Skip to main content

Table 2 Criteria for evaluating adequacy of meeting IRT assumptions and results

From: Development and calibration of a novel social relationship item bank to measure health-related quality of life (HRQoL) in Singapore

Unidimensionality
 Approach Criterion Reference Results Criterion met?
 EFA Percentage of variance accounted for by first factor > 20% PROMIS [16] 18.1% No
Ratio of first to second eigenvalues > 4.0 PROMIS [16] 8.01 Yes
 CFA CFI > 0.95 PROMIS [16] 0.923 No
TLI > 0.95 PROMIS [16] 0.917 No
RMSEA < 0.06 PROMIS [16] 0.098 No
SRMR < 0.08 PROMIS [16] 0.081 No
 Bifactor analyses ARPB < 10% Muthén, Kaplan, and Hollis (1987) [33] 3.6%§ Yes
General ECV > 0.70 Reise, Bonifay and Haviland (2013) [17] 0.805ǂ Yes
OmegaH > 0.80 0.980¥ Yes
General ECV > 0.60 and OmegaH> 0.70 Reise, Schienes, Widaman and Haviland (2013) [15]   Yes
Local Independence
 Residual correlation matrix Average absolute residual correlations < 0.10 PROMIS[16] 0.067 Yes
Percentage of residual correlations above 0.20 Artmann et al. 2010 [18] 1.84% No threshold given
(8 of 435)
  1. Abbreviations: Item response theory (IRT), comparative fit index (CFI), Tucker-Lewis Index (TLI), root mean square error of approximation (RMSEA), standardized root mean residual (SRMR), explained common variance (ECV), item explained common variance (IECV), omega hierachical (omegaH), explanatory factor analysis (EFA), confirmatory factor analysis (CFA)
  2. §Maximum ARPB among three exploratory bifactor models with 2,3 and 4 specific factors. See Table 3
  3. ǂMinimum general factor ECV attained among three exploratory bifactor models with 2, 3 and 4 specific factors. See Table 3
  4. ¥ Minimum OmegaH attained among three exploratory bifactor models. See Table 3