Patterns of digital health access and use among US adults: a latent class analysis

BMC Digital Health

Table 2 Model fit information and selection criteria for latent class models with 1 to 20 classes

Classes	# Parameters	Log-likelihood	Entropy	AIC	CAIC	BIC	SABIC	VLMR-LR
Desideratum:	-	-	> 0.8	Lowest value	Lowest value	Lowest value	Lowest value	Greater magnitude indicates greater improvement over the previous model^*
1	64	-299,539	-	599,206	599,753	599,689	599,485	-
2	129	-214,242	0.99	428,742	429,845	429,716	429,306	164,838
3	194	-172,165	0.99	344,718	346,376	346,182	345,566	81,315
4	259	-159,902	0.96	320,321	322,535	322,276	321,453	23,700
5	324	-154,752	0.96	310,152	312,921	312,597	311,567	9952
6	389	-149,695	0.97	300,168	303,493	303,104	301,868	9772
7	454	-146,777	0.97	294,461	298,341	297,887	296,445	5640
8	519	-143,929	0.94	288,896	293,332	292,813	291,163	5503
9	584	-141,397	0.93	283,962	288,953	288,369	286,513	4893
10	649	-139,854	0.92	281,007	286,553	285,904	283,842	2981
11	714	-139,262	0.91	279,952	286,054	285,340	283,071	1145
12	779	-138,713	0.90	278,984	285,641	284,862	282,387	1061
13	844	-138,198	0.89	278,085	285,298	284,454	281,772	994
14	909	-137,722	0.88	277,262	285,031	284,122	281,233	920
15	974	-137,292	0.88	276,533	284,857	283,883	280,788	831
16	1039	-136,895	0.88	275,867	284,747	283,708	280,406	769
17	1104	-136,561	0.88	275,330	284,765	283,661	280,152	645
18	1169	-136,233	0.88	274,805	284,796	283,627	279,912	633
19	1234	-135,978	0.87	274,423	284,969	283,735	279,814	494
20	1299	-135,749	0.87	274,096	285,198	283,899	279,771	442

AIC Akaike information criterion, BIC Bayesian information criterion, CAIC Consistent AIC, SABIC Sample size adjusted BIC, VLMR-LR Vuong-Lo-Mendell-Rubin adjusted likelihood ratio (of k class model to k-1 class model, hence none for 1 class model)
^*All likelihood ratio tests have p < 0.001, indicating statistically significant improvement of each k-class model compared to previous k-1 class model
Fit for latent class models without covariates
Bold indicates selected model

ISSN: 2731-684X