Skip to main content

Table 5 Parameter Estimates of the 3-Class Solution in the TRacking Adolescent Individuals’ Lives Survey (TRAILS), The Netherlands, 2001–2017

From: Variance constraints strongly influenced model performance in growth mixture modeling: a simulation and empirical study

Model

Parameter

Class 1

 

Class 2

 

Class 3

 
  

Mean

Variance

Mean

Variance

Mean

Variance

0: Unconstrained

Intercept

0.358

0.024

0.325

0.008

0.145

0.009

 

Linear Slope

0.087

0.001

0.008

0.001

0.005

0.001

 

Quadratic Slope

−0.033

0a

−0.048

0a

−0.024

0a

 

Cubic slope

0.003

0a

0.008

0a

0.004

0a

 

T1

0.038

0.049

0.01

 

T2

0.04

0.03

0.008

 

T3

0.041

0.019

0.01

 

T4

0.059

0.01

0.002

 

T5

0.05

0.006

0.001

 

T6

0.049

0.011

0.001

1A: Residual variance time constrained

Intercept

0.460

0.012

0.283

0.008

0.136

0.011

Linear Slope

0.078

0.003

0.039

0.001

0.024

0.001

Quadratic Slope

−0.039

0a

− 0.042

0a

− 0.03

0a

Cubic slope

0.004

0a

0.006

0a

0.004

0a

T1

0.058

0.023

0.005

T2

0.058

0.023

0.005

T3

0.058

0.023

0.005

T4

0.058

0.023

0.005

T5

0.058

0.023

0.005

T6

0.058

0.023

0.005

1B: Residual variance classes constrained

Intercept

0.352

0.026

0.265

0.022

0.369

0.037

Linear Slope

0.078

0.003

0.019

0.001

0.089

0.003

Quadratic Slope

−0.07

0a

−0.033

0a

0.023

0a

Cubic slope

0.012

0a

0.004

0a

− 0.008

0a

T1

0.034

0.034

0.034

 

T2

0.029

0.029

0.029

 

T3

0.028

0.028

0.028

 

T4

0.027

0.027

0.027

 

T5

0.026

0.026

0.026

 

T6

0.005

0.005

0.005

1C: Random effects constrained

Intercept

0.282

0.013

0.466

0.013

0.234

0.013

Linear Slope

0.073

0.001

0.091

0.001

−0.023

0.001

Quadratic Slope

−0.048

0a

−0.033

0a

− 0.025

0a

Cubic slope

0.006

0a

0.004

0a

0.005

0*

T1

0.027

0.067

0.031

T2

0.029

0.061

0.013

 

T3

0.027

0.057

0.011

 

T4

0.028

0.094

0.003

 

T5

0.023

0.09

0.001

 

T6

0.03

0.075

0.004

Model 2A:

Residual variance Constrained time and classes

Intercept

0.302

0.021

0.223

0.006

0.715

0.021

Linear Slope

0.105

0.001

0.043

0.000b

−0.142

0.000b

Quadratic Slope

−0.034

0a

−0.045

0a

−0.029

0a

Cubic slope

0.004

0a

0.006

0a

0.007

0a

T1

0.028

0.028

0.028

T2

0.028

0.028

0.028

 

T3

0.028

0.028

0.028

 

T4

0.028

0.028

0.028

 

T5

0.028

0.028

0.028

 

T6

0.028

0.028

0.028

2B: Random effects and residual variance time constrained

Intercept

0.463

0.01

0.323

0.01

0.148

0.01

Linear Slope

0.094

0.001

0.04

0.001

0.026

0.001

Quadratic Slope

−0.033

0a

−0.044

0a

−0.032

0a

Cubic slope

0.003

0a

0.006

0a

0.005

0a

T1

0.073

0.029

0.006

T2

0.073

0.029

0.006

T3

0.073

0.029

0.006

T4

0.073

0.029

0.006

 

T5

0.073

0.029

0.006

 

T6

0.073

0.029

0.006

2C: Random effects and residual variance classes constrained

Intercept

0.273

0.018

0.491

0.018

0.397

0.018

Linear Slope

0.019

0.001

0.469

0.001

−0.074

0.001

Quadratic Slope

−0.032

0a

−0.261

0a

0.099

0a

Cubic slope

0.005

0a

0.032

0a

−0.015

0a

T1

0.037

0.037

0.037

T2

0.023

0.023

0.023

T3

0.026

0.026

0.026

 

T4

0.033

0.033

0.033

 

T5

0.017

0.017

0.017

 

T6

0.027

0.027

0.027

  1. aThese variances were manually set to be equal to zero
  2. bThese variances were estimated to be very close to zero, which led to a warning that the covariance matrix was not positive definite. Nevertheless, the variances were not set to be equal to zero, in order to keep the model comparable to other models