Long-term objectives are to understand and facilitate the making of permanent residential plans by elderly parents for their mentally retarded (MR) adult offsprings to live with them. Unless plans are made, parents' incapacities or death can lead to inappropriate placements. The specific aim is to test a model of antecedents to making plans using structural equation modeling. The variables hypothesized to contribute directly to plans are relatives willing to assume responsibility; non-use of avoidance coping with planning; and perceived inability to provide homecare. In turn, it is assumed that perceived inability to care is influenced by offsprings' impairment level and adverse changes (in the parent, in the offspring, and other adversities such as illness or death of a spouse); but attenuated by use of informal supports and psychological interdependence between parent and child. Avoidance coping is assumed to be increased by pessimistic attitudes toward offsprings' future and decreased by satisfaction with the use of formal services, as well as by a perceived internal locus of control. In turn, an external locus of control is presumed to be associated with lower SES. Whereas age is associated with adverse changes, it is also inversely related to the use of formal services because, among older cohorts, formal service usage was equated with institutionalization. Lower SES is also assumed to be associated with a greater incidence of adverse changes. The sample of 300 parents (age 65 and over), each from a different family, will include 100 who have not used MR services (the yet to be studied "hidden"). The 1 hour interview consists of a mix of fixed response items and open-ended queries. The model will be evaluated using a structural equation approach based on LISREL VII.In case of a poor model fit, LISREL diagnostics will be evaluated to help isolate the source of model discrepancies. Changes in the model will be made if they can be justified on conceptual and psychometric grounds, and the model will then be reevaluated. The model will also be tested after removing the subgroup who have family members who are perceived as willing to provide home care (an estimated 20%); and also, hypothesized differences between the non-hidden and hidden subgroups will be tested using the multi-group solution in LISREL. The availability of this sample permits further investigation of antecedents and, additionally, assessment of beneficial effects that follow from the sense of relief in having made plans.