. 2022 Jul 22;13:897901. doi: 10.3389/fpsyg.2022.897901

TABLE 5.

Bifactor graded response model parameters for the final DVMSQ model in Prolific and SPARK samples.

	Prolific sample										Spark sample
	a ₁	a ₂	a ₃	a ₄	a ₅	d ₁	d ₂	d ₃	d ₄	a ₁		a ₂	a ₃	a ₄	a ₅	d ₁	d ₂	d ₃	d ₄
Item 1	1.96	1.09	−	−	−	5.69	3.46	0.49	−2.35	1.60		0.77	−	−	−	6.03	4.48	2.06	−0.01
Item 2	5.15	5.83	−	−	−	6.03	0.41	−5.70	−11.90	4.32		5.36	−	−	−	7.38	3.59	−2.27	−6.93
Item 4	0.82	0.55	0.66	−	−	1.46	0.56	−0.98	−2.66	1.00		0.64	−0.07	−	−	1.82	0.78	−0.53	−1.97
Item 5	2.85	−	2.17	−	−	2.09	0.07	−2.44	−5.61	2.34		−	1.00	−	−	4.06	2.80	1.10	−0.90
Item 6	2.07	−	0.97	−	−	2.80	1.16	−0.92	−2.76	2.41		−	1.84	−	−	6.74	5.23	2.64	0.45
Item 7	1.78	1.37	−	−	−	−0.16	−1.85	−3.48	−5.18	1.55		1.40	−	−	−	1.16	−0.29	−2.01	−3.39
Item 8	2.24	−	−	−	−	0.65	−0.68	−2.43	−4.55	1.70		−	−	−	−	2.61	1.29	−0.29	−1.73
Item 9	2.47	−	−	−	−	3.75	1.98	−0.34	−2.75	2.27		−	−	−	−	6.25	4.59	1.91	0.03
Item 10	1.32	−	0.30	−	−	0.18	−0.47	−1.60	−2.91	1.31		−	0.11	−	−	1.73	1.34	0.28	−0.94
Item 12	1.29	−	0.30	−	−	1.39	−0.10	−1.75	−3.49	1.25		−	0.25	0.49	−	3.54	2.21	0.41	−1.20
Item 13	2.53	−	−	2.59	−	−1.67	−4.96	−7.52	−9.60	2.44		−	−	3.25	−	2.68	−0.80	−4.67	−7.70
Item 14	1.38	−	−	1.70	−	−0.79	−2.80	−4.61	−6.52	1.43		−	−	2.42	−	0.91	−1.00	−3.13	−5.14
Item 15	2.12	−	−	2.67	−	−2.75	−4.92	−7.26	−8.83	1.69		−	−	2.25	−	0.75	−1.39	−3.67	−6.04
Item 16	1.79	−	−	2.73	−	−3.36	−5.24	−6.92	−8.87	1.25		−	−	1.98	−	0.04	−0.98	−2.40	−3.73
Item 18	3.52	−	−	2.87	1.98	−0.13	−4.35	−7.66	−10.69	3.00		−	−	2.55	1.99	5.51	1.26	−2.82	−6.30
Item 19	5.33	−	−	4.28	3.52	−0.18	−6.89	−11.19	−15.65	3.57		−	−	2.82	2.74	7.31	1.83	−2.52	−7.34
Item 20	3.35	−	−	2.78	1.84	−2.67	−6.17	−8.19	−10.46	2.56		−	−	1.90	1.83	2.85	−0.30	−3.43	−5.77

Both models assume that all latent variables are orthogonal and have a standard normal distribution (i.e., M = 0, SD = 1) in the population. Differences in intercept terms between the two groups are not significant after considering the higher mean scores on the misophonia latent trait in the SPARK sample. a₁–a₅ = slope parameters (higher values indicate stronger relationships with the latent variables [i.e., stronger factor loadings]); d₁–d₄ = intercept parameters (higher values indicate “less difficult” or more easily endorsed item categories).