Relatively little is known about the psychometric properties of measures that assess the constructs of caregiver strain and well-being. Using structural equation modeling, the authors validated two scales used in the caregiving literature: Caregiver Well-Being Scale and Caregiver Strain Index. Two models were generated for the Well-Being Scale that demonstrate the measure's validity and reliability. Nineteen of the 45 items on the Well-Being Scale validated the scale. Of the original 13 items on the Strain Index, 11 are valid indicators of their respective factors. Structural equation modeling is a useful technique to evaluate these measures, because it identifies the most reliable and valid items for each scale. Key words: caregivers; psychometric properties; stress; structural equation modeling; well-being Well-being and stress are important constructs in measuring the caregiving experience, because they represent both a strengths-based perspective and a detrimental outcome of care-giving. They present particular measurement challenges to researchers studying caregivers. These constructs are viewed as multifaceted, complex phenomena that incorporate both subjective and objective dimensions (George, 1994). Neither construct can be measured by a single observable variable, but each requires multiple items to assess it. Definitions of the are many and varied. Caregiving stress (strain or burden) is defined as enduring problems that have the potential for arousing threat, a meaning that establishes strain and stressor as interchangeable concepts (Pearlin & Schooler, 1978, p. 3). Well-being is defined as at least some periodic states of security and structure within the turbulence of (Goldstein, 1990, p. 273). Because of the complexity of each construct and the methodology used to evaluate the measures, further research is needed to determine the reliability and validity of measures of stress and well-being. Such techniques as factor analysis, correlation, and coefficient alpha have been used to estimate the validity and reliability. There are several limitations inherent in these techniques (Nunnally & Bernstein, 1994). For example, coefficient alpha assumes that the items are homogeneous (Miller, 1995). If the measure consists of heterogeneous items, use of coefficient alpha is not appropriate. Another limitation of coefficient alpha is that it is affected by the number of items in a measure. The measure becomes more reliable by adding more items (Nunnally & Bernstein, 1994). Furthermore, none of the aforementioned techniques is useful with a multidimensional measure. A more sophisticated technique, structural equation modeling (SEM), overcomes many of these limitations when assessing measures' psychometric properties. The capacity to estimate a model's measurement error makes SEM an alternative for assessing validity and reliability of a measure. Because well-being and stress are multifaceted and require multiple indicators, SEM can model different dimensions of the constructs. Research demonstrates a need for more advanced statistical techniques to assess fully the validity and reliability of existing measures. This article presents the results of an assessment of the psychometric properties of two measures--well-being and stress. Using two separate samples, we used SEM to evaluate the reliability and validity of each measure. SEM is presented as an additional method to evaluate the reliability and validity of caregiver well-being and strain. Earlier research evaluated these measures with factor analysis or coefficient alpha (Robinson, 1983; Tebb, 1995). LITERATURE REVIEW Measuring well-being and stress presents challenges for social scientists. Researchers use well-being to explain life satisfaction, emotional status, affective disorder, and stress (Stetz, 1992). Both constructs are used to study family caregiving, but additional data is needed on the psychometric properties of current instruments. …
Read full abstract