Nurses' attitude, belief, practice, and knowledge regarding urinary incontinence in adults: LISREL analysis of a model

This study sought to discover what nurses think, believe, do, and know about urinary incontinence in adults and to determine inferred causal relationships among these constructs. A nonexperimental descriptive design with causal inference analyzed the measurement and structural relationships of an hypothesized four-scale model and a respecified three-scale model. Prior to analysis, items with item-to-scale correlations below 0.3 and at or above 0.7 were eliminated. Following this revision, four researcher-developed measurement scales were psychometrically tested for reliability and validity. Three of the four scales, the "Urinary Incontinence Attitude Scale" ($\alpha$ =.8685), the "Urinary Incontinence Belief Scale" ($\alpha$ =.8669), and the "Urinary Incontinence Practice Scale" ($\alpha$ =.8965) were judged reliable using Cronbach's coefficient alpha and squared multiple correlation. Alpha coefficient for the "Urinary Incontinence Knowledge Scale" ($\alpha$ =.6312) was lower than the 0.7 recommended by Nunnally (1978) for newly-developed scales. Validity of all measures was determined with confirmatory factor analysis using LISREL 8.12a (Joreskog & Sorbom, 1993) and found adequate under Bollen's (1989) definition of validity: all latent constructs were significantly linked to their hypothesized indicators. The inferred causal relationships were tested using structural equation modeling with LISREL 8.12a. There was a significant structural causally-inferred relationship between attitude of nurses regarding care of adults with urinary incontinence to their self-reported incontinence-related practice. No other significant relationships were found in the structural models. Goodness-of-fit statistics indicated a good fit of both the four-scale and the respecified three-scale models to the data.