Skip to main content

Table 5 Averages and 95% confidence intervals for correlations between random intercepts and correlation between random slopes based on simulated data, where data were generated under 3 different scenarios (Scenarios A: different proportions of progressive cases; Scenarios B: different magnitudes of deteriorating rates in progressive cases; Scenarios C: various strength of between-error correlations)

From: Estimating correlation between multivariate longitudinal data in the presence of heterogeneity

Simulation Scenarios

Correlation between random intercepts

Correlation between random slopes

True correlation (no heterogeneity)

0.145

0.111

Scenario A:

 5%

0.199 (0.070, 0.329)

0.522 (0.390, 0.654)a

 10%

0.246 (0.118, 0.373)

0.657 (0.562, 0.753)a

 15%

0.268 (0.153, 0.383)a

0.727 (0.651, 0.804)a

 20%

0.296 (0.178, 0.413)a

0.769 (0.698, 0.840)a

Scenario B:

 50%

0.200 (0.070, 0.329)

0.240 (0.051, 0.429)

 75%

0.207 (0.077, 0.336)

0.382 (0.223, 0.542)a

 100%

0.199 (0.070, 0.329)

0.522 (0.390, 0.654)a

 125%

0.205 (0.078, 0.332)

0.634 (0.532, 0.734)a

 150%

0.202 (0.078, 0.327)

0.706 (0.616, 0.795)a

Scenario C:

 0.0

0.199 (0.070, 0.329)

0.522 (0.390, 0.654)a

 0.2

0.232 (0.104, 0.361)

0.603 (0.479, 0.727)a

 0.4

0.273 (0.147, 0.398)a

0.686 (0.574, 0.797)a

 0.6

0.303 (0.180, 0.425)a

0.769 (0.664, 0.873)a

 0.8

0.334 (0.218, 0.450)a

0.853 (0.756, 0.949)a

  1. a95% confidence interval does not contain the true value