“cannot classify” (CC) or “unresolved” (U) - were also given a forced classification into
one or more of the organised categories, according to the scoring directions. Participants’
and controls’ AAI classifications are reported in Table 1.
-------------
Table 1 about here
AAI classifications can be clustered into two groups (secure vs. insecure or disorganised),
three categories (secure vs. preoccupied vs. dismissing, with disorganised participants
forced into the best-fitting organised category), four categories (secure vs. preoccupied vs.
dismissing vs. disorganised) or each category analysed separately (secure vs. preoccupied
vs. dismissing vs. cannot classify vs. unresolved). Because of the small sample size, we
focus on reporting two- and three-category analyses here. Two categories can be used to
explore the simple question of the rate of attachment security, while using more categories
allows a more detailed exploration of the types of internal working models of attachment
within the sample.
Attachment classifications of participants and controls were compared using two-
tailed chi-square tests. When three categories were used, the difference between
participants and controls approached significance (χ2 = 5.33, df = 2, p = .07). The
difference appears to be due to three controls whose primary classifications are “cannot
classify” or unresolved but whose secondary classifications are secure. The two-category
chi-square, comparing secure to insecure and disorganised classifications, showed no
significant difference between participants and controls (χ2 = 1.29, df = 1, p = .26).
• AAI scale scores
All AAI transcripts were scored on five “probable experience” scales and eight “state of
mind” scales, all of which range from 1 to 9. The “metacognitive monitoring” scale was
14