Updated: 19 December 2016
Over 60 years after its prototype was first deployed and in spite of a welter of criticisms – especially from cross-cultural research – the Strange Situation remains the most popular and most used measure of children’s attachment. Just exactly what the procedure measures and how successful it actually is have been contested by several prominent researchers and theoreticians and a number of limitations have been acknowledged over the years.
Ironically, considering the issues raised by some cross-cultural research, the idea for the procedure came from work in Uganda by Mary Ainsworth. She had worked for a period with John Bowlby in the UK and been much influenced by Bowlby researcher John Robertson’s meticulous attention to detail in recording naturalistic observations, particularly to do with separation. In 1954 Ainsworth went to Uganda, as a result of her husband getting a research position there. She studied mother-child relationships in 6 villages of the Ganda people in Kampala, visiting 26 mothers and their infants, every 2 weeks for 2 hours per visit over a period of up to 9 months. Visits (with an interpreter) took place in the family living room, where Ganda women generally entertain in the afternoon. She was particularly interested in determining the onset of proximity-promoting signals and behaviours, noting carefully when these signals and behaviours became preferentially directed toward the mother. From this Ainsworth (1967) identified 3 types of relationships:-
- ‘Securely attached’: children were generally contented and pacified by the presence of their mother, using her as a secure base for exploring
- ‘Insecurely attached’: children were less inclined to explore and cried frequently, even when with the mother
- ‘Not yet attached’: children were indifferent to the presence of the mother
These categories were a further development of a classification system she and Bowlby (Bowlby et al, 1956) had used with long-stay children in a British TB sanitorium. Ainsworth also found that securely attached children had mothers who enjoyed breastfeeding and were responsive to their children’s needs (caregiver sensitivity).
Back in the United States several years later, Ainsworth conducted a similar study of 26 families in Baltimore. She regularly visited the mothers and their children for upto 4 hours at a time, interviewing them and making detailed notes from her observations.
From these 2 projects Ainsworth (1969) concluded that there are 2 distinctive features of attachment, both of which have adaptive value. Firstly, infants seek to be close to the mother – proximity – especially when they feel threatened by something. (A survival strategy of the BEIGE vMEME.) Secondly, a secure attachment enables the infant to explore – essential for cognitive development – using the mother/attachment figure as a secure base from which to explore and return to.
Mary Ainsworth & Barbara Wittig (1969) devised the Strange Situation controlled observation procedure from the Baltimore studies, to study formally the reactions of young children to brief separations from their mother to determine the nature of attachment behaviours. In particular they were interested in differences between secure and insecure attachments. They hoped that the Strange Situation would prove a reliable and valid measure of attachments . It first major use was by Mary Ainsworth & Sylvia Bell in 1970 with 100 middle-class infants aged between 12 and 18 months.
Strange Situation procedure
The full procedure consists of 8 stages:-
- The parent and their infant are introduced to the experimental room, containing toys in some parts of it – a novel environment: a 9 x 9-foot square room, marked off into 16 squares to facilitate the hidden observer in recording the infant’s movements.
- The parent and the infant are alone. The parent, sitting quietly in a chair, does not participate while the infant explores.
- The stranger enters, converses with the parent and then gradually approaches the infant with a toy. The parent leaves inconspicuously.
- First separation episode: the stranger’s behaviour is geared to that of the infant – leaving the child playing with the toys unless they are inactive, in which case the stranger tries to interest the infant in the toys.
- First reunion episode: the parent returns and waits for the infant to respond. The stranger leaves inconspicuously as parent comforts infant. Once the infant is settled, the parent then leaves again
- Second separation episode: the infant is alone.
- Continuation of second separation episode: the stranger enters and gears behaviour to that of the infant.
- Second reunion episode: the parent enters, waits for the infant to respond and then picks up the child; the stranger leaves inconspicuously.
The video above, created by students of PSY201 class at Aiken Technical College, gives some flavour of how the procedure works.
Clearly there are ethical issues with the Strange Situation as infants are caused anxiety – sometimes severe anxiety! If the child becomes very distressed, the mother returns earlier than planned. The procedure is then repeated with a further ‘stranger’ episode.
The child’s behaviours are recorded every 15 seconds throughout the sequence of events and are also rated for intensity on a scale of 1-7.
Over the years both Ainsworth and others have introduced minor variations into the procedure. It has also been used with children up to the age of 6.
There are 4 measures of attachment used in the Strange Situation:-
- Exploration and proximity seeking: does the child explore the room? If so, does the child keep looking back at the parent as they explore the room or perhaps going up to the parent and touching them before going back to exploring the room?
- Separation protest: does the child react and show upset when the parent leaves the room? If so, how distressed does the child become?
- Reunion behaviour: how does the child react when the parent returns into the room?
- Stranger anxiety: does the child seem anxious when the stranger is in the room? If so, to what degree and does the child seek the protection of the parent?
Categories and their patterns
Ainsworth & Bell, building on Ainsworth’s earlier work, classified the infants in their sample into 3 types:-
- Type B: secure
The infant is distressed by the mother/caregiver’s absence but rapidly returns to a state of contentment after her return. The infant is wary of the stranger, especially when the mother/caregiver is absent. The mother/caregiver shows sensitivity to the child and the child shows a willingness to explore.
- Type C: insecure-resistant
The infant is insecure n the presence of the mother/caregiver, showing little inclination to explore and becomes very distressed when she leaves. The infant resists contact when the mother/caregiver returns and may express anger/resentment – shouting/screaming and even hitting. The infant is highly wary of the stranger.
- Type A: insecure-avoidant
The infant does not seek contact with the mother/caregiver, shows little distress when separated and avoids contact with the mother/caregiver upon her return. The infant treats the stranger in a similar way to the mother/caregiver, often avoiding them. The mother/caregiver may ignore the infant. The infant usually shows a good willingness to explore.
Both insecure types show characteristics of the RED vMEME emerging in an unhealthy state, due to the PURPLE vMEME failing to develop a secure attachment to the caregiver.
Generally speaking, Ainsworth & Bell found that the infants explored the room most enthusiastically when just the mother was present than either when the stranger entered or when the mother was absent. Additionally, they found the type of attachment between a mother/caregiver and child seems related to the mother/caregiver’s sensitivity and responsiveness to the child. (This links to Carl Rogers’ 1961 work on conditional and unconditional love and the way this affects the development of the selfplex.) They concluded the Strange Situation is a reliable procedure for measuring differences in attachment.
Ainsworth’s procedure has beehttp://selfplexn replicated many times to determine whether:-
- The classification shows good reliability over time with the same mother-child pairs
- The category percentages are similar across cultures
Although Ainsworth’s own follow-up study (1973) failed to show good test/re-test reliability, Toni Antonucci & Mary Levitt (1984) – and other studies – found strong consistency between classifications at 7 and 13 months. Mary Main, Nancy Kaplan & Jude Cassidy (1985) assessed infants in the Strange Situation before 18 months and with both mothers and fathers, and then retested them at the age of 6. They found that 100% of the secure infants were still secure and 75% of the insecure-avoidant were still insecure-avoidant. Ainsworth et al (1978) carried out a meta-analysis of the findings of 106 Strange Situation studies on middle-class children and found almost total support for Ainsworth & Bell’s original study. Everett Waters (1978) reported that, in one study, only 2 out of 50 infants, changed their attachment type between 12 and 18 months. A study in southern Germany by Ulrike Wartner et al (1994) found 78% of the children were classified in the same way at the ages of 1 and 6. Edward Melhuish (1993) has suggested that, where variations occur, these are often associated with changes in the form of care – eg: parents separating. This clearly implies that some kind of disruption to the PURPLE vMEME’s need to find safety in belonging will impact upon the child’s strength and type of attachment. See Separation, Deprivation & Privation for more on this.
The graphic above shows the percentage of the 3 categories Ainsworth et al’s meta-analysis found and the relationship of the categories to caregiver sensitivity and Bowlby’s (1953) concept of the inner working model – a key structure in the selfplex and the basis of the Continuity Hypothesis.
Cross-cultural findings and concerns
By and large the Strange Situation has been used successfully in a number of different cultures – though the research has produced some interesting variations on what might be termed the ‘Ainsworth standard’- represented in the graphic above. Some of that research also raises issues around the validity of the procedure, in terms of just what it is that is being measured.
In 1988 Marinus Van IJzendoorn & Pieter Kroonenberg published a key cross-cultural meta-analysis of 32 Strange Situation studies. Research from 8 different nations was compared, which included Western cultures (eg: US, Great Britain, Germany) and non-western cultures (eg: Japan, China, Israel). The cultural spread of studies used and the researchers’ findings are shown in the table left. Considerable consistency in the overall distribution of attachment types was found across all cultures. Secure attachment (Type B) was the most common type of attachment in all 8 nations. However, significant differences were found between the distributions of insecure attachments. For example, in Western cultures the dominant insecure type is insecure-avoidant (Type A), whereas in non-Western cultures it is insecure-resistant (Type C), with China being the only exception, as insecure-avoidant and insecure-resistant were distributed equally. One of the most significant findings was that there is 1.5 times greater variation within cultures than between cultures. Eg: the 2 Japanese studies. One found no insecure-avoidant but a high proportion of insecure-resistant. The other found a pattern much more like the Ainsworth standard.
The overall consistency in attachment types leads to the conclusion that there may be universal characteristics that underpin infant-caregiver interactions. Van Ijzendoorn & Kroonenberg suggested this might be explained by the effects of mass media (eg: TV and books) which spread ideas (memes) about parenting so that children all over the world are exposed to similar influences. However, where significant variations occur, this indicates that universality is limited.
Van IJzendoorn & Kroonenberg’s study has several strengths, not least that studies were excluded if they looked at special groups such as Down’s Syndrome or twins, or they involved fewer than 35 infants. In all 2,000 Strange Situation classifications were examined. However, there was a major discrepancy in the number of studies from different cultures – eg: 18 from the USA and one each from Britain, Sweden and China – and there were far more studies from the West than the East. This potential culture bias and the differences in the sample groups – eg: some were from the middle class, some from the working class – may provide an explanation of the large within-
Some cross-cultural studies have shown more significant variations from the category percentages found by Ainsworth. Most notably Klaus Grossmann et al (1981) found that the majority of German infants demonstrated insecure-avoidant attachments – 49% insecure-avoidant, 33% secure and 18% insecure-resistant. Abraham Sagi, Marinus Van IJzendoorn & Nina Koren-Karie came up with similar findings 10 years later, with 40% of German infants securely attached, 49% insecure-avoidant and 11% insecure-resistant. As a potential explanation, the researchers suggest that German culture requires some distance between parents and children – “the ideal is an independent, non-clinging infant who does not make demands on the parents but rather unquestioningly obeys their commands.”
Sagi, Van IJzendoorn & Koren-Karie were both conducting their own research and used a meta-analysis to create a 4-nation comparision. The results are graphed below.
The Japanese findings – taken from Kazuo Miyake, S J Chen & Joseph Campos (1985) – showed 68% secure and 32% insecure-resistant, with no insecure-avoidant at all in their samples. Keiko Takahashi (1990) also found a significantly higher than average percentage of insecure-resistant types in Japan. She suggested this might be attributed to the excessive stress Japanese infants might experience during the Strange Situation separation, as infant-mother separation is not the norm in the Japanese culture. However, as Mary Ellen Durrett, Midori Otaki & Phyllis Richards (1984) had pointed out earlier, in modern Japanese families mothers do go out to work and leave their children; and attachment types in such families tended to follow the Ainsworth standard.
The Israeli findings – from then-unpublished work by Abraham Sagi et al (1994) – were rather different: 62% secure, 33% insecure-resistant and only 5% insecure-avoidant. The Israeli sample, though, was taken from a kibbutz (communal farm) where the infants were looked after much of the time by adults who were not part of their family (‘metapelets’). As the children still had close relationships with their mothers, they tended not to be insecure-avoidant. They rarely encountered complete strangers in real life which might help explain the high numbers of insecure-resistant. Another reason put forward was that mothers were often absent; while the caregivers rotated shifts and could not always give prompt attention to individual children. (These findings very much replicated Nathan Fox’s 1977 study into infant attachments on kibbutzim.) Sagi, IJzendoorn & Koren-Karie compared kibbutz children who experienced family sleeping arrangements with those who experienced communal sleeping arrangements and found that the children who slept with their family showed the more ‘normal’ attachment patterns.
Only the American results were similar to Ainsworth’s: 71% showing secure attachment, 12% insecure-resistant and 17% insecure-avoidant.
A more recent study by Mary McMahan True, Lelia Pisani & Fadimata Oumar (2001) of the Dogon in West Africa found a complete absence of insecure-avoidant. This was atrributed to the community’s infant care practices which involve responsiveness, constant closeness to mothers and immediate nursing in response to signs of stress.
In 2005 Jin Mi Kyoung & Mee Sook Yoo compared 87 Korean families with 113 American families, using the Strange Situation. Although there were some notable differences – eg: Korean infants stayed less close to their mothers and explored and, when the mothers returned, they were more likely to get down on the floor and play with their infants – there were a similar proportion of securely-attached children in both cultures.
Implications of these significant cross-cultural variations include the linking of the variation in attachment to child-rearing practices in different cultures, demonstrating the power of cultural memes. The Strange Situation was created and tested in the USA which means that it may be culturally-biased (ethnocentric) as it will reflect the norms and values of American culture. The Strange Situation test, which is based on the concept that attachment is related to anxiety on separation, assumes that behaviour has the same meaning in all cultures, when, in fact, social constructions of behaviour differ greatly. For example, in Japan, infants are very rarely parted from their mother. As a result, the Strange Situation represents more of a threat to them than it usually does to Western infants – yet this is interpreted via Mary Ainsworth’s measures as insecure attachment. (Imposing one set of cultural values on another unrelated culture with the idea that those values are universal is an imposed etic.) Thus, the usefulness of the Strange Situation in assessing attachment across cultures may be limited by the subjectivity inherent in observation and interpretation of behaviour.
Also, the greater variation found within cultures suggests that sub-cultural comparison studies may be more valid than cross-cultural comparisons. The significant differences also question the validity of just what the Strange Situation is measuring.
In spite of these limitations, with rare exceptions such as Germany, the Strange Situation tends to find consistently that secure attachment is the main form of caregiver-infant relationship around the world. However, it may be that it can be used to better understand some of the main sub-cultural differences found within any given country.
Further validity issues
Gavin Bremner (1994) has questioned the validity of the measures used in the Strange Situation. Earlier Cassidy (1986) postulated that secure infants will not feel the need to maintain continual proximity. If this point were to be validated, it would undermine a key characteristic of the secure attachment style.
John Bates, Christine Maslin & Karen Frankel (1985) cast doubt on the validity of the Strange Situation studies when they found that attachment style at 12 months did not predict the presence of behaviour problems at 3 years of age – it being expected that securely-attached infants would be better adjusted socially and emotionally at later ages than insecurely-attached children. However, Alan Sroufe (1983) reported that infants rated as secure in their second year have been found later to be more popular, having more initiative, higher self-esteem, less aggression and demonstrating social leadership. (This suggests that healthy PURPLE facilitates the development of healthy RED.) Correspondingly Inge Bretherton & Everett Waters (1985) suggested that insecurely attached children are more likely to be hostile and socially inadequate. The notion that early attachment type influences behaviour in later relationships – via the internal working model – receives support from Cindy Hazan & Phil Shaver’s Love Quiz studies (1987 and 1992) which are based on Ainsworth’s 3 attachment types.
Another validity question relates to the fact that the Strange Situation does not take into account the child’s temperament and the effect that might have on the attachment – see Caregiver Sensitivity vs Temperament Hypothesis. It has even been suggested that the procedure might be measuring temperament rather than attachment (Nathan Fox, Nancy Kimmerley & William Schafer, 1991) .
There is also the question as to whether the Strange Situation is testing the relationship between the caregiver and the infant rather than the resultant attachment type. Mary Main & Donna Weston (1981) found that children behaved differently in the Strange Situation according to whether they were with mother or father. This supported the earlier work of Michael Lamb (1977) who gave the instance of a secure attachment to the mother and an avoidant relationship with the father. However, most children tested through the Strange Situation with more than one attachment figure do appear to behave consistently. Marinus Van IJzendoorn, Abraham Sagi & Miryam Lambermon (1992) took the view that the best way of predicting later development was to effectively average the child’s multiple attachments (mother, father and, perhaps, childminder, for example). This suggests the Strange Situation is a valid procedure but that attachment type must be derived from more than one attachment relationship.
Then there is the question as to whether the Strange Situation measures attachment…or the response to the Strange Situation! Urie Bronfenbrenner (1979) found that attachment appeared much stronger when the Strange Situation procedure was carried out at the family home. Clearly this raises issues of ecological validity. Michael Lamb et al (1985) criticise the procedure for failing to take into account the mother’s behaviour, pointing out that some mothers/caregivers most likely behave differently towards their child in the laboratory than they would do at home, as they know they are being observed. Equally infants who show great separation protest in the Strange Situation might be less distressed by separation at home where they are in familiar surroundings and their PURPLE vMEME feels safer. Harriet Rheingold & Carol Eckerman (1973) had reported that, in natural settings where a child could approach a stranger in their own time, the child often smiled at the stranger and sometimes moved towards them.
However, Mary Ainsworth, Sylvia Bell & Donelda Stayton’s (1974) own analyses of home data found infants who had been resistant toward or avoidant of the mother on reunion in the Strange Situation had a less harmonious relationship with her at home than those (a majority) who sought proximity, interaction or contact on reunion. This suggests the qualities of the domestic relationship do transfer into the laboratory environment.
For all the criticisms and limitations, Lamb et al do acknowledge the Strange Situation to be “…the most powerful and useful procedure ever available for the study of socioemotional development in infancy.”
Type D: disorganised
However, Lamb et al do criticise Ainsworth for developing the criteria for her 3 attachment types from an initial study of just 26 American babies. They argue classification was developed too quickly from an inadequate sample.
Research by Mary Main & Judith Solomon (1990) led to the identification of a fourth type of attachment which they called Type D: disorganised. This type is characteristic of ‘high-risk’ families where children have perhaps been abused or neglected. A child showing disorganised attachment will appear confrontational and/or confused and apprehensive, with no consistent response to the events of the Strange Situation – eg: approaching the mother on her return and then avoiding her. Such children also tend to freeze or show stereotyped behaviours such as rocking. When Main reanalysed data she, Kaplan & Cassidy had collected, she found about 13% of infants they studied were disorganised. This 13% disorganised in their sample is greater than the 10% resistant usually found in the Ainsworth standard.
This raises the issue of whether Ainsworth made a serious category error in her initial classifications. It also leads to the question of how many avoidant and/or resistant classifications from the thousands of Strange Situation studies carried out would be recategorised disorganised if the data were to be reanalysed.
Back in 1978 Ainsworth – Main was her graduate student – had acknowledged difficulty in fitting all the infants observed into one of the 3 classifications (Ainsworth et al). In 1990 she acknowledged the validity of the disorganised category but cautioned that it should be treated as “open-ended, in the sense that subcategories may be distinguished”. She was concerned that the D classification might be too encompassing and might treat too many different forms of behaviour as if they were the same thing.
Type D and the effect of abuse and neglect on the infant brain has become a key area of research as it appears brain abnormalities can be caused by child abuse and neglect. This is particularly so in the limbic system which is associated with emotions. Abuse can cause disturbances that can lead to seizures and other abnormalities that show in the electrical activity measured by an electroencephalogram. In abuse cases investigated this way, the abnormality has been in the left hemisphere and has been linked to Depression and memory deficits.
Type D has been associated, to greater or lesser degree, with a number of psychopathologies. For example, Helen MacDonald et al (2008) have found a correlation between Type D attachment and the development of Post-Traumatic Stress Disorder.
Another related finding is that the corpus callosum is smaller in those who have been abused as children. A reduction in size of 24%-42% has been found in boys who have suffered neglect. In girls who have been sexually abused, a reduction of 18%-30% has been found. However, neglect of girls appears to have no effect on the size of the corpus callosum. It si thought that a reduction in size of the corpus callosum means less activity between the two hemispheres and this has been associated with changes in mood.