Caregiving is stressful and compromises the psychological health of caregivers. Data from a sample of 187 African American (AA) and 247 European American (EA) caregivers from the second Family Relationships in Late Life Project were used to examine variables associated with caregiver depressive symptoms. First, however, to ensure that the instruments used in this study were equally valid for all research participants, confirmatory factor analyses were conducted to determine measurement equivalence of the Care Recipient Controlling and Manipulative Behavior (CRCMB) Scale, Caregiver Resentment Scale (CRS), Mutual Communal Behavior Scale (MCBS), and the Center for Epidemiologic Studies Depression Scale (CES-D). Results suggested that, for each scale, the number of factors and their loading patterns were the same for each group. Thus, AA and EA caregivers conceptualized CRCMB, caregiver resentment, mutual communal behavior, and caregiver depressive symptoms in the same ways. In other words, each construct had the same meaning for both groups. In addition, all four instruments had equivalent factor loadings between the groups, suggesting that AA and EA caregivers perceived and interpreted scale items in the same ways. Last, the CRCMB scale, the CRS, and MCBS had full scalar equivalence (i.e., equivalent intercepts) whereas the CES-D had partial scalar equivalence. Put another way, the scores on the scales had the same meaning between groups. Overall, the reliability coefficients, goodness-of-fit indices, and equivalency outcomes suggested that the instruments were acceptable to make comparisons between AA and EA caregivers in this sample. In primary analyses, moderated mediation was used to investigate whether (a) resentment mediated the relation between CRCMB and caregiver depressive symptoms and, (b) mutual communal behavior moderated the mediated effect (i.e., the mediation effect would vary according to the pre-illness communal strength of the caregiver-care recipient relationship). Since AA caregivers were hypothesized to have stronger communal relationships than EA caregivers, the moderated mediated effect was expected to vary according to race of the caregiver. The model was tested in the total sample, the AA sample, and the EA sample.As expected, resentment mediated the relation between CRCMB and caregiver depressive symptoms in the total sample, the AA sample, and the EA sample. That is, for AA and EA caregivers, CRCMB predicted more caregiver depression largely to the extent that more CRCMB led to more caregiver resentment. There was no association between race and mutual communal behavior. Mutual communal behavior did not moderate the mediated effect in the total sample, the AA sample, or the EA sample. In other words, the mediated effect did not vary according to the strength of the pre-illness communal relationship between caregiver and care recipient, regardless of race.
