Skip to main content

Table 2 Model fit indices and comparison of LCS models for total weekly alcohol consumption and mental health in the Whitehall II study

From: An exploration of the dynamic longitudinal relationship between mental health and alcohol consumption: a prospective cohort study

 

Baseline

Alcohol → ΔMCS

MCS → Δalcohol

Reciprocal

Age and sex adjusted

 Fit statistics

 Log likelihood

-146158.161

-146158.041

-146155.151

-146154.760

 χ2 (df)

274.233 (12)

273.995 (11)

268.214 (11)

267.432 (10)

 RMSEA

0.059

0.061

0.061

0.064

 AIC

292380.321

292382.083

292376.302

292377.520

 SSA BIC

292494.731

292500.068

292494.288

292499.081

 CFI

0.982

0.982

0.982

0.982

 TLI

0.959

0.955

0.956

0.952

 Model comparison (difference in χ2 fit (df))

 Versus baseline

–

0.238 (1), P = 0.63

6.019 (1), P = 0.01

6.801 (2), P = 0.03

 Versus previous best

–

–

–

0.782 (1), P = 0.38

Fully adjusted

 Fit statistics

 Log likelihood

-243989.314

-243989.108

-243985.355

-243984.851

 χ2 (df)

328.239 (54)

327.827 (53)

320.320 (53)

319.312 (52)

 RMSEA

0.028

0.029

0.028

0.028

 AIC

488798.629

488800.217

488792.710

488793.702

 SSA BIC

490264.509

490269.672

490262.165

490266.732

 CFI

0.983

0.983

0.984

0.984

 TLI

0.952

0.951

0.953

0.952

 Model comparison (difference in χ2 fit (df))

 Versus baseline

–

0.412 (1), P = 0.52

7.919 (1), P < 0.01

8.927 (2), P < 0.01

 Versus previous best

–

–

–

1.008 (1), P = 0.32

  1. AIC, Akaike information criterion; CFI, comparative fit index; df, degrees of freedom; LCS, latent change score; MCS, mental health component score; RMSEA, root mean square error of approximation; SSA BIC, sample size adjusted Bayesian information criterion; TLI, Tucker-Lewis index.