ABSTRACT
Claims concerning the stability of attachment classifications are not as central to at least some attachment theorists as are assertions about their antecedents or interpretations. To be sure, it has been claimed (e.g., Waters, 1983) that evidence of high stability is of crucial importance to the interpretability of Strange Situation behaviors, since these classifications are presumed to reflect consistent dimensions of mother–infant interaction. Others (e.g., Thompson & Lamb, 1984b; Thompson, Lamb, & Estes, 1982, 1983; Vaughn, Egeland, Sroufe, & Waters, 1979), however, have argued that if the classifications accurately reflect a dyad’s interactive quality, they should change when events or circumstances occur which influence the quality of dyadic interaction. If the security of attachment did not change in the face of stressful events or changes likely to affect the quality of interaction, there would be reason to doubt that the Strange Situation is a sensitive index of the parent–infant relationship. Thus the reliability and validity of the Strange Situation procedure can be demonstrated by both high and low levels of temporal stability, provided that changes in classification status are meaningfully related to the kinds of circumstances that may affect adult–infant interaction. Evidence for such external correlates contributes to the validity of the Strange Situation procedure and the interpretability of changes in attachment status. Interestingly, however, one can observe a change over time in the implicit goal of researchers. Following an initial concern about demonstrating that Strange Situation classifications were, or at least could be, stable over time, most researchers are now concerned about defining when, and under what circumstances, we should expect stability, and when we would predict instability or change.
