Skip to main content

Table 3 LPA fit indices and model diagnostics for class solutions 1 through 6

From: Multimodal lifestyle engagement patterns support cognitive stability beyond neuropathological burden

Classes

Log likelihood

AIC

BIC

Smallest class size (% of sample)

BLRT

BLRT p-value

Entropy

Minimum average posterior probabilityb

1

−16,573.15

33,170.30

33,237.86

2059 (100%)

-

-

1.00

1.00

2

−15,738.43

31,514.85

31,621.82

383 (18.6%)

1669.45

< .001

0.90

0.92

3

−15,473.25

30,998.49

31,144.87

100 (4.9%)

530.36

< .001

0.90

0.79

4

−15,252.65

30,571.30

30,757.09

91 (4.4%)

441.19

< .001

0.89

0.84

5a

−15,094.37

30,268.74

30,493.94

64 (3.1%)

316.57

< .001

0.89

0.84

6

−14,978.71

30,051.43

30,316.03

62 (3.0%)

231.31

< .001

0.78

0.80

  1. aSelected as the most optimal class solution based on a combination of fit indices. bRepresents the minimal of the diagonal of the average latent class probabilities for most likely class membership