Skip to main content

Table 2 Results data example. Parameter estimates (posterior means) of the multilevel model for the example ‘Trait Anxiety’ data with a random intercept using the IRT-based plausible values technique and the CTT-based scores as outcome variables

From: Why item response theory should be used for longitudinal questionnaire data analysis in medical research

 

IRTa

CTTb

 

Meanc

SD

Meanc

SD

Fixed effect

    

γ Intercept

0.018

0.043

0.015

0.045

Random Effects

    

Between individual (level-2)

    

τ2 Intercept

0.733

0.067

0.685

0.074

Within individual (level-1)

    

σ2 Residual variance

0.294

0.030

0.357

0.042

Intra Class Correlation

    

ρ

0.287

 

0.343

 
  1. aItem response theory based estimates
  2. bClassical test theory based estimates using sum-scores
  3. cMean of the coefficients resulting from fitting the structural model (longitudinal multilevel model) to the five draws of plausible values based on the IRT or CTT measurement models