An approach to modifying self-report instruments

The Rasch model, a simple latent trait model was applied to responses to a 21-item self-report instrument. Items were assigned values reflective of interval scale measurement. To modify the instrument, items below the calibrated mean were eliminated. The proposed revision contains fewer items with known scale values and only those items that are the strongest indicants of the variable in question. The present work suggests that the Rasch and similar models are useful for increasing the measurement properties of self-report instruments.