an advice rationale was present (absent)

Accountability is an indicator variable equal to 1 (0) when accountability is high (low)

Source(s): Authors’ own work

[ViewLarge]

Sharing Unavailable