Exploring the Links Between Parental Attachment Style, Child Temperament and Parent-Child Relationship Quality During Adolescence

Linked with a myriad of developmental outcomes, the parent-child relationship serves an important function, and it is therefore important that determinants of this relationship are elucidated so that optimal outcomes can be promoted. This study investigated links between parental attachment style and child temperament in the prediction of parent-child relationship quality during early adolescence. One hundred parents of 12–15-year-olds completed an online survey. Findings indicated that when examining parent attachment style alone, attachment-related anxiety but not avoidance predicted parent-child relationship quality. When examining child temperament alone, negative affect and affiliation predicted parent-child relationship quality. When examining both parental attachment style and child temperament in the same model, child temperament was uniquely predictive of the parent-child relationship quality. No moderation effects were detected. More precise models and expanded measures of parental attachment style, child temperament and parent-child relationship quality are recommended to fully capture the transactional and dynamic nature of the parent-child relationship. Parental attachment avoidance was not predictive of parent-child relationship quality. Parental attachment anxiety was predictive of relationship quality when examined independently but not when examined with child temperament. Child negative affect and affiliation were predictive of relationship quality. Child effortful control and surgency were not predictive of relationship quality. No moderation effects were detected. Parental attachment avoidance was not predictive of parent-child relationship quality. Parental attachment anxiety was predictive of relationship quality when examined independently but not when examined with child temperament. Child negative affect and affiliation were predictive of relationship quality. Child effortful control and surgency were not predictive of relationship quality. No moderation effects were detected.

Parental attachment avoidance was not predictive of parent-child relationship quality.
• Parental attachment anxiety was predictive of relationship quality when examined independently but not when examined with child temperament.
• Child negative affect and affiliation were predictive of relationship quality.
• Child effortful control and surgency were not predictive of relationship quality.
• No moderation effects were detected.
The parent-child relationship is often considered to be the most enduring and significant relationship in an individual's life (Golish, 2000).Linked with social, developmental, educational, and mental health outcomes for children of all ages, the quality of the relationship that develops between parents and their children has been found to play a pivotal role in successful child adjustment in every life domain (Hong and Park, 2012).Whilst serving an important function for children during every stage of development, the period of adolescence marks an important turning point in this relationship.Parent-child relationships during adolescence are characterised by flux, with both children and parents having to negotiate their respective roles in the context of a different relationship (Moretti and Peled, 2004), thus making the maintenance of emotional connectedness more difficult (Ryan and Lynch, 1989).The relationship goal changes from one of proximity to availability as the child strives towards autonomy with greater reciprocity in the relationship evident between parent and child (Kerns et al., 2015).The challenge for parents is to develop greater tolerance as their adolescent children strive for separateness and a sense of identity, while simultaneously providing support and validation (Grotevant and Cooper, 1985).
Research indicates that a healthy transition to autonomy is not facilitated by detachment from parents (Lamborn and Steinberg, 1993), but rather by secure relationships and emotional connectedness with parents (Ryan and Lynch, 1989).Good quality parent-child relationships during adolescence have been shown to promote the process of separation-individuation, which is crucial for identity formation (Lopez and Gover, 1993) and is linked with a lower risk of engaging in excessive drinking, substance misuse, and risky sexual behaviour (Cooper et al., 1998), fewer mental health difficulties (Raja et al., 1992), less conflict with family and peers (Ducharme et al., 2002), and more constructive coping skills (Howard and Medway, 2004).
Research aiming to elucidate determinants of parentchild relationship quality during adolescence has been the subject of debate, given the array of both parent and child factors identified in the literature.One factor thought to influence parent-child relationship quality during adolescence is the quality of attachment (Moretti and Peled, 2004).Attachment theory (Ainsworth, 1979;Bowlby, 1969) emphasises the parent's emotions, cognitions, and behaviour towards the child as paramount predictors of the quality of the relationship that emerges.According to the theory, early parent-child interactions, based on an interplay of conceptual behavioural systems, influence an individual's capacity to establish good quality interpersonal relationships throughout the lifespan.The quality of the relationship that develops between the parent and the child is the result of the interaction between the child's attachment behavioural system and the parents' caregiving behavioural system.While the primary function of the child's attachment system is to protect it from danger, the parent's role is to provide a "secure base" from which the child can explore and a "safe haven" in times of distress (Kerns et al., 2015).In the context of well-functioning parent-child relationships, both systems work in synchrony (Jones et al., 2015).Predicated by parental attunement and sensitivity, children are thought to develop attachment styles or patterns of interaction, with synchronous systems leading to what Ainsworth and Bell (1970) categorised as a "secure" attachment relationship and asynchronous systems leading to either insecure or disorganised attachments (Main and Soloman, 1990).
Although serving an important function in terms of determining the quality of relationships formed between parents and their children, these early attachment relationships also establish in the child cognitive frameworks or internal working models consisting of mental representations for comprehending the world, self, and others (Bowlby, 1969).Subsequently, these internal working models act as relationship blueprints, influencing thoughts, feelings, and behaviours in all relationships throughout the lifespan (Pietromonaco and Barrett, 2000).Based on this, Bowlby (1969) predicted that parents' own experiences and representations of attachment are likely to influence how they behave toward their children, which in turn influences the quality of their relationship.Thus, the quality of the relationship that develops between parents and their children depends not only on the quality of their early interactions but importantly, is also influenced by the parent's own attachment experiences and representations.With extensive empirical evidence supporting the predictive ability of various parental emotions, cognitions, and behaviours on the quality of parent-child relationships, measuring parents' own attachment experiences in this manner may thus add further depth to the study of this complex relationship because they are thought to provide a template guiding feelings, thoughts, and behaviours in relationships from infancy to old age (Jones et al., 2015).Support for this view comes from studies consistently showing that adults classified as securely attached tend to be more sensitive and responsive parents compared to adults classified as insecurely attached (see van IJzendoorn, 1995, for a meta-analysis).
Temperament theory (Thomas and Chess, 1956) on the other hand emphasises child temperament as being fundamental to relationship quality.Temperamental differences, consisting of relatively stable, biologically based individual differences in reactivity and self-regulation abilities, are thought to affect the way children respond to their environment, thus influencing the type of relationship that evolves between parent and child (Calkins, 2012).For instance, individual differences in affective and motivational responses to stimuli (components of reactivity), are thought to explain the tendency for some children to approach novel stimuli with intrigue while others feel threatened (Clark et al., 2015), with the consequence that these differences can affect the nature of the relationship that develops with their parents.Offering a challenge to the explanation provided by attachment theory, temperament theorists suggest that individual differences in attachment security classifications in the Strange Situation relate to child temperament differences, rather than parental responses, with more "difficult" children exhibiting more extreme reactions to separation and reunion and thus more likely to be classified as insecure than their "easy" counterparts (Kagan, 1982).Proponents of temperament theory thus posit child temperament to be predictive of the quality of relationship formed between parents and their children, with evidence supporting the notion that a secure relationship is more easily formed between a caregiver and a child who has an easier temperament, than with a child who is characteristically more fearful, negative, or socially apprehensive (Durbin et al., 2007).
With a growing understanding of the reciprocity of parent-child relationships (Hipwell et al., 2008), a greater awareness of the complexity of the joint influence of both parent and child factors in the development of parent-child relationships is evident in the literature.Studies have shown that parents not only shape children's temperamental, emotional, and self-regulatory characteristics (Crockenberg et al., 2008), but they are also responsive to temperament, with children eliciting distinct parenting behaviours (Collins et al., 2000).This view has received support from studies such as Dix and Lochman (1990) who found that parents of "difficult" children reported more negative emotions and harsher disciplinary responses compared to parents of "less difficult" children.Children's temperamental differences in emotionality and selfregulation may also influence how responsive or sensitive they are to parenting with a growing body of research supporting this view (Belsky, 2005).
Regardless of whether parent or child factors are more influential in parent-child relationship quality, it is likely that both temperament and parenting contribute uniquely and simultaneously to its development (Kiff et al., 2011).Moreover, it is also probable that the interaction of parent and child factors with one another also influences the quality of the relationship that develops between the dyad.This view is supported by the Goodness of Fit model developed by Chess and Thomas (1991), which suggests that child outcomes including parent-child relationship quality are influenced by the level of congruence or lack of, between the child's temperament and the personalities, attitudes, and parenting practices of the parents.Studies have indicated that an incompatible or "poor fit" between parental responses and child temperament may generate negative interactions leading to adverse parent-child relationships and suboptimal outcomes, while a "good fit", may lead to positive parent-child relationships that safeguard children from risks associated with potentially harmful temperament traits (Hong and Park, 2012).
Traditionally, research on parent-child relationship quality has utilised a variety of measurement instruments for both parental attachment style and child temperament.For instance, while interview-based measures such as the Adult Attachment Interview (George et al., 1985) have been considered the gold standard measurement instrument for adult attachment, recent research syntheses have confirmed the applicability of self-report measures of parental attachment style, which fundamentally measure parental behaviours, cognitions, and emotions, to the study of parentchild relationships (see Jones et al., 2015).Moreover, although in-person and relationship/contextual variability has been indicated in adult attachment styles, these have been found to relate predominantly to romantic relationships rather than to parental relationships whose representations are thought to be more stable over time (Fraley et al., 2011), thus confirming the suitability of attachment style measures to the study of parent-child relationships.Also, while commonly measured categorically, psychometric evidence indicates that attachment styles are better captured by dimensional measures (Brennan et al., 1998) which assess an individual's degree of attachment-related avoidance and attachment-related anxiety.Attachmentrelated avoidance, characterised by a tendency to deactivate the attachment system resulting in discomfort with dependency or closeness in relationships (Shaver and Mikulincer, 2002) has been linked with less sensitive, responsive, and supportive parental behaviour (Berlin et al., 2011), less constructive conflict behaviour from mothers but not fathers (Feeney, 2006), and greater parenting stress (Kor et al., 2012).Attachment-related anxiety, on the other hand, is characterised by a tendency to hyper-activate the attachment system, resulting in intense fears of rejection/abandonment and a strong desire for closeness in relationships (Shaver and Mikulincer, 2002).This has been linked with missing the child's signals and interfering with exploration (Selcuk et al., 2010), less constructive conflict behaviour from both mothers and fathers (Feeney, 2006), greater emotion-focused coping (Mikulincer and Florian, 1999), and overwhelming perceptions of family responsibilities (Kohn et al., 2012).
Regarding temperament, although four different progenies are evident in the literature (Buss and Plomin, 1984;Thomas and Chess, 1956;Goldsmith and Campos, 1982;Rothbart and Derryberry, 1981), there is consensus regarding its structure with Rothbart and Derryberry's (1981) model among the most supported approaches to adolescent temperament (Snyder et al., 2015).Whilst originally thought to comprise several lower-order factors (e.g., fear, frustration, etc.), which combine to produce three higher-order factors including Negative Affectivity (NA), Positive Affectivity (PA), and Effortful control (EC) (Putnam et al., 2001), recent psychometric evidence suggests that PA is not adequately measured by temperament instruments (Snyder et al., 2015).It is therefore suggested that adolescent temperament is better captured by four factors: Negative Affectivity (i.e. the tendency to be easily frustrated, irritable, fearful, and sad), Effortful Control (i.e. the ability to control one's behaviour, regulate emotions, and sustain attention), Surgency (i.e. the tendency toward novelty and sensation-seeking, high-intensity pleasure, and positive affect) and Affiliativeness (i.e. the tendency toward emotional empathy, concern for others, and a desire for closeness with others) (Snyder et al., 2015).

The Current Study
To date, much of the research incorporating parental attachment styles in the study of parent-child relationships pertains to very young children, with few studies relating to adolescents.Similarly, Jones et al.'s (2015) review found no studies that had considered the joint effects of both parental attachment style and child characteristics in the prediction of parent-child relationships.Given the overwhelming evidence for the importance of good quality parent-child relationships for overall adolescent adjustment, in addition to the reciprocal and interactive nature of the parent-child relationship, consideration of the joint influence of factors associated with each member of the dyad is essential in further understanding this complex relationship.The current empirical study aims to address this gap by (a) investigating the degree to which parental attachment style and child temperament are predictors of parent-child relationship quality during early adolescence, and (b) establishing whether parental attachment style moderates the relationship between child temperament and parent-child relationship quality.
It is hypothesised that both attachment-related avoidance and attachment-related anxiety will significantly predict parent-child relationship quality during early adolescence.A considerable body of research exists supporting the positive association between responsiveness and warmth, characteristics generally linked with secure parents (i.e., those with lower levels of attachment-related avoidance and attachment-related anxiety) and the quality of the parentchild relationship (Cooper et al., 1998).Noted as a time of increased parent-child conflict and autonomy development, supporting a child during the period of early adolescence may be particularly challenging for insecure parents (i.e.those with higher levels of attachment-related avoidance and attachment-related anxiety), who may adopt less constructive conflict behaviour and/or struggle to support their adolescents' attachment needs for availability as opposed to proximity, leading in turn to poorer quality parent-child relationships.
It is also hypothesised that negative affectivity, surgency, affiliativeness, and effortful control will all significantly predict parent-child relationship quality during early adolescence.A study by Chiah and Baharudin (2012) found that positive affectivity was related to strengths or pro-social adjustment, while negative affectivity was related to difficulties or antisocial behaviour.Given that social-emotional adjustment reflects an individual's emotion regulation ability and social functioning, including in relationships (Santrock, 2008), it is anticipated that the current study will duplicate these findings concerning parent-child relationships.It is thus hypothesised that higher levels of adolescent NA will result in lower parent-child relationship quality, while higher levels of affiliativeness and surgency (formally thought to represent PA), will result in higher quality parent-child relationships.Similarly, as individual differences in self-regulation have also been implicated in peer relationships (Olson and Lifgren, 1988), as well as childteacher relationship quality (Diaz et al., 2017), it is hypothesised that lower levels of adolescent effortful control will result in lower levels of parent-child relationship quality during early adolescence.
In line with research on the Goodness of fit model (Chess and Thomas, 1991), it is hypothesised that parental attachment style will moderate the relationship between child temperament and parent-child relationship quality.This view is supported by research that indicates that the influence of temperament is mitigated by a caregiver's ability to modify their behaviour to fit the needs of the child (Hong and Park, 2012).In cases where children exhibit more difficult temperamental characteristics, a poorer quality parent-child relationship may therefore be evident where parents are less able to adapt their parenting behaviour due to factors such as personality or stress levels (Hong and Park, 2012) or in this case, parental attachment style.The links between parental attachment style and adaptability (Hamarta et al., 2009), and parental attachment style and stress management (Berant et al., 2003) have all been well documented, giving further weight to the suggestion that parental attachment style will moderate the relationship between child temperament and parentchild relationship quality.

Participants
Participants for this study were custodial parents of children between the ages of 12 and 15 years.A parent was defined as any individual with parental responsibility for an early adolescent child, with the child living with them on a fulltime basis.Participants were required to be proficient in the English language, to avoid language barriers.There was no discrimination regarding employment status, with included parents working both within the family home (e.g., full-time parents/homemakers) and outside of the family home (e.g., employed/self-employed). Parents with a diagnosed mental health disorder or active addiction were excluded from the study.
Demographic information for the study participants is outlined in the supplementary section.A total of 100 participants (91% female) took part in the study.Participants were predominantly Irish (81%), aged between 40-49 years (60%), and co-parenting with a partner or spouse (86%).Seventy-seven percent reported having attended third-level education with 53% of participants employed in the professional sector.Of all participants, 78% reported being in either part-time or full-time paid employment with 16% identifying themselves as full-time parents.The number of children in the household varied, with 9% of participants reporting having one child, 28% having two children, 33% having three children, 22% having four children, and 8% having five or more children.Categorisation of the children described by parents in the study yielded the following characteristics: Sex (male: 48%; female: 52%), Age (12 years = 32%; 13 years = 22%; 14years = 20%; 15 years = 26%) and Birth Order of Target Child (1st born = 42%; 2nd born = 37%; 3rd born = 10%; 4th born = 8%; 5th born = 1%; other = 2%).

Procedure
Once ethical approval was obtained from the University of Derby research ethics committee, several non-random sampling methods were employed to recruit participants.Convenience sampling was used, which involved posting a link to the online study on social media sites and on a university research recruitment site, with snowballing used to increase survey reach (i.e., participants were asked to forward the link to other eligible parents known to them).An email was also sent to principals of all primary and postprimary schools in the counties of Galway and Mayo in the West of Ireland, seeking their assistance in participant recruitment.Two primary schools in Galway agreed to distribute information about the study via email to eligible parents on their database and a link to the survey was forwarded to the principals of these schools.No schools in the county of Mayo agreed to distribute information about the study.
Once potential participants clicked on the study link, they were presented with information about the study in the form of a parent/guardian participant brief and consent form.Participants were required to indicate their informed consent before progressing to the study questionnaires.The survey took approximately 15 minutes to complete.The survey remained live online for a period of six weeks, during which time people could participate in the study.

Parental attachment style
Parental attachment style was measured using the Relationship Structures Questionnaire (ECR-RS), a self-report instrument adapted by Fraley et al. (2011) from the widely used Experiences in Close Relationships-Revised Questionnaire (ECR_R; Fraley et al., 2000).Designed to assess attachment patterns in close relationships, the ECR-RS, similar to its predecessors, the ECR-R and Experiences in Close Relationships questionnaire (ECR; Brennan et al., 1998) assesses individual differences in two fundamental dimensions underlying attachment patterns, attachmentrelated anxiety (intense fears of rejection/abandonment and a strong desire for closeness in relationships), and attachment-related avoidance (discomfort with dependency or closeness in relationships) (Shaver and Mikulincer 2002).With research demonstrating that the alpha reliability estimates of the scores obtained using the ECR-RS (α = 0.85, Fraley et al., 2011) are highly comparable to those derived from the longer versions of this scale (α = 0.91 for the ECR, Brennan et al., 1998; α = 0.90 for the ECR-R, Fraley et al., 2000), the ECR-RS is an attractive measurement instrument due to its shorter length and reduced participant burden.
The ECR-RS consists of 9 items (statements) that are designed to assess participants' global attachment orientations.Participants are asked to read each statement and rate on a 7-point Likert scale the extent to which they believe the statement best describes their feelings about close relationships in general.Each point, ranging from 1 (strongly disagree) to 7 (strongly agree) is labelled to make the task as concrete as possible.The avoidance dimension, representing the extent to which individuals are comfortable depending on others, is captured in the first 6 items, with statements such as "It helps to turn to people in times of need".The anxiety dimension, representing the extent to which individuals tend to worry about the availability and responsiveness of an attachment figure, is captured in the last 3 items, with statements such as "I often worry that other people do not really care about me".Scores on each item are added to obtain an overall avoidance score (items 1 to 6, with the first 4 items reverse scored), and an overall anxiety score (items 7 to 9), with lower scores on each dimension indicative of greater attachment security.
Mean scores for global attachment representations (computed by dividing the total overall score per dimension by the number of items per dimension) using this measure in two prior large-scale studies have been found to range between 2.53 and 4.47 for attachment-related anxiety, and between 3.18 and 3.75 for attachment-related avoidance, with standard deviations ranging between 1.19 and 1.67, and between 0.96 and 1.23, respectively (Fraley et al., 2011(Fraley et al., , 2015)).With lower scores on each of the attachment style variables indicating greater levels of attachment security, the mean scores observed in these studies suggest that the average person is relatively secure (Fraley et al., 2011).This is convergent with studies using interview-style measures of attachment style, which have reported secure attachment in 46% of the population (Bakermans-Kranenburg and van IJzendoorn, 2009).

Child temperament
Child Temperament was assessed using the Early Adolescent Temperament Questionnaire-Revised (EATQ-R)-Parent Report Form, developed by Ellis and Rothbart (1999), a revision of an earlier instrument devised by Capaldi and Rothbart (1992).Designed to assess temperament in adolescents aged 9 to 15, the EATQ-R-Parent Report Form assesses parental perceptions of their child's temperament across 10 dimensions, ultimately loading onto 4 temperament factors as follows: Effortful Control, Surgency, Negative Affect, and Affiliativeness.Psychometric testing has revealed alpha values of between 0.61 and 0.74 for each of the EATQ-R subscales, indicating modest but sufficient internal consistency, with test re-test correlation coefficients indicating modest to good validity (Muris and Meesters, 2009).
The 62-item instrument requires participants to rate on a 5-point Likert scale, ranging from 1 (almost always untrue) to 5 (almost always true) the answer which best describes how true each statement is for their child.A measure for Effortful Control was obtained by summing the total scores of the scales measuring Attention (6 items, e.g., "finds it really easy to concentrate on a problem"), Inhibitory Control (5 items, e.g., "has a hard time waiting his/her turn to speak"), and Activation Control (7 items, e.g., "has a hard time finishing things on time").Similarly, Surgency was measured by summing the total scores of the scales measuring Surgency (9 items, e.g., "would like to drive a racing car"), Fear (6 items, reversed scored, e.g."worries about getting into trouble"), and Shyness (5 items, reverse scored, e.g."is shy").Negative Affect was measured by summing the total scores of the scales measuring Frustration (6 items, e.g., "is annoyed by little things other kids do"), Depressive Mood (5 items, e.g., "feels like crying over very little on some days"), and Aggression (7 items, e.g., "if very angry, might hit someone").Finally, Affiliativeness was measured by summing the total score of the scale measuring Affiliation (6 items, e.g., "is quite a warm and friendly person").Total scores for each construct were utilised, with higher total scores indicative of higher levels of each temperamental attribute.

Parent-child relationship quality
Parent-Child Relationship Quality was assessed using the Parent-Child Interaction Questionnaire-Revised (PACHIQ -Parent version) designed by Lange et al. (2002) to assess how parents view their relationship with their child.Conceptually based on learning theory and structural systems theory, the PACHIQ-R reveals a two-factor structure, namely conflict resolution and acceptance.For the purposes of this study, however, the instrument was used in its broader sense to obtain an overall measure of parent-child relationship quality, with higher scores indicative of a more positive relationship.Psychometric evaluation of the PACHIQ-R has revealed coefficient alphas of 0.86 for both mother and father reports, indicating good reliability and validity of the overall measure (Lange et al., 2002).Participants are asked to read 21 statements and select on a 5-point Likert scale ranging from 1 (never) to 5 (always) the answer that best describes how true each statement is for them and their target child (e.g., "I show my appreciation clearly when my son/daughter does something for me").Overall parent-child relationship quality scores were obtained by summing the scores of each item (with reverse coding required for negatively formulated items).In line with suggestions for use by instrument developers (Lange et al., 2002), total scores, rather than subscale scores were used to measure parent-child relationship quality.

Socio-economic status
Socio-economic status (SES) was identified as a potential confounding variable and was therefore controlled within the analyses.Aarø et al. (2009) identified parental education level as a satisfactory measure of SES; therefore, this was used in the current study.Participants were asked to select from a series of multiple-choice options to indicate which option best described their level of education (e.g., primary school, post-primary school, college certificate or diploma, bachelor's degree, etc.).This was then transformed into a dichotomous variable to enable inclusion within the analyses.Responses were divided into high parental education (university bachelor's degree or higher) and low parental education (college certificate/diploma or lower).

Analytic Strategy
Prior to conducting the main analyses, data were screened to ensure that the assumptions of hierarchical multiple regression analysis were met.Descriptive statistics and bivariate correlations were then conducted, followed by hierarchical multiple regression analysis (using the Enter method) to examine the extent to which the dependent variable (parent-child relationship quality) could be explained by the addition of one or more predictor variables (parental attachment style and child temperament) while controlling for the effects of the confounding variable (SES).To control for the effects of SES, parental education level was entered at the first stage of the hierarchy.
The predictor variables, parental attachment style (measured by degree of attachment-related anxiety and attachment-related avoidance) and child temperament (measured by degree of effortful control, negative affect, surgency, and affiliativeness) were first examined independently, whilst controlling for SES.The full model of parental attachment style and child temperament was then examined by entering all six variables at block 2 of the hierarchy, again controlling for SES.Effect sizes were calculated using the following formula: R 2 /(1 − R 2 ), to obtain Cohen's f 2 values (Coolican, 2014).
Moderation analyses were then conducted to examine whether parental attachment style moderated the relationship between child temperament and parent-child relationship quality.

Preliminary Data Screening
All data were initially screened for missing data, with no missing data identified.Normality was assessed by visual inspection of Q-Q plots, histograms, and boxplots, which all suggested approximate normal distribution.Further examination of normality by calculation of Zskew and Zkurtosis values revealed no problematic skew or kurtosis, with all values falling within the normal range (values all within 2.58 for sample size ≥100).Examination of standardised residuals (z-scores) revealed no extreme outliers (all absolute values within ±3 SD), with level of error within the model deemed acceptable (1% of standardised residual absolute values ≥2.5 and >5% with absolute value ≥1.96).Shapiro-Wilk and Kolmogorov-Smirnov statistics suggested non-normality in avoidance, anxiety, and affiliation, however, Elliot and Woodward (2007) cautioned the validity of these tests with sample sizes >50.Given the size of the current sample (n = 100), and that all other indicators suggested approximate normal distribution, the assumption of normality was assumed.

Data Screening for Multiple Regression Analysis
Examination of the correlation matrix indicated the absence of multicollinearity, with all values <0.9.This was further confirmed by observation of low variation inflation factor (VIF) values with all values <10 and all tolerance values observed at >0.2.Normality of the residuals was assessed by visual inspection of probability-probability (P-P Plot), with observed values close to the diagonal line, suggesting relatively normal distribution.Inspection of the histogram and scatterplot also indicated normality of residuals with no outliers identified (all data points between −3/+3).Moreover, examination of casewise diagnostics and studentized deleted residuals confirmed the absence of outliers, with no values greater than ±3 SD.Average leverage values all measured <0.2 and residual statistics pertaining to cook's distances were all <1, suggesting that no individual cases had undue influence on the model.Observation of the variance of each of the residuals, through visual inspection of the scatterplots measuring standardised predicted (ZPRED) against standardised residuals (ZRESID) revealed no funnelling or circular patterns indicative of heteroscedasticity, thus suggesting homoscedasticity.Linearity of the individual predictor variables with the DV was assessed through the production of scatterplots (with regression line for each predictor) and through examination of the partial regression plots.Linearity of the combined IVs with DV was assessed through visual inspection of the ZPRED against ZRESID plot, with findings suggesting that all variables (individual and combined) met with the assumption of linearity.There was independence of residuals, as assessed by a Durbin Watson statistic of 1.86, suggesting the absence of serial autocorrelation.

Descriptive Statistics
Central tendency (mean), variability (standard deviations), and score range for each of the study variables are presented in Table 1.
Mean scores for attachment-related anxiety (3.04) and attachment-related avoidance (2.38) fell within the range reported by previous large-scale studies (Fraley et al., 2011(Fraley et al., , 2015)).Inspection of the mean scores on the EATQ-R temperament scales indicated that surgency (70.39) and effortful control (59.60) were the most often endorsed temperament dimensions by the current sample.These results are somewhat like the findings of Muris and Meesters (2009), who reported higher endorsement of activation control, a subcomponent of effortful control in non-clinical youth, however less like their findings in relation to affiliation.The mean score for parent-child relationship quality (85.64) is like the findings of Lange et al. (2002), who reported mean scores of 87.84 (father-report) and 89.74 (mother-report) for normal populations, and mean Pearson's correlations between each of the study variables are presented in Table 2. Examination of the correlation coefficients revealed that parent-child relationship quality was significantly positively correlated with three of the variables, child effortful control (r = 0.34, p < 0.001), child surgency (r = 0.35, p < 0.001), and child affiliation (r = 0.43, p < 0.001).Significant negative relationships were found with parental anxiety (r = −0.25,p = 0.006) and child negative affectivity (r = −0.64,p < 0.001).Parental avoidance was not significantly correlated with parentchild relationship quality (r = 0.05, p = 0.314).Parental anxiety was also significantly positively correlated with child negative affectivity (r = 0.43, p < 0.001) and significantly negatively correlated with child surgency (r = −0.27,p = 0.004).Parental avoidance was not significantly correlated with any of the other variables.

Inferential statistics
Hierarchical Multiple Linear Regression (Enter method) was used to determine whether parental attachment style (as measured by degree of attachment-related avoidance and attachment-related anxiety) and child temperament (as measured by degree of child related effortful control, negative affectivity, surgency, and affiliativeness) predicted parent-child relationship quality (DV), controlling for SES (as measured by parental education level).
To test the relative contributions of each of the predictor variables and to gauge the degree to which some of the correlated variables shared variance, the constructs were initially examined separately.Controlling for SES, the first model examined the independent predictive ability of parental attachment style, and the second model examined the independent predictive ability of child temperament.Following this, the third and final model examined the predictive ability of both parental attachment style and child temperament combined.
Independent analysis of the parental attachment style variables as predictors of parent-child relationship quality during early adolescence revealed a statistically significant result for the overall model (SES, attachment-related avoidance and attachment-related anxiety), with the model explaining 10% of the variance in parent-child relationship quality, R 2 = 0.10, F(3, 96) = 3.65, p = 0.015, adjusted R 2 = 0.07 (see Table 3).After controlling for SES, parental attachment style explained an additional 8% of the variance in parent-child relationship quality, R 2 change = 0.08, F change (2, 96) = 4.51, p = 0.013.Of the two parental attachment style variables, only one, attachment-related anxiety was statistically significant, t(96) = −2.96,p = 0.004.This variable made an independent contribution to the model (beta = −0.29),with the model predicting a 0.58 unit decrease in parent-child relationship quality for every one unit increase in parental attachment anxiety if the effects of all other variables are held constant.Neither attachment-related avoidance, t(96) = 0.90, p = 0.369 nor socio-economic status, t(96) = −1.83,p = 0.070 were statistically significant.Independent analysis of the child temperament variables also revealed a statistically significant result for the overall model (SES, effortful control, negative affectivity, surgency, and affiliativeness), with the model explaining 49% of the variance in parent-child relationship quality, R 2 = 0.49, F(5, 94) = 17.77, p < 0.001, adjusted R 2 = 0.46 (see Table 4).After controlling for SES, child temperament explained an additional 47% of the variance in parent-child relationship quality, R 2 change = 0.47, F change (4, 94) = 21.38,p < 0.001.Of the four child temperament variables, two were statistically significant.Negative affect, t(94) = −6.07,p < 0.001) made the strongest independent contribution to the model (beta = −0.56),with the model predicting a 0.40 unit decrease in parent-child relationship quality for every one unit increase in child negative affect if the effects of all other variables are held constant.Affiliativeness, t(94) = 2.45, p = 0.016 also independently contributed to the model (beta = 0.21), with the model predicting a 0.44 unit increase in parent-child relationship quality for every one unit increase in child affiliation if the effects of all other variables are held constant.Additionally, socio-economic status, t(94) = −2.38,p = 0.019, contributed to the model (beta = −0.18),with the model predicting a 2.93 unit decrease in parent-child relationship quality for every one unit increase in SES.
Overall, the full model of SES, parental attachment style, and child temperament as predictors of parent-child relationship quality during early adolescence was statistically significant, explaining 50% of the variance, R 2 = 0.50, F(7, 92) = 12.92, p < 0.001, adjusted R 2 = 0.46 (see Table 5 for full details on each regression model).Model 1, which contained the covariate SES, was not statistically significant F(1, 98) = 1.82, p = 0.180.However, the addition of parental attachment style and child temperament variables explained an additional 48% of the variance in parent-child relationship quality, after controlling for SES, R 2 change = 0.48, F change (6, 92) = 14.52, p < 0.001.In Model 2, only three variables were statistically significant.Negative affect, t(92) = −5.66,p < 0.001, made the strongest independent contribution to the model (beta = −0.57),with the model predicting a 0.40 unit decrease in parent-child relationship quality for every one unit increase in child negative affect if the effects of all other variables are held constant.Affiliativeness, t(92) = 2.29, p = 0.024 also independently contributed to the model (beta = 0.20), with the model predicting a 0.42 unit increase in parent-child relationship quality for every one unit increase in child affiliativeness if the effects of all other variables are held constant.Additionally, socio-economic status, t(92) = −2.37,p = 0.020) contributed to the model (beta = −0.18),with the model predicting a 2.93 unit decrease in parent-child relationship quality for every one unit increase in SES.The effect size (f 2 = 0.98) was large.
Moderation analyses were conducted using Hayes' (2017) PROCESS v3.5 macro to examine whether attachment-related anxiety or attachment-related avoidance

Discussion
The primary goal of the current study was to investigate whether parental attachment style and child temperament predicted parent-child relationship quality during early adolescence.A secondary aim was to assess whether parental attachment style moderated the relationship between child temperament and parent-child relationship quality.Independent examination of parent-attachment style revealed attachment-related anxiety but not attachmentrelated avoidance to be predictive of parent-child relationship quality after controlling for SES.Independent examination of child temperament found that two dimensions of this variable, negative affect and affiliativeness, predicted parent-child relationship quality while surgency and effortful control did not.When examining both parental attachment style and child temperament in the same model, child temperament was found to be uniquely predictive of parent-child relationship quality.Contrary to expectations, no moderation effects of parental attachment style on the relationship between child temperament and parent-child relationship quality during early adolescence were found.
The findings of the current study lend some support to the premise of attachment theory and the findings of a myriad of previous studies that link parental attachment style with relational outcomes, including parent-child relationship quality (e.g., Selcuk et al., 2010;Feeney, 2006).Specifically, the current study findings highlight the importance of parental attachment style, particularly attachment-related anxiety, in predicting the quality of relationships formed between parents and their early adolescent child.This is consistent with previous research by Feeney (2006) who found parental attachment security to be both directly and indirectly associated with how their children navigated relationships, including with their parents (i.e., parental attachment security was found to be directly associated with offspring's relationship anxiety and indirectly associated with offspring's loneliness and discomfort with closeness, which are key components of relationship quality).Contrary to expectations, parental attachmentrelated avoidance was not found to be predictive of parentchild relationship quality in the current study.Li et al. (2021) found that maternal anxiety negatively predicted the quality of relationship formed between mothers and their adolescent children, whilst maternal avoidance negatively predicted the quality of the relationship formed between fathers and their adolescent children.Given that most respondents in the current study were mothers, it is therefore not surprising that parental attachment-related anxiety predicted the quality of the relationship between these mothers and their adolescent children, whilst parental attachment-related avoidance did not.The findings would suggest that to examine the true impact of maternal attachment-related avoidance on parent-child relationship quality, the quality of relationships between these adolescents and their fathers would also need to be explored.
Two dimensions of temperament (negative affect and affiliativeness) were found to be predictive of parent-child relationship quality, thus offering support to temperament theorists in their assertions as to the influence of child temperament on parent-child relationship quality.Specifically, the current study findings highlight the important role played by individual differences in reactivity measures, with two of the three reactivity variables producing significant results.This is consistent with previous research by Chiah and Baharudin (2012), who found both negative and positive affectivity to be associated with parent-child relationship quality and overall adolescent social-emotional adjustment.However, while Chiah and Baharudin (2012) found the correlations between negative affectivity and difficulties, and positive affectivity and strengths, to be equivalent, analysis of the beta-coefficients in the current study would suggest that negative affectivity is the most predictive, whereas affiliativeness is less so.Considering the vast body of research highlighting the links between child negative affect and sub-optimal child outcomes, along with the role of parents in mediating the effects of temperament (Kiff et al., 2011), the findings of the current study thus further emphasise the importance of establishing and maintaining positive relationships between children and their parents, to successfully achieve optimal child outcomes during adolescence.Unlike negative affect and affiliativeness, the third reactivity measure (surgency), was not found to predict parent-child relationship quality.While this was unexpected given that surgency is generally thought to represent a component of positive affectivity, it perhaps offers further support to Snyder et al.'s (2015) findings which suggest that surgency is a separate construct from positive affectivity and needs to be viewed as an independent factor.Similarly, effortful control was not found to be predictive.This was also unexpected given that an association between child self-regulation abilities and other relational outcomes, including those with peers, has previously been documented (e.g., Morris et al., 2013).One possible explanation could be that the pathway between effortful control and parent-child relationship quality is more ambiguous than the pathway of some of the other temperament dimensions.This explanation receives some support from studies that have found that whilst effortful control has a genetic, temperamental basis (Goldsmith et al., 2008); it is also associated with and influenced by environmental factors such as parental sensitivity (Spinrad et al., 2012), highlighting its intricacy.Contrary to expectations, child temperament was found to be uniquely predictive of parent-child relationship quality when examining both parental attachment style and child temperament in the same model.This may suggest that child factors (i.e., temperament) are more salient than parental factors (i.e., attachment style) in predicting parentchild relationship quality during early adolescence.This suggestion receives some support from personality research which has found that youth personality plays a more dominant role than parental personality in determining the quality of relationships formed between parents and their adolescent children (Werneck et al., 2014).Although child temperament and youth personality traits are not equivalent constructs, there is evidence in the literature of their associations (Tackett et al., 2013).Similarly, associations have been reported between parental attachment style and adult personality traits (Abou-Amerrh et al., 2013).It may therefore be possible to infer that mechanisms similar to those that underscore parent-child relationships in the personality literature are also at play with parental attachment style and child temperament.If this is the case, rather than dismiss the influence of parental attachment style, the findings of the current study could suggest that whilst parental attachment style is still influential, child temperament plays a more dominant role in determining the quality of relationship that is formed between parents and their children during this specific developmental stage.It is plausible that with the shift towards increasing autonomy, independence, and identity formation, certain temperament traits may become more prominent during adolescence.This hypothesis is consistent with Zohar et al.'s (2019) findings, which identified a biphasic pattern during early adolescence with some traits increasing up until the age of 14 years and decreasing thereafter.This increase in trait presentation, in conjunction with increasing peer influences which are thought to supersede that of parents may explain why parental attachment style, thought to be expressed in how parents think, feel, and behave in their relationships (Jones et al., 2015), may not be as influential in determining parent-child relationship quality during early adolescence as it is during other key stages of childhood.
Another possible explanation could be that with the transmission of attachment patterns over generations, children who themselves become insecurely attached often mirror the attachment patterns of their parents (Bartholomew and Horowitz 1991), and though not optimal, lead perhaps to a more compatible relationship during early adolescence with neither of the dyad's attachment-related demands exceeding the attachment-related capabilities of the other.
Considering the extensive body of research, however, linking parental attachment style with parental responses (Berlin et al., 2011;Selcuk et al., 2010), and parental responses with parent-child relationship quality at every developmental stage (Zeegers et al., 2017), alternative explanations for the findings of the current study must also be considered.Given that the same person reported on all measures, it is conceivable that a mono-reporter bias impacted which predictors were uniquely predictive in the context of the full model.For instance, the quality of daily interactions between parent and child in the days/weeks leading up to survey completion could have influenced how parents perceived and consequently reported on their child's temperament and, potentially, their own behaviours related to attachment.It is also possible that one or more of the variables impacted the predictably of the others.An example of this can be seen in the case of SES, which was controlled in the current study using a measure of parental education level.Although this was not independently predictive when entering the model initially, it did become significantly predictive with the addition of the parental attachment style and child temperament variables.One possible explanation for this finding is that when the parental attachment style and child temperament variables were added to the model containing SES, suppression effects occurred, removing outcome-irrelevant variance, and revealing the true relationship between SES, parental attachment style, child temperament, and parent-child relationships.Because the operation of suppression in regression models with multiple variables is complex, further exploration of the validity of this hypothesis is outside of the scope of this paper.Some evidence of reciprocal suppression is suggested, however, in the fact that negative marginal correlations were observed between SES and both parent-child relationship quality and parental attachmentrelated anxiety, and a strong negative correlation was observed between parental attachment-related anxiety and parent-child relationship quality.Additionally, the fact that partial regression coefficients of some of the variables were found to be greater than their simple regression coefficients with no sign reversal would also indicate potential reciprocal suppressor effects, and may explain why parental attachment-related anxiety was no longer significant in the full model.
Contrary to expectations, parental attachment style was not found to moderate the relationship between child temperament and parent-child relationship quality during early adolescence.This is at odds with the Goodness of Fit Model and previous studies that have shown that children can develop different quality relationships with mothers, fathers, and other caregivers (Goossens and van IJzendoorn, 1990), suggesting that parents can modulate the effects of their children's temperament by influencing their environment.A key message in the literature appears to be that the influence of children's temperament (or other attributes) on parent-child relationship quality may be mitigated if parents can adjust their caregiving behaviours to better fit the needs of the child (Hong and Park 2012).In the case where a parent's capacity to modify their behaviour is limited, however, due to either external (e.g., stress) or internal (e.g., internal working models, personality) factors, then children with more difficult temperament traits are at risk of developing poorer quality relationships with their parents.The lack of association found in the current study could be purely methodological; power within the models may have been too low to detect these relations.Alternatively, it is also conceivable that the constructs of parental attachment style, child temperament, and parent-child relationship quality might not always be analogous, particularly within the contexts of the present study.To fully capture the transactional and dynamic nature of goodness of fit in the parent-child relationship, more precise models and expanded measures of parental attachment style, child temperament, and parent-child relationship quality could offer further insight.
Overall, the findings of the current study show that in accordance with temperament theory, child temperament is the strongest predictor of parent-child relationship quality during early adolescence.Contrary to expectations, this study questions the utility of parental attachment style in predicting relationship quality, suggesting instead a decline in parental influence during this developmental stage.However, given that parental attachment-related anxiety was independently predictive of parent-child relationship quality, rather than dismissing the predictive utility of parental attachment style, other reasons for the results need to be considered.The findings of this study tentatively suggest therefore that while temperament is the most important predictor of parent-child relationship quality during early adolescence, the influence of parental attachment style, though diminished, cannot be ruled out, thus highlighting the importance of careful consideration of the interplay of both parent and child factors in the study of parent-child relationships.
The current study has several strengths.While much of the literature to date incorporating adult attachment styles pertain to parents of young children, the current study extended this research by focusing on parent-child relationships during early adolescence, which is often a challenging developmental stage for both members of the dyad.Moreover, it incorporated parent and child factors, focusing on the predictive utility of enduring characteristics of the parent and the child (i.e., parental attachment style and child temperament), an approach which is believed to better explain individual differences in patterns of adolescent adjustment than focusing on the changes in response to maturation of the child (Laursen and Collins, 2009).However, this study also had several limitations.The final sample was relatively homogeneous, which may explain the large effect size obtained.Research utilising more diverse samples would allow for additional insights into factors affecting parent-child relationship quality during early adolescence.It is also important to acknowledge that although both self-report and interview-based measures of attachment style are reliably associated with various facets of parenting (Jones et al., 2015), they are in no way the same.In fact, studies have indicated that some aspects of parenting are more strongly predicted by one measure rather than the other or a combination of both (Scharf and Mayseless, 2011).Including only a self-report measure of parental attachment style in this study may therefore have affected its predictive ability.This is particularly salient since the full range of potential emotions, cognitions, and behaviour that parents may express, particularly in interaction with their children, are difficult to access using questionnaires alone.Future research should therefore include both self-report and interview-based measures of parental attachment style to comprehensively tap into this construct.Also, because the current study aimed to tap into participants' global attachment representations in close relationships rather than relationship-specific attachment representations, a broad measure of global parental attachment style was deemed to be the most suitable option.However, given that relationship-specific measures of attachment have been found to be more predictive of intra and interpersonal outcomes (Fraley et al., 2011), employing relationship-specific measures of parental attachment style in future studies may offer better insights into the predictive ability of parental attachment style on parent-child relationships.
While parent reports of child temperament have been found to be acceptably convergent with adolescent reports (Ellis, 2002), the reliance on parent reports alone in this study may have served as a limiting factor and impacted not only the generalisability of the findings, but also which predictors were uniquely predictive in the context of the full model.To understand the true nature of child temperament, including both parent and adolescent reports in future studies would be beneficial.This also applies to parent-child relationship quality, as reports from each member of the dyad could offer a more comprehensive depiction of their relationship quality.Future studies should therefore examine the constructs using multiinformant measures, including observational measures where possible, which would also serve to reduce the social desirability risks associated with self-report measures.The relatively small sample size may also have disadvantaged this study and identifies the need to treat the findings with caution.Further investigation using much larger samples is warranted to establish if and how the constructs impact parent-child relationship quality as children advance through childhood and beyond.This would allow for the establishment of appropriate community-based prevention and intervention strategies aimed at supporting parent-child relationship quality, whilst recognising the potential impact of parent and child factors at each stage of development.Understanding the interplay of parental attachment style and child temperament in predicting parent-child relationship quality would also have important implications for fostering and adoption agencies seeking to ensure appropriate matching of placements to ensure goodness-of-fit (Chess and Thomas, 1999), particularly since one in four long-term placements have been found to break down during adolescence (Vinnerljung et al., 2017).
Overall, this study shows that the prediction of parentchild relationship quality during early adolescence is complex and is most likely influenced by both child and parentrelated factors.Child temperament appears to be particularly influential, underscoring the need to account for temperamental factors when developing programs aimed at promoting and improving parent-child relationship quality.Although parental attachment style was not directly predictive in the current study, suggesting perhaps that parental influence declines during early adolescence, parents still have a role in influencing relationship quality with their early adolescent children.These findings are important in ensuring that optimal child and adolescent outcomes can be successfully achieved through the development of interventions to promote and improve parent-child relationship quality at each developmental stage.

Table 1
Means (M), standard deviations (SD), Min/Max, and range of possible scores for study variables

Table 2
Correlations between study variables

Table 4
Hierarchical multiple regression predicting parent-child relationship quality from SES and child temperament