ABSTRACT
The 15-item version of the Geriatric Depression Scale (GDS-15) is a self-report screening instrument widely used. The current study aimed at providing evidence of the measurement precision of the GDS-15 applying Item Response Theory (IRT). The relative contribution of each item and the reliability of the whole scale in measuring the trait level around the cutoffs were investigated employing data collected from a sample consisting of 1344 old people (M=76.44years, SD=5.13; 58.7% women) involved in a large gerontological research project. The unidimensional two-parameter (2PL) logistic model was employed to item estimate location and slope parameters as well as the Test Information Function (TIF). Nine out of fifteen items were located around the cutoffs and the slopes showed that the majority of the items had substantial discrimination ability. The TIF peaked in correspondence of the cutoffs attesting the good local reliability of the scale. These findings support to the utility of the GDS-15 in detecting depression among older people.
