Introduction

Attachment theory (Bowlby, 1969/1982, 1973, 1980) has two theoretical propositions regarding the concordance of attachment classification between children and their parents: the association between parental caregiving (especially sensitivity) and children’s attachment security, and the link between parental attachment representation (e.g., attachment style) and parental caregiving.

The association between parental sensitivity and children’s attachment security is one of the core propositions of attachment theory because attachment is the relationship construct, not the trait (Waters et al., 2015). Secure children and adolescents have typically been found to have parents who are responsive to children’s attachment needs (for meta-analytic review, see Brumariu et al., 2018; Koehn & Kerns, 2018).

A link between parental attachment style and caregiving is also suggested in attachment theory, with Bowlby (1969/1982) proposing that the attachment behavioral system and caregiving behavioral system interact dynamically, and cognitive representations of early relationships serve as templates for functioning in future relationships (Jones et al., 2015). However, the influence of maternal attachment style on parental responsiveness and children’s attachment security might change as children age. For example, maternal attachment avoidance, rather than attachment anxiety, was demonstrated to be the dominant predictor of less observed parental sensitivity among 3- to 8-year-old children in previous studies (for review, Jones et al., 2015). In contrast, maternal attachment anxiety, not attachment avoidance, was reported to be associated with children’s attachment security in 7- to 13-year-old in Australia (Keenan et al., 2016) and in fourth- to eighth-graders in Canada (Doyle et al., 2000).

In addition, middle childhood might be a time when attachment is likely to change for both parents and children. Kerns and Brumariu (2016) suggested that children typically reorganize child–parent attachment relationships between the ages of 8 and 10. The phenomenon by which the child–parent attachment relationship moves to a supervision partnership phase between the ages of 10 and 14 (Koehn & Kerns, 2022) might cause changes in parents’ attachment style, in accord with Jones et al. (2015) suggestion that child behavior might not only activate the parental caregiving system but may also activate the parental attachment system directly (e.g., a new mother who is overwhelmed with parenting duties).

To the best of our knowledge, no previous studies have examined the associations among maternal attachment style, maternal responsiveness, and children’s attachment security using data from children-parent dyads in middle childhood in Japan. From a developmental perspective, it is essential to understand developmental transformations in the parent–child attachment relationship in middle childhood and identify the factors that facilitate or inhibit the parent–child negotiation process in these changes (Kerns & Brumariu, 2016).

Obtaining data from non-Western cultural contexts is essential for enhancing the cross-cultural validity of attachment theory (Mesman et al., 2016), and data from Japanese populations may be particularly useful. Although the proposal was debunked by Van IJzendoorn & Sagi (2001), the accusation of “cultural blindness” regarding Western bias in attachment theory made by Rothbaum et al. (2000) was specifically based on a comparison with the Japanese cultural context.

Therefore, the current study aimed to reveal the associations among maternal attachment style, maternal responsiveness, and children’s attachment security in fifth and sixth graders in Japan, on the basis of findings reported by Doyle et al. (2000), Keenan et al. (2016), and Kerns et al. (2001). We tested three predictions on the basis of previous findings: 1. maternal attachment anxiety would be negatively correlated with children’s attachment security; 2. maternal responsiveness would be positively correlated with children’s attachment security; 3. if our data supported Predictions 1 and 2, maternal attachment anxiety would be negatively correlated with maternal responsiveness in our sample.

In the current study, a survey was conducted using two samples. In Sample 1, a web survey was used to recruit mother–child pairs of samples from all over Japan. In Sample 2, the findings of Study 1 were replicated in a specific region of Japan (Kyoto) to test the reproducibility of the results.

Methods

Sample 1

A web-based survey was conducted via a Japanese internet research company (Macromill) with 104 pairs of mothers and children in the fifth and sixth grades (N = 55, 49, respectively). The mean age of the mothers was 42.29 years. The mean age of children was 10.77 years, with 52 boys and 52 girls.

Sample 2

The survey included 53 pairs of mothers and children in the fifth and sixth grades (N = 15, 38, respectively). The mean age of mothers was 43.70 years. The mean age of children was 11.40 years old, with 21 boys and 32 girls.

Maternal attachment style

Mothers, as mature adults, typically form attachments to a wide range of intimate interpersonal relationships, not just romantic relationships (Mesman et al., 2016). Thus, we used a 20-item Japanese version of the Experiences in Close Relationships Inventory for the generalized other (ECR-GO20J; Kanemasa, 2007; Nakao & Kato, 2004) to measure the maternal attachment style. The wording of the ECR-GO20J, back-translated into English, was nearly identical to the language of the Experience of Relationships Scale (Richards & Schat, 2011), with attachment anxiety: items 2, 4, 6, 8, 12, 14, 16, 18, 22, and 32, and attachment avoidance: items 1, 3, 9, 15, 17, 23, 25, 27, 29, and 31. The grading scale was a seven-point scale from 1 = disagree strongly to 7 = agree strongly.

Maternal responsiveness

Willingness to serve as an attachment figure is conceptualized as an indicator of parental responsiveness (Koehn & Kerns, 2018). A Japanese version of the Child Rearing Practices (CRP) scale (Kerns et al., 2001), with the item cluster willingness to serve as an attachment figure, was created in this study using a translation and back-translation procedure. Participants were asked to rate 20 statements (10 CRP items, 10 filler items) from 1 = not at all descriptive of me to 6 = highly descriptive of me.

Children’s attachment security

To measure children’s attachment security, we used a Japanese version of the Security Scale (SS; 14 items; Nakao & Murakami, 2016). Higher scores on the SS indicate more attachment security. Children were asked to choose an answer from a pair of statements, then choose whether the answer was “really true for me” or “sort of true for me.” Scoring was performed on a four-point scale from 1 to 4, following Kerns et al. (2001).

Procedure and ethical considerations

In the current study, a web-based survey was conducted on Sample 1 from July 25 to July 27, 2014. The web survey was completed by the mothers themselves. In contrast, children’s responses were entered by their mothers while they were both shown the survey screen together because of the terms and conditions of the web research company, which restricted unaccompanied under-aged users from participating in the web-based survey. Mothers were instructed as follows: “We would like your child’s answers to be as honest as possible. Therefore, while you are answering the questions on behalf of your child by selecting the most appropriate answers from the list, please do not express your own opinions.” In addition, as an ethical consideration, the following information was provided to the mothers: 1. The results of the research will not be used for any purpose other than research and education; 2. Your privacy will not be violated; 3. You can stop answering the questionnaire in the middle of the questionnaire if the psychological burden is too much.

Regarding Sample 2, survey forms were distributed and collected from a private tutoring school and an after-school childcare facility for schoolchildren, both in Kyoto city, in May 2016. In addition, survey forms were distributed and collected in a snowball format through the first author and first author’s acquaintances. As an ethical consideration, the same information as in Study 1 was clearly mentioned on the cover of the questionnaire to the mothers. The following information was clearly stated on the cover of the questionnaire to the child; 1. We will not tell anyone else what answers you have written (no invasion of privacy); 2. The results of the questionnaire have nothing to do with school performance; and 3. By answering this questionnaire, we assume that you have agreed to participate in this survey.

Results

Table 1 shows descriptive statistics, alpha coefficients, and Pearson’s correlation coefficients for each subscale. In both Samples 1 and 2, the results revealed that there was (1) a significant negative correlation between maternal attachment anxiety and maternal willingness to serve as an attachment figure, and (2) a significant positive correlation between willingness to serve as an attachment figure and children’s attachment security. The results revealed a significant negative correlation between maternal attachment anxiety and children’s attachment security in Sample 1, but not in Sample 2. Prediction 1 was supported only in Sample 1. Predictions 2 and 3 were supported in Samples 1 and 2.

Table 1 Summary of descriptive statistics and Pearson product-moment correlation coefficients in Samples 1 and 2.
Full size table

Prediction 1 appeared to not be supported in Sample 2 because of the small sample size and the effect of an outlier. We calculated Spearman’s correlation coefficients for maternal attachment style and children’s attachment security in Sample 2, which yielded a significant result (rs = −0.31, p = 0.02, Table 2). As Pearson’s correlation coefficients describe a linear association between two variables and Spearman’s correlation coefficients describe a monotonic relationship between two variables (Schober et al., 2018), we demonstrated at least a monotonic negative association between maternal attachment anxiety and children’s attachment security in Samples 1 and 2. In so doing, Prediction 1 was eventually partially supported in Samples 1 and 2.

Table 2 Summary of Spearman rank correlation coefficients in Samples 1 and 2.
Full size table

Discussion

We demonstrated an association among maternal attachment anxiety, maternal willingness to serve as an attachment figure, and fifth and sixth graders’ attachment security in Japan. Our data could be useful for enhancing the cross-cultural validity of child–parent attachment among children in our target age-group, and suggest that the influence of maternal attachment styles on parental responsiveness changes as children age in middle childhood.

We consider that this developmental transformation might occur between the ages of 8 and 10 years, when children begin to judge the social behavior of others on the basis of their motives, rather than on the results (Piaget & Inhelder, 1966/1969), and when they reorganize attachment representation (Kerns & Brumariu, 2016). For children under the age of 8 years, child-rearing behavior of parents is critical for their attachment security based on their cognitive constraints. As this observed sensitivity requires parents to not avoid social interaction and to not deal with attachment signals defensively (Mikulincer & Shaver, 2016), parental attachment avoidance but not attachment anxiety would be expected to be associated with parental responsiveness when children are under the age of 8. There was no significant correlation between attachment anxiety and the two variables in the findings of previous studies (for review, Jones et al., 2015) because attachment-anxious people are typically able to interact with their children despite their lack of confidence in their own abilities and social value (Mikulincer & Shaver, 2016).

For children over 10 years old, perceived parental availability rather than observed parental sensitivity is critical for attachment security (Kerns & Brumariu, 2016). We consider that children’s need for parental availability, rather than sensitivity, requires parents to become “stronger and wiser” (Bowlby, 1969/1982) pillars of support. As attachment-anxious people’s lack of confidence in their abilities and social value makes it difficult for them to act as an available attachment figure (Mikulincer & Shaver, 2016), parental attachment anxiety, not attachment avoidance, would be expected to be associated with parental responsiveness when children are over the age of 10. There was no significant correlation between attachment avoidance and the two variables because attachment-avoidant people might try to maintain a positive self-image through defensive information processing (Mikulincer & Shaver, 2016).

Moreover, in late adolescence, again, parental attachment avoidance would also be expected to begin to affect the child’s attachment security, because children might be able to consider perceived availability and observed sensitivity simultaneously as children’s cognitive abilities develop. Kanemasa (2007) administered a questionnaire survey to 209 pairs of university students and their mothers in Japan, and found significant correlations between parent and adolescent scores, with r = 0.14 (p < 0.05) for attachment anxiety and r = 0.15 (p < 0.05) for attachment avoidance.

We consider that the findings observed in the current study are unlikely to be limited to Japan. Similar results have been obtained in Australia (Keenan et al., 2016) and Canada (Doyle et al., 2000). Nevertheless, given the cultural specificity of Japan, our findings should be discussed from the perspective of Amae (passive love, acting on the goodwill of others in relationships, Doi, 2001, p. 65), a characteristic of Japanese culture. Japanese infants exhibit affective Amae behavior toward their parents (e.g., snuggling, seeking to be held), and begin to perform instrumental Amae behavior (e.g., being clingy, acting helplessly, exhibiting temper tantrums) in childhood (Behrens, 2004). Endo (2012) suggested that the frequency of children’s Amae behavior is highest from early to middle childhood.

Endo (2012) also suggested that Japanese parents are implicitly required to accept these Amae behaviors as much as possible, because one of the main functions of Amae behavior is to confirm the representational proximity of the attachment figure (e.g., Japanese school-aged children behaving in a clingy way, and acting helplessly in non-stressful situations to ensure the prospect of being attached to an attachment figure in a time of need). In addition, the supervision partnership phase in an attachment relationship requires parents’ availability and accessibility, but also a willingness to communicate about plans, goals, and life events, and a willingness to negotiate and recognize that both the parent and child have a right to contribute to the decision-making process (Koehn & Kerns, 2022).

These situations may be overwhelming for attachment-anxious parents in Japan, whereas attachment-avoidant people tend to deal with these situations defensively to protect themselves. Therefore, we conclude that attachment anxiety, not attachment avoidance, is associated with willingness to serve as an attachment figure, and children’s attachment security.

The current study was limited by testing a small sample size and did not investigate dyads containing children under 10 years of age and their mothers in middle childhood in Japan. Future studies should examine the parent–child negotiation process in the developmental transition and identify the factors that facilitate or inhibit this process (cf. Kerns & Brumariu, 2016).

Conclusion

Our data could enhance the cross-cultural validity of attachment in fifth and sixth graders, and suggest that the influence of maternal attachment styles on parental responsiveness changes in middle childhood.