Skip to main content

Table 4 Model fit changes for short form selection

From: State of the psychometric methods: patient-reported outcome measure development and refinement using item response theory

ModelCronbach’s alphaAICBIC-2log likelihoodΔ in -2log likelihoodRMSEAM2 (df)
51 items0.98365,230.1866,432.6064,720.180.43163,378.86 (1071)***
Short Form 1 Prioritizing Content (10 items)0.94613,762.9914,234.4013,562.9951,157.190.011469.53 (1420)
Short Form 2 Prioritizing Precision (10 items)0.94513,825.1214,296.5413,625.1251,095.060.011513.90 (1420)
  1. Cronbach’s alpha = measure of internal consistency/reliability from Classical Test Theory (criterion: ≥.90).
  2. AIC Akaike information criterion (criterion: the lower the number, the better the fit)
  3. BIC Bayesian information criterion (criterion: the lower the number, the better the fit)
  4. -2log likelihood = if models are nested, subtract at each step to see if step is significant
  5. RMSEA Root mean square error of approximation (criterion: ≤ .05).
  6. M2 = model fit.
  7. ***p < .001 (Note: a significant value for model fit indicates that the model does NOT fit well)
  8. df Degrees of freedom.
\