Issues regarding aggregation of data in nursing systems research.

The unit of analysis at which variables are conceptualized and measured is an important consideration when conducting patient outcomes research. Outcomes research ranges from study of individual responses to nursing intervention to investigation of group level outcomes associated with the contexts and processes through which nursing care is delivered. Because many variables salient to the investigation of group level patient outcomes are not amenable to direct measurement, aggregation procedures are used to generate proxy measurement of organizational context or process variables. In this article, key conceptual and methodologic issues encountered when using data aggregation procedures are reviewed. Following this, estimation of aggregated data reliability is illustrated using data from a study conducted to investigate skilled nursing unit structures.