Design: Comparison of two versions of the same theoretical model of a healthy nursing practice environment; one using single indicators, one using multiple indicators for each concept.

Method: Structural Equation Modeling using the chi-square test of model fit.

Data: The Canadian Nurse Survey portion of the International Hospital Outcomes Study (n=17,403).

Concepts Studied: Causal variables included familiarity with the practice environment, organizational support for staff development, collaboration with physicians, sufficient staffing resources, professional practice support, and supervisory support. Outcome variables included improved relationship between management and staff, enhanced staff participation in scheduling, perception of increased competence among nursing colleagues, increased nursing autonomy, increased recognition, and ability to spend time with patients.

Findings: The chi-square of the multiple indicator model showed very poor fit with the data (chi-square 25,825, df 270). The chi-square of the single indicator model showed markedly closer fit with the data (chi-square 354, df 24).

Conclusions: Models with multiple indicators may lead to erroneous conclusions. Theoretically, multiple indicators actually measure different concepts and reduce the measurement parsimony required for well-fitting models. Measurement implications include an increase in statistical coordination that is require between the causal variable and its multiple indicators. This is sufficient to drastically reduce model fit.

Back to Posters
Back to Evidence-Based Nursing: Strategies for Improving Practice
Sigma Theta Tau International
July 21, 2004