Validity of the Neuromuscular Recovery Scale: a measurement model approach.

OBJECTIVE

To determine how well the Neuromuscular Recovery Scale (NRS) items fit the Rasch, 1-parameter, partial-credit measurement model.

DESIGN

Confirmatory factor analysis (CFA) and principal components analysis (PCA) of residuals were used to determine dimensionality. The Rasch, 1-parameter, partial-credit rating scale model was used to determine rating scale structure, person/item fit, point-measure item correlations, item discrimination, and measurement precision.

SETTING

Seven NeuroRecovery Network clinical sites.

PARTICIPANTS

Outpatients (N=188) with spinal cord injury.

INTERVENTIONS: Not applicable.

MAIN OUTCOME MEASURE: NRS.

RESULTS

While the NRS met 1 of 3 CFA criteria, the PCA revealed that the Rasch measurement dimension explained 76.9% of the variance. Ten of 11 items and 91% of the patients fit the Rasch model, with 9 of 11 items showing high discrimination. Sixty-nine percent of the ratings met criteria. The items showed a logical item-difficulty order, with Stand retraining as the easiest item and Walking as the most challenging item. The NRS showed no ceiling or floor effects and separated the sample into almost 5 statistically distinct strata; individuals with an American Spinal Injury Association Impairment Scale (AIS) D classification showed the most ability, and those with an AIS A classification showed the least ability. Items not meeting the rating scale criteria appear to be related to the low frequency counts.

CONCLUSIONS

The NRS met many of the Rasch model criteria for construct validity.