Table 2. Open in new tab Nested models...

Model

Predictor

R²

AIC

χ2

ChatGPT

Warmth direction

0.407

7.68

<0.001

Competence direction

0.193

2.85

0.005

0.141

1204.14

Warmth direction

0.408

7.78

<0.001

Competence direction

0.152

2.27

0.023

Warmth representativeness

−0.396

−2.23

0.026

Competence representativeness

0.640

4.23

<0.001

0.203

1179.00

29.14

<0.001

Warmth direction

−0.079

−0.57

0.302

Competence direction

−0.032

−0.37

0.745

Warmth representativeness

−0.448

−2.93

0.008

Competence representativeness

0.255

1.22

0.177

Warmth interaction

1.700

8.17

<0.001

Competence interaction

0.599

2.71

0.009

0.316

1113.87

69.13

<0.001

SBERT

Warmth direction

0.033

9.25

<0.001

Competence direction

0.026

7.59

<0.001

0.113

−2765.83

Warmth direction

0.028

9.411

<0.001

Competence direction

0.025

8.010

<0.001

Warmth representativeness

−0.108

−6.076

<0.001

Competence representativeness

0.042

2.004

0.045

0.161

−2802.37

40.54

<0.001

Warmth direction

−0.008

−1.99

0.046

Competence direction

0.003

0.83

0.407

Warmth representativeness

0.006

0.32

0.752

Competence representativeness

−0.093

−3.95

<0.001

Warmth interaction

0.313

13.62

<0.001

Competence interaction

0.280

8.72

<0.001

0.337

−3040.00

241.63

<0.001

Table 2.

Open in new tab

Nested models showing incremental validity of Warmth and Competence representativeness and its interaction with direction, for ChatGPT and SBERT, in predictions of internal valence.

Model	#	Predictor	b	t	P	R²	AIC	χ2	P
ChatGPT	1	Warmth direction	0.407	7.68	<0.001
	1	Competence direction	0.193	2.85	0.005	0.141	1204.14
	2	Warmth direction	0.408	7.78	<0.001
		Competence direction	0.152	2.27	0.023
		Warmth representativeness	−0.396	−2.23	0.026
		Competence representativeness	0.640	4.23	<0.001	0.203	1179.00	29.14	<0.001
	3	Warmth direction	−0.079	−0.57	0.302
		Competence direction	−0.032	−0.37	0.745
		Warmth representativeness	−0.448	−2.93	0.008
		Competence representativeness	0.255	1.22	0.177
		Warmth interaction	1.700	8.17	<0.001
		Competence interaction	0.599	2.71	0.009	0.316	1113.87	69.13	<0.001
SBERT	1	Warmth direction	0.033	9.25	<0.001
	1	Competence direction	0.026	7.59	<0.001	0.113	−2765.83
	2	Warmth direction	0.028	9.411	<0.001
		Competence direction	0.025	8.010	<0.001
		Warmth representativeness	−0.108	−6.076	<0.001
		Competence representativeness	0.042	2.004	0.045	0.161	−2802.37	40.54	<0.001
	3	Warmth direction	−0.008	−1.99	0.046
		Competence direction	0.003	0.83	0.407
		Warmth representativeness	0.006	0.32	0.752
		Competence representativeness	−0.093	−3.95	<0.001
		Warmth interaction	0.313	13.62	<0.001
		Competence interaction	0.280	8.72	<0.001	0.337	−3040.00	241.63	<0.001

Model	#	Predictor	b	t	P	R²	AIC	χ2	P
ChatGPT	1	Warmth direction	0.407	7.68	<0.001
	1	Competence direction	0.193	2.85	0.005	0.141	1204.14
	2	Warmth direction	0.408	7.78	<0.001
		Competence direction	0.152	2.27	0.023
		Warmth representativeness	−0.396	−2.23	0.026
		Competence representativeness	0.640	4.23	<0.001	0.203	1179.00	29.14	<0.001
	3	Warmth direction	−0.079	−0.57	0.302
		Competence direction	−0.032	−0.37	0.745
		Warmth representativeness	−0.448	−2.93	0.008
		Competence representativeness	0.255	1.22	0.177
		Warmth interaction	1.700	8.17	<0.001
		Competence interaction	0.599	2.71	0.009	0.316	1113.87	69.13	<0.001
SBERT	1	Warmth direction	0.033	9.25	<0.001
	1	Competence direction	0.026	7.59	<0.001	0.113	−2765.83
	2	Warmth direction	0.028	9.411	<0.001
		Competence direction	0.025	8.010	<0.001
		Warmth representativeness	−0.108	−6.076	<0.001
		Competence representativeness	0.042	2.004	0.045	0.161	−2802.37	40.54	<0.001
	3	Warmth direction	−0.008	−1.99	0.046
		Competence direction	0.003	0.83	0.407
		Warmth representativeness	0.006	0.32	0.752
		Competence representativeness	−0.093	−3.95	<0.001
		Warmth interaction	0.313	13.62	<0.001
		Competence interaction	0.280	8.72	<0.001	0.337	−3040.00	241.63	<0.001

Model comparison metrics are provided, comparing the three levels of nested regressions for each language model. Higher values for the outcome indicate more positive language model ratings of the social categories.

This Feature Is Available To Subscribers Only