Table 2.

Comparing model fit for different classes profiles in studied models.

Classes	AIC	BICa	BLMRT (p)	Entropy	Latent Class Prob. [Range]	Class Prevalence [Range]
Unconstrained modelss
1	9315.57	9329.59	---	---	---	---
2	9260.06	9288.61	p = 0.0099	0.575	[0.8–0.92]	[0.39–0.61]
3	9239.02	9282.10	p = 0.0396	0.761	[0.87–0.92]	[0.18–0.48]
4	9197.48	9255.10	p = 0.0099	0.835	[0.86–0.99]	[0.08–0.54]
5	9201.40	9273.55	p = 0.2871	0.855	[0.89–0.94]	[0.07–0.41]
Constrained modelss
1	9315.57	9329.59	---	---	---	---
2	9304.84	9322.48	p = 0.0099	0.686	[0.91–0.91]	[0.49–0.51]
3	9278.02	9299.31	p = 0.0099	0.795	[0.9–0.96]	[0.02–0.5]
4	9268.63	9293.55	p = 0.0198	0.81	[0.75–0.96]	[0.02–0.48]
5	9269.58	9298.13	p = 0.3069	0.754	[0.71–0.88]	[0.01–0.44]

Legend: AIC—Akaike Information Criterion, BIC—sample size adjusted Bayesian Information Criterion, BLRT(p)—bootstrapped Lo-Mendell-Rubin test. Lower AIC and BICa values indicate better fitting models. Significant p-values for the BLRT(p) indicate that the model is a better fit than a model which is 1 class lower. Entropy is a measure of how accurate a model is at classifying people into latent profiles. This is a number ranging from 0 to 1 and the value of 1 indicates the highest confidence. Latent class probabilities [range] provide a range (i.e., minimum and maximum) of probabilities of assigning each observation to a particular class. The higher the lower value of the range, the greater the likelihood of assigning observations to particular classes. Class prevalence [range]—the percentage of observations in the smallest and largest class. It should be higher than 0.05. In our case, the smallest class in the unconstrained models contains 8% of the observations for the whole group, so it is acceptable.