Using Self-Report Romantic Attachment Measures in Therapeutic Practice.
Richard J. Atkins - revised and updated 30th March 2010
Over the eleven plus years since I began hosting this website I have been contacted a number of times by therapists, often couple therapists, asking which self-report measure of attachment they should use to assess the underlying attachment characteristics of their clients. Based on the comments I have made here, excepting those who are trained in attachment from a clinical perspective, I do not believe attachment concepts or measures are appropriate for use in therapeutic practice. I set out the two main reasons for this below.
What is being Measured?
Self-report attachment measures are relatively easy to administer. Certainly they are less involved than interview or observational measures such as the Strange Situation Procedure or the Adult Attachment Interview and their use and scoring requires little training and can often be inferred from journal articles where the measures were first presented or later used. Also, because they are generally used for research purposes, many of them have been published in academic journals and are available for use without charge. This appears to make them a very attractive option for a therapist who wishes to assess a client's attachment characteristics as it initially looks like it would be an easy thing to do. However there is a problem:
Results from self-report attachment measures describe individuals in terms of their attachment characteristics (secure, avoidant, anxious etc.) and attachment is generally understood as being about persisting trait-like individual differences in relationship-related cognitions (and consequent behaviours) that were formed during the early years and then affected by subsequent experiences. As a result it is easy to slip into the belief that these self-report measures are assessing underlying trait-like attachment characteristics and I have often been approached by therapists who want to use these measures in precisely this way. Unfortunately, the results of extensive research using these self-report measures (and indeed the names of many measures) indicate these measures primarily describe current or most recent relationship functioning. Since there is a correlation between current relationship functioning and underlying attachment characteristics, these measures are appropriate for assessing attachment in a research context such as where you are looking for correlations with other factors using large groups of individuals. In such contexts current relationship function is effectively a confounding influence that would ideally be controlled for but which will (hopefully) have little systematic effect on results other than to increase the residual variance present in measurements. This is why the measures are useful to psychologists engaged in personality and social psychological research who are using relatively large samples of participants.
Unfortunately this does not mean the measures are suitable for clinical use. If they are more closely related to current relationship functioning, then although there is little doubt that current relationship functioning is affected by persisting working models and early experiences, results using these instruments will actually be about the former rather than the latter. Using such a measure to assess an individual's underlying attachment 'style' is completely inappropropriate. For example, it seems likely that an individual whose early experiences led them to a relatively secure attachment 'style' might well appear to be anxoiusly attached if they are experiencing relationship difficulties, as applying these measures DOES NOT give you their underlying attachment characteristics but a reflection of their current attachment within a relationship under threat.
What is being Understood?
On the more general point, I am not entirely convinced that romantic attachment concepts, let alone measures, should be used in therapeutic practice at all. While I acknowledge the excellent reintegration of 'developmental' attachment theory into the psychodynamic therapeutic movement, psychodynamic therapists and counselling psychologists will generally have a thorough and extensive grounding in the psychology behind attachment theory with detailed exposure to Bowlby's theories and Ainsworth's mother-infant research as well as later therapeutic integrations such as Object Relations theory. My concern here is more with the use of attachment theory concepts by counsellors and psychotherapists who have minimal exposure to the underlying theory or whose main exposure is to the personality/social psychology approach.
Unlike systems such as Transactional Analysis with it's ego-state model and 'OK corral,' and systemic theory with it's inclusion/affection/control model, since its terms were first established, attachment theory has developed in such a way that the common meanings of its value-laden terminology are no longer adequate or even appropriate. Furthermore the apparent simplicity of categorical models and early ideas such as individuals having a definate 'attachment style' encourages labelling with these easily misunderstood terms and this leads to the misrepresentation and misuse of attachment concepts. To my dismay (and despite my best efforts) more than once I have been in the refectory after delivering a lecture on romantic attachment and have overheard students using attachment concepts as weapons with which to assassinate the characters of their mutual acquaintances and (more especially) their friends' relationships and partners. It is interesting to note that in every case, the assassinators have claimed the labels 'secure' or 'anxious' for themselves and have damned others with the label 'avoidant'. Adopting the Karpman (1968) drama triangle model, it appears that the self-labelled 'secure' individuals are often acting as rescuers, individuals labelled as 'anxious' (the socially acceptable form of insecurity?) are viewed as the victims of their 'avoidant' persecutors (avoidance being the socially unacceptable form of insecurity). Obviously this is a wildly inaccurate oversimplification of attachment theory and one which facilitates potentially unhelpful interpretations, yet it appears to be an interpretation of attachment 'styles' that is commonly and readily adopted.
Summary
Unlike other psychological and 'pop psychology' models that have genuinely accessible, readily understandable and easily applied terminologies and models, attachment theory has a genuinely misinterpretable and readily misunderstood but unfortunately an easily remembered terminology and it is for this reason I am hesitant about its presentation to clients as an easily grasped model that can be used for personal insight and as a context for understanding their relationships, particularly when working with vulnerable individuals and couples experiencing problems in their relationships. Add to this the fact that self-report attachment measures do not actually measure attachment as it is commonly understood and I think this constitutes a convincing argument that therapists and other practitioners should be strongly discouraged from using these measures and concepts in clinical practice, unless they have the appropriate training to really understand what they are doing.
It may look easy and simple but its not!