The AAI classification system assigns a rating of unresolved (U) to an interview on the basis of that particular portion of the interview pertaining to the discussion of loss or traumatic events. However, those individuals assigned the (U) category also display an underlying secondary attachment classification of the existing types. Thus it is common to split the analysis of attachment data into examination of the traditional three categories, and a four-category system involving the (U) category. Including the (U) category in the analysis, 22% (n=9) of the sample were dismissing (D), 10% (n=4) secure-autonomous (F), 7% (n=3) preoccupied (E), and 51% (n=21) unresolved (U). In terms of the traditional three category system the distributions are slightly different, with 12% (n=5) secure-autonomous, 51% (n=21) dismissing, and 27% (n= ll ) preoccupied. In both these classifications four subjects
w ere assig n ed the C an n o t C lassify c a te g o ry , but becau se o f th eir sm all num ber, w ere ex c lu d e d from fu rth e r analysis. F igure 1 show s graphically the distrib u tio n of a tta c h m e n t p attern s in the sam ple, again on the basis o f the trad itio n al th re e w ay split, the m o re re c e n t four-w ay split show n in F igure 2 (these c h a rts include the c a n n o t classify ca te g o ry ). T h e high rates o f attach m en t insecurity in the sam ple are in line w ith p re v io u s studies both o f adolescents and adults (R o sen stein & H o ro w itz , 1996; D o z ie r, 1990). G iven the relatively small sam ple size in th e p re se n t study h o w ev er, fu rth e r analyses w ere carried o u t w ith re sp ect to a nu m b er o f tw o -w a y splits in the d ata, including dism issing vs. p re o ccu p ied , resolved vs. u n re so lv e d , secu re vs. insecure.
F ig u re 1. Illu strates the distribution o f attach m en t pattern s in the trad itio n al three- w ay split (including the cannot classify category)
cannot classify 10% insecure-preoccupied 2 7 % secure-autonomous 12% in secu re-dismissing 5 1 %
F ig u re 2. P ie ch art illustrating the distribution o f A dult A ttach m e n t C lassificatio n s in the sam p le w hen the U nreso lv ed category is included.
secure 10% unresolved 5 1 % ■ ^ dismissing 22% preoccupied 7 % cannot classify 10%
A ttachm ent Insecurity - Dismissing and Preoccupied Styles
T h e initial analysis o f d istributions o f attac h m e n t classification w as ca rried o u t by w ay o f chi-square tests, exam ining dism issing and p re o ccu p ied classifications in term s o f age, g ender, and the presen ce o f sexual abuse. H o w e v e r the resu lts o f these analyses d id not indicate any differences betw een the tw o classification o f attac h m e n t insecurity in term s o f age = 0 .2 6 , ns), g en d er = 0 .0 0 6 , ns), o r th e p re sen ce o f o th erw ise o f sexual abuse {'i = 0.03, ns).
A ttachm ent and Resolution of loss, abuse and trau m a
T o look at the w ay in w hich the categ o ry o f u n re so lv e d atta c h m e n t statu s, w hen included in the sam ple, relates to o th e r variables in the study a n u m b er o f chi- sq u are analyses w ere carried o u t com paring type o f attac h m e n t insecurity, diagnostic c a te g o ry , reflective self-functioning, abuse/n o n -ab u se, g en d e r, and age. T h e only significant difference w ith re sp ect to the u n resolved c a te g o ry w as ev id en t b etw een th o se individuals p resenting w ith high and low levels o f R eflective S elf-functioning
(RSF), those individuals with high RSF more likely to be resolved regarding traumatic experiences =5.53, df 1, p<0.05). Attachment status appeared to have no significant impact overall on whether individuals were able to resolve trauma or loss.
As outlined in the methods section, assignment of the unresolved category depends on ratings given to three scales concerning resolution of loss, physical and sexual abuse. Table 2 gives demographic information on the distribution of individuals actually reporting experiences of loss, and/or physical and sexual abuse, as well as numbers of participants displaying lack of resolution regarding these experiences.
Table 2. Summarises demographic information regarding those participants reporting
Type of Trauma/Abuse No. of participants reporting experiences No. of participants receiving unresolved classification Loss 13 3 Physical Abuse 2 1 Sexual Abuse/Loss 4 2 Physical Abuse/Loss 6 6 Sexual/Physical Abuse 2 2 Loss/Sexual/Physical Abuse 7 6 Total 34 22
Note : Lack of resolution is rated on a nine-point scale - five is the cut-off point, above which an individual is considered to be unresolved with respect to the particular trauma.
Table 2 indicates that 34 out of the 41 participants had experienced some form of childhood loss or abuse, either alone or in combination with another form of trauma. No participant reported sexual abuse in isolation. The table demonstrates that for those experiencing loss (n=13) alone, only 3 received the classification of unresolved. This indicates that these three individuals scored above the cut-off point of 5 on the resolution of loss rating scale. By contrast if this loss was accompanied by physical abuse the table suggests that these participants would be far less likely to resolve the combination of trauma, those reporting (n=6), those receiving unresolved classification (n=6). Numbers were too small to fully examine statistically whether
this effect was due alone to the presence of physical abuse, or whether it is dependent on the experience of loss in combination.
Further analysis of these sub-scales examined the effects of attachment patterns on resolution of loss, sexual abuse, and physical abuse. For this analysis each type of trauma was considered in isolation, in contrast to considering the overall unresolved category. Chi-square analyses were carried out to look at whether there were differences between secure vs. insecure, and those rated as dismissing vs. preoccupied individuals in their capacity to resolve loss, sexual, or physical abuse. The only significant result of these analyses indicates that those individuals rated as insecure-dismissing are more able to resolve loss than those who present preoccupied discourse = 10.4, df 1, p<0.001). No differences were observed between the capacity of secure vs. insecure individuals to resolve the various types of trauma or loss.