ABSTRACT
As the preceding chapters suggest, Ainsworth’s classification system has provided a foundation for a large body of research, in all of which it has been assumed that there exist clear differences among groups of infants defined by the A, B, and C categories. The A–B–C classification system has recently been questioned on several grounds (Campos, Barrett, Lamb, Goldsmith, & Stenberg, 1983; Connell & Goldsmith, 1982; Gardner & Thompson, 1983; Lamb et al., 1984), however. First, it has been argued that “security of attachment” may not be the only factor influencing infant behavior in the Strange Situation. Because other aspects of socioemotional functioning, such as temperamental differences (Goldsmith & Campos, 1982) may be influential, exclusive focus on the A, B, C categories, which are interpreted as measuring only security of attachment, may obscure other important contributions to Strange Situation behavior. Second, individual differences in Strange Situation behavior may be better represented by variations along continuous dimensions rather than the discrete categorical differences presupposed by the A, B, C group typology. This possibility is suggested in part by what seem to be continuities underlying the definitions of the subgroups: for example, the distal to proximal interactive style continuum along which the B subgroups can be arrayed, and the possible “borderline secure” status of the B1 and B4 subgroups proposed by some (e.g., Goldberg et al., 1984; Hazen & Durrett, 1982; van IJzendoorn et al., 1983). Third, it has been argued that there is little a priori reason to expect that individual differences in Strange Situation behavior will be discrete rather than continuous, since natural selection is more likely to have fostered behavioral continua rather than discrete types of behavior (Lamb et al., 1984 and Chapter 4). Finally, if a valid multidimensional set of continuous measures could be developed to represent individual differences in Strange Situation behavior, it would be of substantial utility in research on early socioemotional development. Continuous variables have many desirable statistical properties not shared by categorical variables. For example, they are likely to produce higher predictive validity coefficients. In addition, continuous variables are far more tractable in the structural equation modeling techniques that are essential for addressing the complex longitudinal questions currently needed in Strange Situation research (see Chapters 7, 10, and 14).
