Background: Vignettes are used by nurse researchers use to determine how clinical judgments about patient care situations are made. However, when vignettes are designed there is often a restriction on the number of characteristics studied, which oversimplifies the richness and complexity of real world healthcare situations.
Objectives: The purpose of this article is to describe a factorial survey. Its multilevel design of independent variables allows for real world complexity in a way not tested by a sample set of four to six identical vignettes. Nurses' judgments about patients' confusion and the application of restraints are used to illustrate the method.
Method: The factorial survey is an experimental design that can be developed in three steps: (a) identifying and using the variables, (b) writing a coherent vignette, and (c) randomly generating the vignettes.
Results: The unit of analysis is the vignette and Ordinary Least Squares (OLS) regression is used for analyses. In the example provided on confusion recognition and restraint use, patient characteristics accounted for the majority of explained variance in confusion recognition of (40%, R2 = 0.40) and restraint intervention for (43%, R2 = 0.43). The results for both models were strikingly similar as the same patient characteristics all were significant predictors for confusion recognition and restraint use.
Conclusions: The versatility of the factorial survey lies in the researcher's ability to use it to test judgments in a variety of complex clinical simulations, to aid in concept development, and to identify consensus and disagreement among nurses. The multilevel design of the independent variables allows for real world complexity in a way not tested by a sample set of four to six identical vignettes.