Motivational profiles and change in physical activity during a weight loss intervention: a secondary data analysis

Table 1 Fit Statistics, Homogeneity, and Separation

Fit Statistics
							H0: K classes; H1: K + 1 classes
Model (K-class)	Model Log-likelihood	Number of parameters	AIC	BIC	CAIC	AWE	LRT	Adj LMR P value	Bootstrapped LRT P value
1-class	− 937.06	8	1890.12	1915.16	1923.16	1980.20	–	–	–
2-class	− 868.06	17	1770.12	1823.33	1840.33	1961.54	135.08	0.0003	< 0.0001
3-class	− 835.14	26	1722.28	1803.66	1829.66	2015.03	64.45	0.0203	< 0.0001
4-class	− 812.95	35	1695.90	1805.44	1840.44	2089.99	43.44	0.2813	0.0400
5-class	− 798.62	44	1685.95	1822.95	1866.95	2180.67	28.06	0.7423	0.3750
Homogeneity for 3-class Model (values > 0.90 = low degree of homogeneity, values < 0.60 = high degree of homogeneity in bold)
	External	Introjected	Identified	Intrinsic	Class Label
class 1	0.78	0.77	0.64	0.50	Moderate Combined
class 2	0.13	1.17	0.50	0.25	High Autonomous
class 3	0.18	0.44	0.30	0.19	High Combined
Separation for 3-class Model (Cohen’s d) (values < 0.85 = low separation; values > 2.0 = high separation in bold)
		External	Introjected		Identified			Intrinsic
class 1 vs. 2		1.90	0.46		1.71			2.68
class 1 vs. 3		1.09	1.42		2.16			2.49
class 2 vs. 3		5.10	0.72		0.30			0.38

Fit statistics from class enumeration process using latent profile analysis where covariances were fixed to zero, but variances were allowed to differ across classes; AIC Akaike Information Criteria, BIC Bayesian Information Criteria, CAIC Consistent Akaike’s Information Criteria, AWE Approximate Weight of Evidence Criterion, LRT Likelihood ratio test, Adj LMR Adjusted Lo-Mendell-Rubin Likelihood Ratio Test; Homogeneity presented as within-class variance term, with low values indicating whether individuals within a class are similar to each other with respect to item responses. Separation presented as Cohen’s d, with high values indicating that individuals across two classes are dissimilar with respect to item responses

ISSN: 1479-5868