Attachment as an Organizational Construct

underlying developmental research are often only implicit, yet they guide data collection and interpretation of results. For example, a number of researchers have provided data concerning intercorrelations among behaviors presumed to be indices of attachment (e.g., Coates, Anderson & Hartup 1972; Maccoby & Feldman 1972). Noting that such "index" behaviors do not intercorrelate highly, do not show temporal stability, and are strongly influenced by context, critics have concluded that the attachment construct itself is wanting, that concepts such as attachment relationship and affective bond are superfluous, and that varying patterns of attachment behavior among infants are of little consequence (Cairns 1972; Gewirtz 1972ax, 1972b); Masters & Wellman 1974; Rosenthal 1973; Weinraub, Brooks, & Lewis 1977) . It has been suggested that individual differences be disregarded (e.g., Masters & Wellman 1974) and that research on attachment be reduced to study of contingencies within the contemporary interaction of care giver-infant dyads (e.g., Cairns 1972; Gewirtz, 1972a,b; Rosenthal 1973). The intercorrelational research and the critiques based upon it reflect a particular view of the attachment con[Child Development, 1977, 48, 1184-1199].