Models, measures, and methods: variability in aging research.

The purpose of this paper is to review the models and measurement strategies used in studies evaluating the predictors of nursing home placement, hospitalization, functional impairment and mortality. To do so we examine 167 multivariate equations abstracted from 78 longitudinal studies published between 1985 and 1998 that assess the risk factors of one or more adverse outcomes. We find that both comparatively straightforward concepts such as age and income and widely used scales such as activities of daily living and the short-portable mental status questionnaire display considerable variability in operationalization and coding. We also find that few researchers employ explicit conceptual models to assist with variable choice, while some predictors-demographics, physical and cognitive functioning-were studied much more frequently than others-service, market, and policy characteristics. Variability in measurement highlights the lack of standardization in this area of aging research and leaves room for improvements in validity and reliability. Limited use of conceptual models has led researchers to include some predictors in their analyses to the exclusion of others.