The Good News
Many informed, thoughtful people have worked hard to develop placement guidelines. While their can be no definitive answers, certain principles can be stated. For young infants it is best for there to be a primary home, with away from home overnights not occurring or quite restricted for the first 18 months. Visitations by the non-custodial parent can and should be frequent, all other things considered. Both attachment relationships can thereby be supported, while the child’s task of organizing attachment behavior is simplified. For infants, alternating nights in the two homes, which has been advocated by some, is definitely not advised. With increasing age, overnights and one and then more nights can gradually occur. Gradual transitions are also recommended when a child is to be moved from one placement to another permanent placement.

The Bad News
There is no single answer as to which parent should be primary. The infant will have a primary attachment to whoever has be the most consistently interactive and emotionally involved with them. There is no test, including the Strange Situation, that can provide definitive answers to this question. The reason is that the Strange Situation would be impacted by lengthy separations. Therefore, if one parent has not been with the child for a period of weeks, the child may behave in an anxious way, even if they had previously had a secure attachment with this parent. Further, the procedures would have to be separated in time so that the infant no longer remembered the previous testing. It also would be normal for an infant to be unsettled following a visit with the non-custodial parent, because it has experienced separations from one, then the other parent. This would not necessarily indicate a problem in the other parent-infant relationship. Excellent information can be found about the complexities of these issues, as well as useful ideas regarding custody and placement in a Special Issue of the journal, Family Court Review, 2011, July.


The Strange Situation, conducted as Ainsworth did it, derives its validity from its links to actual attachment/exploration and crying in the home, based on many hours of observation of each child. This is the absolute criterion. Any other measure that claims to be a measure of attachment must be shown to relate to actually observed attachment behavior. This even applies to variations in the Strange Situation (for example, if one used only a single brief separation). Validity would have to be established for this or any other laboratory measure by linking it to home observation.

To date, there is only one alternative, the Attachment Q-Sort, developed by Everett Waters. This procedure in fact involves extensive observation before summarizing the infant’s attachment behavior by sorting a set of cards according to how characteristic or uncharacteristic they are of this relationship. This sorting is then compared to a prototype secure relationship pattern. While for group studies parents may be used as reporters, to assess an individual case with confidence observers are used. Since this procedure involves actually viewing secure-base behavior in the natural environment, it can directly claim to be an attachment measure. It also does correlate with Strange Situation assessments. Moreover, unlike the Strange Situation, which is primarily validated up to age 18 months, the Q-sort can be used until age 4 ½. For complete information about the Q-sort go to JohnBowlby.com.

Finally, Susan Berger at Northwestern (sberger@northwestern.edu) has developed a set of behavioral indicators of attachment based on observations in the pediatrician’s office. She is continuing to validate these procedures.


Basically, no. There are some promising leads. There have been measures based on drawings, story-stem endings and other narratives that have been show to be related to infant strange situation assessments, and this grants them some validity. These measures are interesting and have a great deal of face validity as well(see Bretherton ; Fury et al., 1997 in our references). However, one cannot know for sure whether they are measures of attachment or simply correlates of attachment. Infant attachment assessments relate to many outcomes, for example, math achievement in high school. No one would claim that math achievement is therefore a measure of attachment. One would not want to go into court claiming a math achievement score confirms a secure attachment, nor could you do it with a child’s drawings. It is a hard problem to directly access attachment exploration/balance in middle childhood.


Attachment refers to the relationship between infant and caregiver or between to people in a longstanding, vital relationship. Like any important relationship it develops over considerable time; it cannot be instantaneous. In infants, it is not consolidated until the second-half year of life. Bonding technically refers to the mother’s tie to the child and is argued to often occur in the early days or even hours of life. For these reasons, attachment researchers usually do not use the word “bonding”, even though attachment does entail an emotional bond between the two partners.


It is definitely established that the quality of an attachment relationship can change notably, given notable life changes. This has been shown in infancy. For example, when the life stress of parents declines between 12 and 18 months, infants who were anxiously attached are more likely to now be securely attached than is the case than in families where stress remained elevated notably, given notable life changes. This has been shown in infancy. For example, when the life stress of parents declines between 12 and 18 (Vaughn, Waters, Egeland, & Sroufe, 1979). More generally, anxious attachment histories do not always lead to problematic outcomes and secure attachments do not always forecast development free of problems. Subsequent experience, especially in important relationships can alter a developmental course. Change has even been shown in adulthood. For example, adults who were non-autonomous on the AAI at 19 years are more likely to be autonomous at age 26 years if they have formed a stable partnership in the
intervening period (Van Ryzin, M.J., Carlson, E.A., & Sroufe, L.A. (2012).


The term “attachment style” comes from the adult social-psychological literature and refers to responses to paragraph descriptions regarding adult relationship preferences or to brief questionnaires. These instruments have face validity but have not shown strong ties to any measures of attachment or loss, although they do show correlates with personality questionnaires. Attachment pattern refers, first, to the Ainsworth infant categories and, subsequently, to the adult states of mind regarding attachment derived from the AAI. Both of these, as well as the Attachment Q-sort have been amply validated vis a vis other measures of attachment and attachment behavior. Generally, those working within the Bowlby/Ainsworth attachment tradition do not use the term “attachment style.”