Research on the etiology of adult psychopathology and its relationship with childhood trauma has focused primarily on specific forms of maltreatment. This study developed an instrument for the assessment of childhood and adolescence trauma that would aid in identifying the role of co-occurring childhood stressors and chronic adverse conditions. The Complex Trauma Questionnaire (ComplexTQ), in both clinician and self-report versions, is a measure for the assessment of multi-type maltreatment: physical, psychological, and sexual abuse; physical and emotional neglect as well as other traumatic experiences, such rejection, role reversal, witnessing domestic violence, separations, and losses. The four-point Likert scale allows to specifically indicate with which caregiver the traumatic experience has occurred. A total of 229 participants, a sample of 79 nonclinical and that of 150 high-risk and clinical participants, were assessed with the ComplexTQ clinician version applied to Adult Attachment Interview (AAI) transcripts. Initial analyses indicate acceptable inter-rater reliability. A good fit to a 6-factor model regarding the experience with the mother and to a 5-factor model with the experience with the father was obtained; the internal consistency of factors derived was good. Convergent validity was provided with the AAI scales. ComplexTQ factors discriminated normative from high-risk and clinical samples. The findings suggest a promising, reliable, and valid measurement of early relational trauma that is reported; furthermore, it is easy to complete and is useful for both research and clinical practice.
Over the past two decades, evidence regarding the harmful impact of early maltreatment has accumulated significantly. Two-thirds of adolescents and adults report having suffered from child relational trauma, which represents a root cause of major public health issues (Copeland et al., 2007; van der Kolk and d’Andrea,2010). Previous research has primarily focused on single types of maltreatment sequelae; however, most maltreated children experience more than one form of abuse and neglect (Kessler, 2000; Spinazzola et al.,2005; Pynoos et al., 2008). Indeed, in literature on child abuse, the focus has shifted from individual and specific conditions of risk to a multi-determined etiology of the traumatic experience: initially, the literature focused on a punctiform trauma but now investigates the concept of complex trauma, describing a traumatic developmental context in which the child is immersed. Complex trauma can be understood as experiences of cumulative, chronic, and prolonged traumatic events, most often of an interpersonal nature, involving primary caregiving system, and frequently arising in early childhood or adolescence (Cook et al., 2005; Courtois,2008). Till date, growing evidence has acknowledged the co-occurrence of multiple types of severe adversities (Mullen et al., 1996; Higgins and McCabe, 2001; Diaz et al., 2002; Clemmons et al., 2003; Dong et al., 2004; Stevens et al., 2005; Arata et al., 2007; Finkelhor et al., 2007, 2009; Turner et al., 2010; Greeson et al., 2011; Trickett et al., 2011) and their greater risk for subsequent trauma exposure and cumulative clinical impairment compared with singly traumatized youth (Schumm et al., 2006; Finkelhor et al., 2007,2009; Cloitre et al., 2009; Margolin et al., 2009; Shen, 2009; Heim et al., 2010). However, numerous studies highlight the additive effect of child and adolescent multi-type maltreatment on later symptom complexity and psychopathology, including internalizing (Danielson et al., 2005; Schumm et al., 2006; Anda et al., 2007; Sachs-Ericsson et al., 2007; Widom et al., 2007; Ford et al., 2010), externalizing (Brown and Anderson,1991; Herrenkohl et al., 1997; Finkelhor et al., 2009; Ford et al., 2009, 2010; Shen, 2009), and trauma symptoms (Boney-McCoy and Finkelhor, 1996; Mulder et al., 1998; Schaaf and McCanne, 1998; Finkelhor et al., 2007, 2009; Vranceanu et al., 2007; Shen, 2009; Ford et al., 2010). Following this large amount of studies, it is understandable that trauma may be referred not only as a present-or-absent construct but also includes dimensional aspects, considering the multiplicity of maltreatment forms observed as well as the frequency of traumatic exposure.
Literature on attachment has generated stimulating outcomes regarding early traumatic experiences and how they shape later responses in adulthood, becoming a framework that enables a more sophisticated comprehension of the relational trauma sequelae on mental health (Lyons-Ruth and Jacobvitz, 2008). Occurrence of abuse and neglect in early life may interfere with the development of a healthy secure attachment relationship (Baer and Martinez, 2006). Research findings suggest that traumatized children are at risk for attachment disruption, specifically developing an insecure disorganized attachment pattern (Lyons-Ruth et al., 1987; Carlson et al., 1989; Cicchetti and Barnett, 1991; Beeghly and Cicchetti, 1994; Barnett et al., 1999; van IJzendoorn et al., 1999; Cicchetti et al., 2006; Cyr et al., 2010). Similarly, during the administration of the Adult Attachment Interview (AAI; George et al., 1984; Main et al., 2003), the surfacing of memories of attachment-related traumatic experiences is not rare. In such cases, disorganized attachment pattern is most typically detected in adult respondents with a history of complex trauma (Lyons-Ruth and Jacobvitz, 2008; Bakermans-Kranenburg and van IJzendoorn, 2009; Murphy et al., 2014).
The assessment of complex trauma is by definition “complex,” as it involves both a delineation of a wide range of traumatic events as well as their appraisal, and because these types of experiences rarely occur in isolation and are highly interrelated (Felitti et al., 1998; U.S. Department of Health and Human Services, Administration on Children, Youth, and Families, 2011). Ascertaining the presence and degree of early maltreatment required for clinically based studies and epidemiological research has been enabled by observer-rated interviews and self-report questionnaires, though retrospective reports usually provide underestimates of the trauma incidence (Hardt and Rutter, 2004). Self-reports generally inquire about limited types and generate quantitative rating of trauma, and are unable to detect when defense mechanisms may distort responses (Ravitz et al., 2010). However, this format requires less time to complete and may elicit more disclosure facilitating valid responses to questions on sensitive issues (Newman et al., 1996). Several guided or semi-structured interviews have been developed to measure a broader range of potential trauma areas to reduce response biases (Roy and Perry, 2004), allowing investigators to obtain uniform information in an interpersonal interaction context. Nevertheless, these instruments are usually time-consuming and require administration and scoring by trained investigators.
Given the pervasiveness of the trauma reported and the outcomes that arise, these findings emphasize the vitality of a comprehensive assessment of early trauma history for making appropriate service recommendations and interventions within child and adolescent welfare (Kisiel et al., 2009; Briggs et al.,2013). Despite several measures that have been created to assess the occurrence of childhood trauma, researchers often base assessment tools on a narrowed conceptualization of maltreatment and focus on creating a single or limited abuse types-instrument (Bremner et al., 2007; DiLillo et al., 2010), making it difficult to extend the investigation of the multiple aspects of traumatic experience and to compare the impact of specific forms of trauma. Moreover, most studies use a priori categorical or dimensional descriptions of complex trauma, while only few define polyvictimization empirically using statistical techniques. Lastly, although evidence regarding the nature and frequency of traumatic exposure is significant to research and practice on child maltreatment (Manly et al., 2001; English et al., 2005), many instruments restrict assessment to the occurrence of trauma in a dichotomous (presence vs. absence) classification. In response to these gaps in measurement, this study aimed to develop a retrospective questionnaire concerning early history of interpersonal trauma and to outline its preliminary psychometric characteristics by presenting data which support its reliability and validity. In this paper, the Complex Trauma Questionnaire (ComplexTQ) was particularly applied to AAI transcripts conceived as a stimulus which would enable activation of the attachment system at emotional levels through an intensive series of probes regarding the interviewee’s history with the attachment figures (Dozier and Kobak, 1992; Dias et al., 2011; Farina et al., 2014).
Materials and methods
Development and features of the Complex Trauma Questionnaire
The Complex Trauma Questionnaire (ComplexTQ) was developed to enable researchers and clinicians to measure adverse childhood experiences displayed in their major forms and frequency of occurrence, to cover multidimensional aspects of complex trauma, including child neglect which represents one of the most widespread forms of maltreatment (Gilbert et al., 2009; U.S. Department of Health and Human Services, Administration on Children, Youth, and Families, 2011), which appears as a deceitful phenomenon and is less easily detectable and assessable compared with active types of abuse.
The ComplexTQ was designed in both clinician and self-report versions1. In this study, the questionnaire is completed by a clinician and is applied to interview transcripts for evaluation of trauma history prior to age 15. Item construction was based on clinical experience and performed following an extensive review of the literature on child abuse and neglect. Clinician responds to a series of specific screener questions that reveal whether the participant experienced a range of caregiver’s behaviors encompassing acts of both commission and omission. To capture the intensity of these experiences, clinician provides ratings on a 4-point Likert scale (from 1 = never to 4 = often) reflecting how frequently the episodes occurred, except for the last item concerning the occurrence of significant bereavements (yes/no). Thus, scoring allows detecting presence and severity (sum and mean of frequency scores) of each maltreatment type. Cumulative occurrence across all single forms of traumatic experience is provided through a total score. Given research and theory suggesting that the caregiver gender may be relevant to trauma effects (Schore, 1999; Briere and Rickards, 2007), scores are differentiated between relationship with the mother, that with the father as well as with another significant figure.
Based on the literature on childhood trauma, nine domains of interpersonal maltreatment experiences were considered and operationalized. Items regarding the Neglect domain are intended to measure withdrawal of caretaking and failure to respond and engage in behavior that is necessary to meet the developmental needs of a child, such as physical needs (e.g., providing food, clothing, and medical treatment), supervision and educational needs (e.g., checking on friendship environment and leisure activities, monitoring school attendance, or assisting with homework), and emotional needs (e.g., demonstrating affection, support, and companionship). These conducts attempt to capture the heterogeneity of child neglect, ranging from a caregiver who is unavailable, inattentive, uninvolved, or psychologically inaccessible (e.g., lacking of love expressions) to the one whose acts involve abandonment of the child. Consequently, the child may feel unloved, unnoticed, or invisible to the caregiver. Reject items describe a cold or hostile caregiving (e.g., episodes of child’s distress and illness are minimized or ridiculed) as an active turning away of the child’s expression of need and attachment. The caregiver seems to prefer the child’s absence by acting on feelings of anger, enmity, irritation, bitterness, or resentment. The child may feel disliked, avoided, and unwanted. Items referring to Role reversal behaviors label caregiver who seeks comfort, intimacy, and physical or psychological unmet needs from the child. Child’s presence and attention seem essential for the caregiver’s sense of welfare. The child is thus expected to take the role of a parent, spouse, or peer toward the caregiver and attempts to reduce his/her distress. Items of Psychological abuse assess verbal and demonstrative acts by the caregiver intended to cause psychological pain or fear, as excessive and repeated ridiculing, blaming, insults, isolation, threats, and coercive attitudes toward the child. The caregiver’s intent seems the control or intimidation of the child, and items range from shaming or embarrassing the child to parental alienation syndrome, intrusive and excessive medical care, or emotional cruelties (e.g., threat to hurt the child or the child’s loved one or to force him/her to violate the law). Questions regarding Physical abuse inquire about the occurrence of physically aggressive behaviors (e.g., hitting, grabbing, knocking down, beating with an object, tying, and burning), resulting injuries (e.g., cuts, bruises, broken bones, and burns), and medical treatment needed. Supplementary queries probe regarding the nonabusive caregiver’s responses to maltreatment perpetrated by other household members (e.g., defending, reassuring, and comforting the child). The Sexual abuse domain includes any sexual intercourse that involves (e.g., inappropriate touching, kissing, fondling, and penetration) or does not involve (e.g., voyeurism, exposure to erotic language, and participation in pornographic activities) physical contact. In addition, items explore if the participant has been threatened not to recount, if the caregiver did not believe the report, or failed to comfort and protect the child from abuse. Items regarding the Witnessing domestic violence area describe verbal/emotional abuse and aggressive behaviors that occur between parents to which the child is exposed. Finally, significant Separations andLosses up to age 14 are explored.
The study has been conducted among 229 individuals, including participants at high-risk and with psychiatric diagnoses (dissociative disorders, N = 14; parents of maltreated children, N = 53; gender identity disorder in the procedure of sex reassignment surgery, N = 41; personality disorders, N = 42), and a nonclinical sample (N = 79). Subjects were recruited from clinical centers in Rome (Italy) and took part in different research projects (Mirizio et al., 2011; Farina et al., 2014; Speranza and Maggiora Vergano, 2015; Lingiardi et al., submitted). Although important differences characterize participants within the clinical/at-risk sample, we decided to combine those groups due to their history of trauma exposure as suggested by literature. Indeed, several studies highlighted early trauma in dissociative patients (Dutra et al., 2009; Schmahl et al., 2010), parents of maltreated children (Dixon et al., 2005a,b, 2009), participants with gender identity disorder (Kersting et al., 2003; Gehring and Knudson, 2005; Veale et al., 2010), and those with personality disorders (Afifi et al., 2011). The combined clinical/at-risk sample (N = 150) was composed of 37 men, 72 women, 27 male-to-female and 15 female-to-male transsexuals, ranging from 16 to 62 years of age (M = 30.44; SD = 11.24), and mostly Italians (94%). The majority of nonclinical participants were females (82%) and Italian (99%) with the mean age of 31.87 years (SD = 5.76) and age range of 25–65 years. The nonclinical participants were recruited from previous longitudinal studies regarding the quality of parenting (Ammaniti et al., 2005, 2006), representing the control group which had no depressive or psychosocial risks.
The overall sample was recruited from different research studies expressively asking for a participation secured through a written informed consent procedure that required active consent from participants. The current study has been attended within the Ph.D. context at the Department of Dynamic and Clinical Psychology (Faculty of Medicine and Psychology, Sapienza University of Rome) and received the approval from the Ethical Committee of the Department.
Adult attachment interview
Adult Attachment Interview (AAI; George et al., 1984; Main et al., 2003). The AAI is a semi-structured interview, an hour-long protocol, which explores adult’s mental representations of attachment while discussing childhood experiences. After an overview of the family composition, respondents are asked to describe their early relationship with each parent, supplying five adjectives to be supported by providing specific descriptive incidents. Interviewees are additionally probed regarding situations of distress (e.g., emotionally upset, hurt, or ill), and instances of threat or abuse. Following questions concern reaction to separations from the caregiver and significant losses. Two types of variables are rated on 9-point Likert scales. Narratives of parent–child relationships are coded on scales for “probable childhood experiences” which describe rejecting, involving/role reversing, neglecting when present, pressuring to achieve, and loving behaviors of each attachment figure. Likewise, “present state of mind” with respect to attachment is rated on scales defining idealization of parents, dismissing derogation of attachment figures or relationships, insistence on lack of recall, fear of loss of the child, passivity or vagueness in discourse, current involving/preoccupying anger toward caregivers, metacognitive monitoring, coherence of transcript, and overall coherence of mind. AAI scoring system is based on the participant’s ability to produce coherent narratives regarding childhood experiences with caregivers, thus classifying interviewee as Secure/Autonomous (F), Dismissing (Ds), Preoccupied with respect to attachment (E), or “Cannot classify” category (CC) when a global breakdown in the organization of discourse arises. An interview may also be assigned an Unresolved/disorganized state of mind (Ud) concerning past abuse or loss in association with a best-fitting primary classification. Psychometric testing and meta-analyses have provided evidence of stability and discriminant and predictive validity of the AAI in both clinical and nonclinical populations (Bakermans-Kranenburg and van IJzendoorn, 1993, 2009; Sagi et al., 1994; van IJzendoorn, 1995; Roisman et al., 2007; Hesse, 2008; van IJzendoorn and Bakermans-Kranenburg, 2008). Each Adult Attachment Interview was conducted by researchers trained by A.M. Speranza to administer the AAI, recorded and transcribed verbatim. The interviews were classified using Main et al. (2003) coding system2 by the first and the third author who are certified AAI coders. A hundred and thirty-five transcripts (59%) were double-coded, blind to subject conditions: percent agreement for the 4-way classification (F, Ds, E, Ud/CC) was 81% (k = 0.77). When there was disagreement, a third independent coder also categorized the transcript, with final agreement reached after discussion by all three coders.
Complex Trauma Questionnaire
Complex Trauma Questionnaire (ComplexTQ). The ComplexTQ is a 70-item scale for the retrospective assessment of multi-type maltreatment, measuring lack of care (physical and emotional neglect), abuse (psychological, physical, and sexual abuse), and other traumatic experiences, such as rejection, role reversal, exposure to domestic violence, separations, and losses. The instrument is available in two different versions, clinician and self-report. The questionnaire assesses adverse experiences from childhood to usage of 14 years separately involving maternal, paternal, and other attachment figures. The clinician version, compiled by the first and the third author, requires approximately 15–20 min to complete (depending on the length of the interview’s transcript) and scores for presence and frequency of traumatic experiences in each domain are automatically provided by the software.
Although the data used in the present study were collected by a single rater who evaluated all transcripts, we conducted a preliminary analysis aimed to test rater’s ability to code data. This analysis was based on a pilot study of 54 participants later included in this study, each evaluated by the first and the third author of the manuscript, blind to each other, using the complete 70-item set. The inter-rater reliability for ComplexTQ ratings was assessed based on the Intraclass Correlation Coefficient (ICC), a measure of similarity of ComplexTQ ratings made by independent coders. The analysis yielded good inter-rater reliability for the 70 items (ICC-s = 0.85 and 0.89 for items concerning mother and father, respectively) as well as for the items retained after factor analysis (ICC-s = 0.81 and 0.87 for items concerning mother and father, respectively).
A preliminary descriptive analyses of ComplexTQ items revealed that 27 and 30 items, for patients’ traumatic experience with mother and father, respectively, had near zero variance in the samples studied. These items were excluded from the analyses that were completed on 40 and 37 items. However, the retained items not only were ordered categorical variables but the inspection of the item response pattern also revealed in most cases an asymmetric response distribution. Standard factor analysis assumes a multivariate normal distribution and item distributions that approach an equal intervals scale. As both the assumptions were violated, we analyzed the polychoric correlation matrix, instead of the standard Pearson’s correlation matrix. This procedure is strongly recommended for ordered categorical items from two to five response categories (Panter et al., 1997; Holgado-Tello et al., 2008; Timmerman and Lorenzo-Seva, 2011; Garrido et al., 2013). Specifically, a polychoric correlation is based on the assumption that each of the ordered categorical items represents an approximation of an underlying continuous variable. Accordingly, the computational procedure estimates what would be the correlation between these underlying variables based on the collected empirical approximation. An unweighted least square factor analysis with promax rotation was conducted by Factor 9.3 (Lorenzo-Seva and Ferrando, 2006). The Kaiser–Meyer–Olkin measure of sampling adequacy attained a fairly high value for both mother and father datasets (KMO = 0.72 and 0.80, respectively), thus showing that the sample correlation matrix was appropriate for factor analysis to proceed.
Factor retention criteria
Despite there being no “gold standard” for determining the “true” number of factors to retain in exploratory factor analysis, we integrated the scree-plot with information from Minimum Average Partial (MAP) and Parallel Analysis (PA), which are considered the most accurate diagnostics (Ruscio and Roche, 2012). The MAP test is based on a series of factor analyses, each followed up by a quantitative assessment of partial correlations. Specifically, as the first step, a factor analysis is performed and the first factor is partialed out of the correlation matrix among questionnaire items. Then, the squared off-diagonals partial correlations are averaged to obtain a summary index of the variance in the correlation matrix that is due to systematic sources. As the second step, both first and second extracted factors are partialed out and again the squared off-diagonals partial correlations are averaged, and this procedure is repeated up to k-1 factors are partialed out. The MAP test lines up the average squared partial correlation indexes obtained on each step, and the minimum value is considered as indication of the appropriate number of factors in the dataset. PA is based on comparing eigenvalues resulting from a factor analysis of real data to those resulting from a Monte Carlo simulation study. Specifically, repeated factor analyses are performed on a large number of randomly generated data matrices (e.g., 1000) that are parallel to the real data (i.e., have the same number of variables and cases). Factors whose eigenvalues are greater than the average random ones are deemed as reflecting systematic sources and therefore, retained as meaningful psychological dimensions (for a recent review of factor retention criteria see Courtney, 2013).
ROC curve analysis
The clinical validity of the ComplexTQ was assessed by plotting the performance of each factor’s derived scale score as a classifier of clinical vs. nonclinical group membership (see participants). A Receiver Operating Characteristic (ROC) curve analysis of these plots was performed by SPSS 23 to determine the optimal threshold for ComplexTQ. On each analysis, the “true positive” rate at different threshold is plotted against the “false positive” rate and the resulting curve connecting data points was drawn. The area under the curve (AUC) is a measure of the scale score classification accuracy, and its statistical significance supports clinical validity.
The transcripts were also coded through AAI scales for inferred experiences in infancy with mother and father. Convergent validity of ComplexTQ scales was then assessed as correlation coefficients with AAI scales.
The analysis of ComplexTQ items yielded 16 and 15 eigenvalues that were greater than one for mother and father datasets, respectively. Regarding mother dataset, the scree-plot showed three major drops in eigenvalue size after the third, the sixth, and the eighth eigenvalue. The MAP test suggested two factors to be retained, while the PA criterion indicated six factors. Regarding father dataset, the scree-plot showed two major drops in eigenvalue size after the second and the fifth eigenvalue. The MAP test suggested three factors to be retained, while the PA criterion indicated five factors. Thus, there is agreement between scree-plot and PA on six and five factors for mother and father, respectively. Besides that, the six- and five-factor solutions also had a better fit than solutions based on a smaller number of factors (GFI-s = 0.94 and 0.95 for mother and father datasets, respectively).
The mother dataset showed the following factors accounting for 49% of the variance in trauma experience (see Table Table1).1). (F1) Role Reversal loaded on eight items evaluating the extent to which the mother placed herself in the center of child’s attention, involving the child in her physical and/or psychological care, or making the child feel responsible for her own wellbeing. Moreover, other items inform about situations of domestic conflict and violence witnessed by the child. (F2) Physical Abuse loaded on seven items describing forms of maltreatment and physical abuse perpetrated by the mother. Furthermore, the caregiver tended to deride or get angry at the child’s signals of attachment needs, arousing child’s fear and worries. (F3) Psychological Abuse/Rejection loaded on eight items assessing rejection and avoidance of the child, especially child’s expression of affection, attention, dependence, and attachment. The mother’s abuse and aversion, exhibited through criticism, insults, or humiliations, made the child feel unwanted and disliked. (F4) Emotional Neglect loaded on seven items describing indifference, inattention, or psychological unavailability to the child’s emotional needs. It includes failure to provide nurturance or companionship within child–caregiver relationship and lack of emotional support to the child when expecting comfort and reassurance. Therefore, the child may not rely upon the mother for sharing difficult or painful experiences. (F5) Failure of Protection loaded on three items regarding a parent who abdicated the role of caregiving, by failing to defend and reassure the child in abusive situations perpetrated by other caretaking figures. (F6) Material Neglect loaded on seven items considering physical neglect (e.g., failure to provide adequate food, clothing, shelter, and medical care), educational neglect (e.g., failure to ensure proper education and learning opportunities), and supervisory/social neglect (e.g., unawareness/inattention to child’s whereabouts and socializing).
The father dataset showed similar factors that emerged, accounting for 51% of the variance in trauma experience (see Table Table2).2). Unlike the mother dataset, no role-reversal factor was identified, the factor ordering was different and the factors loaded on different number of items but preserved the original meaning: (F1) Emotional Neglect (9 items), (F2) Failure of Protection (3 items), (F3) Material Neglect (7 items), (F4) Psychological Abuse/Rejection (7 items), and (F5) Physical Abuse (11 items).
ComplexTQ factor-derived scales
Factor markers with factor loadings greater than 0.20 were selected based on inspection of the factor loading matrix. Our aim was to retain most representative items for each factor and to discard items that failed to load any factors. Then, selected items were resubmitted to factor analysis to verify whether item selection biased factor interpretation. Since no substantial changes in factor labeling and content emerged, ComplexTQ factor-derived scales were based on 33 and 29 items respectively for mother and father datasets (reported in bold font on Tables Tables1,1, ,22).
As one can see from Table Table3,3, ComplexTQ factor-derived scales were used as predictors of group membership and discriminant properties were assessed as AUC-s. For mother and father data, ComplexTQ scores were significantly different for clinical and nonclinical participants. In particular, fairly high AUC values were found for the Emotional Neglect scale for both the mother and the father, while Material Neglect attained the fair AUC threshold for father only. It is worth noting that AUC values for the total scores approached the standard for good diagnostic accuracy.
As shown in Table Table4,4, the mother’s ComplexTQ factors were correlated with AAI scales for inferred experiences. AAI “Pressure to achieve” probable experience scale was omitted as no correlations were found with ComplexTQ factor-derived scales. We reviewed correlation coefficients following Cohen’s (1988) “rule of thumb.” There was a very large overlapping between ComplexTQ Role Reversal factor and AAI Involving/role reversing scale. Physical Abuse factor showed a medium-large correlation with AAI Rejecting scale. Psychological Abuse factor showed a large and small-medium correlation with AAI Rejecting and Neglecting scales, respectively. Emotional Neglect factor showed a very large correlation with AAI Neglecting and Rejecting scales. Failure of Protection factor showed a small-medium correlation with AAI Rejecting and Neglecting scales. Material Neglect showed a large correlation with AAI Rejecting and Neglecting scales. It is noteworthy that all ComplexTQ factors were negatively associated with the AAI Loving scale, although the effect size varied from small (Role Reversal) to very large (Emotional Neglect).
The father’s ComplexTQ factors (see Table Table5)5) had similar correlation patterns as mother for Neglect and Psychological Abuse factors. Instead, Failure of Protection factor was significant only with AAI Rejecting scale showing a small effect size, and Physical Abuse factor was medium-large with both AAI Neglecting and Rejecting scales. All ComplexTQ factors were negatively associated with the AAI Loving scale.