The Strange Situation

Mary Ainsworth, who had collaborated with John Bowlby on attachment theory, wanted to go beyond simply describing attachment and find a way to measure its quality. Working with Silvia Bell, she published the Strange Situation procedure in 1970 (Ainsworth and Bell, 1970), with the full classification system following in Ainsworth et al. (1978).

The procedure takes place in a small, unfamiliar playroom equipped with toys. Observers watch through a two-way mirror or via cameras, recording the infant's behaviour at regular intervals (time sampling every 15 seconds). The procedure consists of seven episodes, each lasting approximately three minutes:

Five key behaviours are assessed throughout:

• Proximity-seeking — does the infant stay close to the caregiver?
• Exploration and secure-base behaviour — does the infant use the caregiver as a safe point of contact from which to explore?
• Stranger anxiety — how does the infant respond when an unfamiliar adult approaches?
• Separation anxiety — how does the infant react when the caregiver leaves?
• Reunion behaviour — how does the infant behave when the caregiver returns?

The Strange Situation is classified as a controlled observation because the setting, sequence, and timing of events are standardised, yet the infant's behaviour is observed rather than manipulated. It is a non-participant observation — the researchers observe but do not interact with the infant.

Types of Attachment

Using the Strange Situation, Ainsworth et al. (1978) identified three distinct patterns of infant behaviour, each reflecting a different quality of attachment to the caregiver. These are the three attachment types named on the AQA specification.

Type B — Secure attachment. Securely attached infants explore the room happily but regularly return to their caregiver for reassurance (proximity-seeking and secure-base behaviour). When the caregiver leaves, they show moderate distress, and they display some wariness towards the stranger. Crucially, when the caregiver returns, they greet them with pleasure, seek physical contact, and are quickly comforted. Secure attachment is the most common type — approximately 60–75% of British infants and around 65% of American infants are classified as secure.

Type A — Insecure-avoidant attachment. These infants explore freely and independently but do not use the caregiver as a secure base. They show little or no distress when the caregiver leaves and little anxiety around the stranger. On reunion, they make minimal effort to seek contact with the caregiver and may actively turn away or avoid them. Around 20–25% of British infants are classified as insecure-avoidant.

Type C — Insecure-resistant attachment. These infants seek greater proximity to the caregiver than others and explore the room less. They become extremely distressed when the caregiver leaves and are very wary of the stranger. However, when the caregiver returns, their behaviour is contradictory — they simultaneously seek contact and resist comfort, for example reaching out to be picked up but then pushing the caregiver away. This is the least common attachment type, found in approximately 3% of British infants and around 12–14% of American infants.

Evaluating the Strange Situation (AO3)

Before turning to cross-cultural research, it is important to assess the strengths and limitations of the Strange Situation itself, because all of the cultural variation research relies on this procedure.

Predictive validity. Research has consistently shown that attachment type as measured by the Strange Situation predicts later developmental outcomes. Ward et al. (2006) found that children classified as insecure-resistant were more likely to develop mental health problems in adulthood, while securely attached children tend to achieve more in school and enjoy healthier relationships (McCormick et al., 2016; Kokkinos, 2007). This suggests the Strange Situation is measuring something real and meaningful about the infant–caregiver relationship, not just a snapshot of behaviour on one day.

Good inter-rater reliability. Bick et al. (2012) tested inter-rater reliability for the Strange Situation among a team of trained observers and found 94% agreement on attachment type classifications. This high reliability is likely because the procedure is standardised and because the key behaviours — such as crying, crawling towards or away from the caregiver — involve large, easily observable movements. High reliability means that the classification does not depend on the subjective judgement of any single observer.

Counterpoint — temperament hypothesis. Jerome Kagan (1982) argued that what the Strange Situation actually measures may be the infant's innate temperament (genetically influenced anxiety levels) rather than the quality of the attachment relationship. If temperament explains the infant's behaviour, then the Strange Situation tells us about the child's personality rather than about their bond with a specific caregiver. However, research has shown that the same child can be classified differently with different caregivers, which supports the view that attachment type reflects the relationship, not just the child's temperament.

A fourth attachment type. Main and Solomon (1986) identified a fourth classification — Type D: disorganised attachment — characterised by an inconsistent, confused mix of avoidant and resistant behaviours. Type D infants often freeze, display contradictory movements, or show fear of the caregiver. This attachment type is relatively rare and is strongly associated with experiences of abuse or severe neglect. The identification of Type D suggests that Ainsworth's original three-category system, while highly useful, does not capture the full range of attachment patterns.

Cultural Variations in Attachment

Child-rearing practices differ across cultures: in some societies, infants are rarely separated from their mothers; in others, independence is encouraged from a young age. The question is whether these differences in caregiving lead to different patterns of attachment, or whether attachment types are universal, as Bowlby's theory would predict.

Van IJzendoorn and Kroonenberg (1988) — the key study

Van IJzendoorn and Kroonenberg (1988) — Conducted a meta-analysis of 32 studies from eight countries, all of which had used the Strange Situation to classify infant attachment type. The combined data covered 1,990 children. The 32 studies were spread across the US (18 studies), Germany (3), the Netherlands (4), the UK (1), Japan (2), Israel (2), China (1), and Sweden (1). The meta-analysis involved combining the results of all studies and weighting each study according to its sample size to produce overall proportions for each country.

Key findings:

• Secure attachment was the most common type in every country studied, ranging from approximately 50% (China) to 75% (UK). This provides support for Bowlby's claim that secure attachment is a universal norm.
• In individualist Western cultures (US, UK, Germany, Sweden, Netherlands), insecure-avoidant attachment was the more common insecure type, and insecure-resistant rates were relatively low (under 14%).
• In collectivist and non-Western cultures (Japan, China, Israel), the pattern reversed: insecure-resistant attachment was notably higher (25–29%), while insecure-avoidant attachment was considerably lower (5–7% in Japan and Israel).
• A critical finding was that within-country variation was 150% greater than between-country variation. For example, within the US alone, secure attachment rates ranged from 46% to 90% across different studies.

The overall conclusion is twofold. First, secure attachment appears to be the dominant type across all cultures, consistent with Bowlby's idea that attachment is an innate, biologically-driven process. Second, the specific distribution of insecure attachment types varies according to cultural child-rearing practices — cultures that prioritise close physical proximity between mother and infant (such as Japan) tend to produce more insecure-resistant classifications, while cultures that value early independence (such as Germany) produce more insecure-avoidant classifications.

Supporting studies

Simonelli et al. (2014) assessed 76 Italian infants aged 12 months and found only 50% were securely attached, with 36% classified as insecure-avoidant. This was a lower rate of secure attachment than previous Italian studies had found. The researchers attributed the shift to changing social patterns — more Italian mothers were working long hours and using professional childcare, reducing the time available for consistent, responsive caregiving.

Jin et al. (2012) used the Strange Situation to assess 87 Korean infants and found that, while overall proportions of secure and insecure attachment were similar to other countries, the distribution of insecure types was distinctive: the vast majority of insecurely attached Korean infants were classified as insecure-resistant, with only one baby classified as insecure-avoidant. This pattern closely mirrors the Japanese findings in van IJzendoorn and Kroonenberg's study, which makes sense given the similar child-rearing practices in Korea and Japan, where mother–infant separation is culturally unusual.

Evaluating Cross-Cultural Research (AO3)

Cross-cultural attachment research raises several important methodological and interpretive issues that examiners frequently target.

The Strange Situation as an imposed etic. Ainsworth developed the Strange Situation based on observations of American infants. When the same procedure is applied to infants from cultures with very different child-rearing norms, it may not measure the same thing. For example, Takahashi (1986) found that Japanese infants showed extremely high levels of separation anxiety in the Strange Situation — not because they had insecure attachments, but because Japanese mothers very rarely leave their babies with anyone else, making the separation episode an abnormally distressing experience. Similarly, in Germany, where independence is culturally valued, a child's lack of distress at separation might be interpreted as insecure-avoidant when it actually reflects culturally appropriate behaviour. This is an example of imposed etic — assuming that a measure developed in one cultural context is equally valid in all others.

The problem of studying countries rather than cultures. Van IJzendoorn and Kroonenberg's meta-analysis compared attachment patterns across countries, but a country is not the same as a culture. Multiple distinct cultures, subcultures, and child-rearing practices can exist within a single nation. Sagi and van IJzendoorn found that within Israel, kibbutz-raised children had an overrepresentation of insecure-resistant attachment, while urban Israeli children showed attachment distributions similar to Western norms. This demonstrates that within-country cultural variation can be substantial and that treating an entire country as a single culture is an oversimplification.

Use of indigenous researchers. A notable strength of the cross-cultural attachment literature is that many studies were conducted by researchers from the same cultural background as the participants — for example, Takahashi in Japan and Grossmann et al. in Germany. This reduces the risk of researcher bias, miscommunication, or misinterpretation of culturally specific behaviours, thereby improving the validity of the data collected.

Confounding variables in meta-analyses. The 32 studies combined in van IJzendoorn and Kroonenberg's meta-analysis were not originally designed to be compared with each other. They differed in sample characteristics (social class, urban/rural, age of infants), environmental conditions (room size, toys available), and exact procedural details. These uncontrolled differences act as confounding variables, making it difficult to be certain that any variation in attachment patterns is genuinely due to cultural differences rather than methodological inconsistencies between studies.

Evaluation Bank (AO3)

Strength: The Strange Situation has strong predictive validity, meaning that attachment classifications made in infancy are associated with meaningful developmental outcomes later in life. Ward et al. (2006) found that children classified as insecure-resistant were more likely to develop mental health difficulties in adulthood, while securely attached infants tend to achieve more in school (McCormick et al., 2016) and experience healthier adult relationships. This pattern of findings suggests that the Strange Situation is capturing something genuinely important about the infant–caregiver relationship, rather than simply measuring a momentary behavioural response. The fact that attachment type predicts outcomes across many years strengthens confidence in its construct validity and supports the broader claim that early attachment quality has lasting developmental significance — consistent with Bowlby's theory that internal working models formed in infancy shape later relationships.

Limitation: A significant criticism of cross-cultural attachment research is that the Strange Situation represents an imposed etic — a procedure developed within a specific cultural context (the US and UK) that is then applied to other cultures as though it has universal meaning. Takahashi (1986) demonstrated this problem when studying Japanese infants: they displayed extremely high levels of separation distress, leading to disproportionate classification as insecure-resistant. However, Takahashi argued that this distress reflected the culturally unusual nature of the separation (Japanese mothers very rarely leave their infants) rather than genuine attachment insecurity. This means that attachment type classifications produced in non-Western cultures may not be valid, because the behaviours being observed have different meanings in different cultural contexts. This relates to the broader issue of cultural bias (ethnocentrism) in psychology — researchers risk imposing Western norms and interpretations onto other cultures, producing findings that reflect the assumptions of the researchers rather than the reality of the participants' experiences.

Limitation: The finding that within-country variation in attachment type distributions was 150% greater than between-country variation poses a serious challenge to the usefulness of cross-cultural comparisons. Van IJzendoorn and Kroonenberg (1988) found, for example, that within the US, secure attachment rates ranged from 46% in one sample to 90% in another — a far greater range than the differences found between the US and Japan or Germany. This suggests that factors such as social class, poverty, urban or rural location, and subculture within a country are more powerful influences on attachment patterns than broad national differences. As a result, statements about "Japanese attachment" or "German attachment" are misleading oversimplifications. This limitation connects to the wider methodological issue of the ecological fallacy — drawing conclusions about cultural groups from data that does not adequately represent the diversity within those groups. It suggests that future research should focus on identifying specific child-rearing practices associated with different attachment types, rather than making broad national comparisons.