Development and evaluation of the Parenting to Reduce Child Anxiety and Depression Scale (PaRCADS): assessment of parental concordance with guidelines for the prevention of child anxiety and depression

Objectives and hypotheses

To assess parental concordance with recommendations in the Guidelines, we developed a new self-assessment parenting scale as a criterion-referenced measure. We chose to develop the scale to facilitate a criterion-referenced (CR) interpretation rather than a norm-referenced (NR) interpretation for several reasons (Glaser, 1994; Shrock & Coscarelli, 2007). In contrast to NR interpretations, in which an individual’s knowledge or skills are defined relative to that of others, CR interpretations define an individual’s knowledge or skills with reference to a pre-established set of specific objectives or domains of knowledge (the ‘criterion’). Accordingly, a CR measure would be composed of items based on specific instructional objectives or domains of knowledge and is suited to determining whether one is able to demonstrate a pre-defined level of knowledge or mastery of a skill (Hambleton & Rogers, 1991; Glaser, 1994). By employing a CR approach to measure parenting, we would be able to assess the knowledge and competencies a parent currently has, rather than whether a parent does as well as other parents (as per a NR measure). In addition, the areas that a parent could improve on are clearly defined by the criterion and can be targeted with specific education or intervention strategies. Changes in a parent’s parenting behaviours can then be assessed against the same criterion to provide an indication of progress that is specific to the individual parent.

In this paper, we describe the development and initial validation of the Parenting to Reduce Child Anxiety and Depression Scale (PaRCADS). To address the current gap in the measurement of parenting, we developed the PaRCADS to assesses the range of parenting risk and protective factors that are known to influence the development of anxiety and depression in children. Based on an assembly of modifiable parenting factors for childhood anxiety, depression and internalising problems, the PaRCADS was designed to be compatible with a transdiagnostic preventive approach. To evaluate the psychometric properties of the PaRCADS, concurrent studies were conducted with a sample of parent–child dyads and a panel of international experts. Given the similarity in some of the parenting practices tapped by the measures and the use of a common-method (i.e., parent self-report), we hypothesised that convergent validity for the PaRCADS would be demonstrated by: (a) moderate to strong correlations with other parent reports of parenting; and (b) moderate to strong correlation with parent report of general family functioning. Consistent with the multi-informant approach to assessment (De Los Reyes et al., 2013), we expected small correlations between child- and parent- reports of parenting. For discriminant validity, we hypothesised that the PaRCADS would have weaker correlations with parents’ health-related quality of life than with other parent-report measures of parenting, given that PaRCADS was expected to assess parenting practices rather than parental health-related quality of life. In addition, based on prior work on the association between parenting and child internalising symptoms (e.g., Yap & Jorm, 2015; Pinquart, 2017), we predicted small to moderate correlations between PaRCADS and child anxiety and depressive symptoms and child health-related quality of life.

Participants and procedures

Measures

Study 1 (parent–child dyad sample)

Study 2 (expert sample)

In the expert evaluation study, all participants were provided the following review materials: (1) a copy of the PaRCADS with the associated Guidelines and objectives each item purports to assess, and (2) a review form. For each PaRCADS item, experts were requested to provide a set of ratings and comments including two which were used in this study, namely, item-objective congruence (i.e., whether the item assesses the intended Guidelines), and item relevance (i.e., the extent to which the item is relevant to the Guidelines). Item-objective congruence was rated on a dichotomous scale (Yes or No), and item relevance was rated on a 4-point scale from 1 (not relevant) to 4 (very relevant).

Data analytic approach

As the child report versions of the CRPBI-Acceptance subscale and the PCS were administered only to children 10 years or above, data on these parenting measures were available only for a sub-sample (n = 144). Results from a missing data analysis showed that about 11% of the participants in Study 1 had at least one missing value (missed 1–3 items). However, the overall proportion of missing data was less than 0.1% and the proportion of item-level missingness was less than 3% on all measures. To assess the pattern of missingness, Little’s Missing Completely at Random test (MCAR) was run. Little’s test suggested that item-level missing data can be reasonably assumed to be missing completely at random, χ²(5941) = 5984.82, p = .34. Missingness was further examined in relation to participant characteristics, such as age, gender, language spoken at home, education level and number of children, while controlling for child participation status and survey administration. At the respondent-level, younger children had more missing data (p = .01). At the dyad-level, parents with younger participating children (p = .01) had more missing data. Given that the proportion of missing data was very low, where information about imputation from the scale developers was not accessible, missing data were imputed by person mean of completed items within each subscale, to be consistent with the imputation method recommended for RCADS-25 and AQoL-8D. There were no missing data from the experts’ ratings in Study 2.

Item analysis of the PaRCADS was conducted using techniques suitable for criterion-referenced measures. The item difficulty index (also known as item easiness or facility) was computed for each item to review the proportion of respondents who answered the item correctly (i.e., in this case, number of respondents who obtained a score for concordance with the Guidelines). The corrected point-biserial correlation (r_pb) between an item score (scored 0 or 1) and total score was used as an item discrimination index to assess the extent to which performance on an item corresponds to overall performance in the expected direction (Shrock & Coscarelli, 2007). The corrected point-biserial correlation provides a more robust discrimination index as it removes the contribution of the item to the total score, and hence allows an examination of the impact of item scores on the total test (Millman & Green, 1989). To assess the quality of each item in distinguishing respondents who scored above the cut-point for concordance from those who scored below, the B-index for each item was also calculated (Brennan, 1972). The B-index was derived from the difference in item difficulty indices (see definition for item difficulty index) between respondents who scored above the cut-point and those who scored below the cut-point.

All item indices can have values ranging from −1 to +1. While there is no reference value for evaluating the B-index, a higher value maximally separates the masters from non-masters on a criterion-referenced test (Brown & Hudson, 2012). Some researchers have suggested removing items with a difficulty index or a discrimination index below or above certain values since these items do not maximise the information about differences among respondents (Henning, 1987; Kline, 2013). However, such items may be useful in their contribution to the measure or reflect critical attributes about the criterion (Popham & Husek, 1969). Moreover, item indices may differ across different samples of respondents or contexts (e.g., clinical vs community samples). Hence, where item indices fell below the recommended range, each item was inspected carefully with reference to the Guidelines and consideration of experts’ ratings and feedback from Study 2 before a decision was made to discard or retain the item.

Descriptive statistics were used to examine the frequencies of participants’ mental health history and parental concern about their child’s risk of developing anxiety or depression. Independent sample t tests were conducted to explore possible differences in demographic characteristics and total child anxiety and depressive symptoms between children who participated in the study with their parents and those who declined to participate.

To assess the reliability or consistency of the PaRCADS in classifying parents as concordant or non-concordant with the Guidelines, we calculated the agreement coefficient (p_o) for the subscale scores and total score using the formula and approximate values provided in Subkoviak (1988) for single test administration. Test consistency for criterion-referenced measures is sometimes referred to as dependability or consistency rather than reliability so as not to be confused with approaches employed in classical test theory (Subkoviak, 1988; Brown, 1990). Where an absolute standard is chosen (as with the cut-point for concordance in the PaRCADS), and the primary interest is to measure classification consistency based on a cut-point (i.e., consistency of the PaRCADS in classifying parents as concordant or non-concordant with the Guidelines), the threshold loss agreement approach and the agreement coefficient were deemed to be more appropriate than other agreement indices, such as the kappa coefficient (Berk, 1984).

As the PaRCADS was intentionally created to be a criterion-referenced measure, with each domain containing items that represent a set of skills or knowledge necessary to be concordant with a parenting domain in the Guidelines, we examined the correlations between the PaRCADS subscales scores and total score. In particular, we expected moderate to high correlations between the subscale scores and total score. Conventional factor analyses were not conducted as the items in each domain were not expected to uniformly represent a single factor in criterion-referenced measures. Construct validity was assessed by comparing the Pearson’s correlation coefficients of the relationships between the PaRCADS and other measures. To explore the relationships between PaRCADS total score, participant demographics, mental health characteristics, child anxiety and depression and child HRQoL, we conducted bivariate correlation analyses and t-tests where appropriate.

Study 1 Parent and child characteristics

Table S1 presents the descriptive statistics for parental concerns about their child’s risk of developing depression or anxiety and parent-reported mental health history. Most parents reported at least “a little” concern about their child’s risk of developing depression (84.8%) or anxiety (88.5%). Nearly two-thirds of parents reported having had a prior mental health diagnosis, with 17.7% of parents experiencing current mental health problems. About two-fifths of parents reported that their children had a mental health or behavioural diagnosis in the past. Slightly more than a third of the children were reported to be experiencing current mental health or behavioural problems that had been formally diagnosed.

Non-participating children (n = 13) were slightly younger than children who participated with their parents (n = 342), t(14.09) =  − 2.59, p = .02. Notably, non-participating children were also reported by their parents to have higher child anxiety and depressive symptoms on the RCADS Total symptom score, t(353) = 3.13, p < .01.

Study 2 Expert ratings

Item-objective congruence was summarised by the proportion of experts who endorsed the item as congruent with the intended Guidelines. Using Lynn’s (1986) guidelines, the item content validity index (I-CVI) was computed for each PaRCADS item to summarise experts’ ratings on item relevance. Adopting the cut-points recommended in the literature (Polit, Beck & Owen, 2007; Almanasreh, Moles & Chen, 2019), 76 out of 83 PaRCADS items were endorsed by at least five out of six experts for item-objective congruence, and 64 items had the highest ratings from the experts for item relevance. Details of item ratings are available in File S2.

The subscale content validity index (S-CVI) was also calculated by averaging the I-CVIs for item relevance across items within each subscale. Results suggested that all subscales in the PaRCADS displayed good subscale content validity (Davis, 1992; Polit, Beck & Owen, 2007), except Child’s relationship with others (S-CVI/Ave = .71) and Health habits (S-CVI/Ave = .77).

In addition to expert ratings, other item indices appropriate for criterion-referenced measures were computed for consideration as part of the item selection process.

Psychometric properties of the PaRCADS

Item analysis

Item statistics are available in File S2. All but nine items had difficulty indices above .30, indicating that most of the items did not suffer from floor effects. Eight items had item-difficulty indices ≥.90, suggesting ceiling effects for these items.

Examination of other item discrimination indices revealed that all items had positive corrected point-biserial correlations, indicating that the item scores functioned as expected in relation to the total score. Further, 54 items (65%) had corrected point-biserial correlations ≥.25. Items with the lowest discrimination indices (in this case, weakest corrected item-total point biserial correlations) were found in the subscale Getting help when needed. This was unsurprising, as items in this subscale were hypothetical questions about a parent’s likelihood of taking certain actions if they noticed a change in their child’s mood or behaviour, whereas most other items in the PaRCADS require parents to report on their usual parenting behaviour. An examination of the B-index for items in this subscale further indicates that the values were low, suggesting that the items did not strongly differentiate parents who scored at or above the cut-off score and parents who scored below the cut-off score for overall concordance with the Guidelines.

Reliability analysis

Agreement coefficients (p_o) were used to provide estimates of the reliability/dependability of the scale. The overall measure (e.g., total score) was highly consistent (p_o = .95). Eight subscales had acceptable to high agreement coefficients (.75 to .92). The remaining two subscales had agreement coefficients (p_o = .74) just below the recommended cut-point of .75 (Subkoviak, 1988).

Item reduction

After considering the inter-item correlations within subscales, corrected item-total point biserial correlations, experts’ endorsement on item-objective congruence, item-relevance CVIs and feedback from experts, four items were removed due to low ratings in three or more aspects of the aforementioned. Deletion of these four items improved their corresponding subscale agreement coefficients for three subscales (p_o increased by .07 for Relationship with your child, by .02 for Child relationship with others, and by .15 for Getting help when needed) and resulted in a reduction for one subscale (p_o reduced by .02 for Health Habits). While there were other items with discrimination indices and expert ratings that fell outside the recommended ranges (e.g., items 2.7, 4.3), they were retained because they represented key parenting practices for the prevention of child anxiety and depression in the literature. Although items with high facility indices (i.e., too easy), such as items 1.1 and 10.1, did not maximally differentiate between respondents in this sample, they might display statistical variability in a clinical sample and were thus retained.

The agreement coefficient for the resultant total score was still very high (p_o = .94), indicating that reliability/dependability was high for the measure as a whole. At the subscale level, the revised agreement coefficients ranged from .74 to .90. The 79 items retained after this process were subjected to subsequent analyses. Table 2 shows the descriptive and dependability statistics for the PaRCADS revised subscale and total scale scores. As shown in Table 3, there were positive correlations of varied strength between the PaRCADS subscales scores, and moderate to high correlations between each subscale score and the total score.

Parental concordance with guidelines

The descriptive statistics for the PaRCADS revised subscale and total scale scores are presented in Table 2. Concordance rates varied across subscales. The lowest average concordance rate was observed in Health Habits (11.5%), while the highest was observed in Getting help when needed (91.3%). More than 30% of parents scored within the concordant range for six out of ten subscales. The mean PaRCADS total score was 50.18 (SD = 10.60).

Associations between parental concordance and participant characteristics

Using the PaRCADS revised total score as an indicator of overall parental concordance with the Guidelines, we found small, negative correlations between the PaRCADS total score and parental concern about their child’s risk of depression or anxiety, with higher PaRCADS total scores associated with less parental concern about their child’s risk (see Table 4). There were also small, negative correlations between the PaRCADS total score and parent reported child’s current mental health problem (r =  − .14, p = .01) and history of mental health diagnosis (r =  − .11, p = .03). Of the parent’s demographic characteristics, only parental age was correlated with the PaRCADS total score (r = .10, p = .046).

Construct validity

To assess its construct validity, we examined the Pearson’s correlations between the PaRCADS revised total score and the subscale scores of other parenting, family functioning and HRQoL measures.

Convergent validity

Results of the convergent validity tests are shown in Table 5. The correlations between the PaRCADS total score and parent reports on the CRPBI-Acceptance subscale score (r = .54, p < .01) and the PCS subscale score (r =  − .42, p < .01) were strong and in the predicted directions. A higher PaRCADS total score was also associated with a lower level of problematic family functioning (r =  − .52, p < .01).

Parent-child agreement on closely related measures could provide another form of convergent validity. There was a small positive correlation between PaRCADS and child-report on the CRPBI-Acceptance subscale [r(144) = .16, p = .03], but not the PCS [r(144) =  − .13, p = .06]. An examination of parent–child agreement on the CRPBI-Acceptance subscale and PCS showed that there were small significant correlations between parent- and child-reports on the CRPBI-Acceptance subscale and PCS (rs .15 and .24).

Associations between parental concordance, child symptoms and child health-related quality of life

Given that the PaRCADS was intended to assess parenting practices to reduce the risk of child anxiety and depression, we expected small negative correlations between the PaRCADS total score and the RCADS subscale and total symptom scores. Further, we predicted that the PaRCADS total score will be positively correlated with child HRQoL on the KIDSCREEN.

There were small correlations in the expected directions for both parent and child reports of RCADS anxiety, depression and total scores, with lower child symptom scores being associated with higher PaRCADS total scores (see Table 6). To further explore the differences in PaRCADS total score between parents of children with borderline or clinically elevated symptoms (i.e., RCADS T-score >65), and those without, independent sample t tests were conducted separately for parent and child report of RCADS (two subscales and total scale scores). Results showed significant group differences with small effect sizes in PaRCADS total score for depressive symptoms [t (353) = 2.81, p = .01, g =  − .37] and total symptoms [t (353) = 2.64, p = .01, g =  − .37] based on parent report only. There were no group differences based on child report of RCADS (see Table S2).

As predicted, PaRCADS total score was also moderately correlated with parent report of child HRQoL in the KIDSCREEN Parent relations and autonomy dimension, r = .39, p < .01. There were small to moderate correlations between PaRCADS and parent reports of child HRQoL in non-parenting-related dimensions such as, Physical well-being, Psychological well-being, Peers and social support, and School environment (see Table 6). Correlations between PaRCADS and child-self report of HRQoL were also significant (rs .11 to .20), except for the Physical well-being and Peers and social support dimensions. There were moderate to strong positive correlations between parent- and child- reports on the various dimensions of the KIDSCREEN. Specifically, there was a moderate, positive correlation between parent- and child-report on the Parent relations and autonomy dimension (r = .30, p < .01).

Parental concordance with guidelines

We suggest utilising the total score on the PaRCADS for assessing the extent to which a parent’s overall parenting practices align with evidence and expert consensus on parenting risk and protective factors for child anxiety and depression. Except for one subscale (Involvement in child’s life, p_o = .74) where its estimate of reliability fell just below the acceptable level, the individual subscale scores may be used if it is of interest to obtain a profile of parenting practices across different domains or to assess changes in parenting at the domain level.

In line with the notion that a carer’s parenting competence can vary across different domains of social interaction (Grusec & Davidov, 2010), the rates of parental concordance with Guidelines varied widely across different domains of the PaRCADS. Based on the revised scale, Relationship with your child and Getting help when needed had the highest concordance rates (82.2% and 91.3%, respectively), while the remaining eight subscales had concordance rates ranging between 11% to 46%. Given that parents in the current sample were mostly well educated and had self-selected to participate in the study where they could receive an online preventive parenting program, and with most of their children agreeing to participate as well, it is unsurprising that parents had high scores on the two mentioned subscales. By contrast, low concordance rates were found in the Rules and consequences (16.1%), Managing emotions (15.5%) and Health habits (11.5%) domains. This finding may be interpreted in the context of the study sample where nearly two-thirds of the parents reported having experienced past mental health problems and 17.7% of parents were experiencing current mental health problems. However, we did not find a correlation between parents’ total score on the PaRCADS and their mental health status (past or current). There were also no significant differences in the mean PaRCADS total scores between parents with or without a past (p = .85) or current (p = .07) mental health history.

In Rules and consequences, only 12% of parents reported involving their children in the development of family rules. This finding may be interpreted in a number of ways. Parents may think that their children lack the maturity to contract reasonable boundaries for themselves and hence do not engage their child in the process of setting rules. Alternately, parents may think that their child should be involved in developing rules but do not know how to. Notably, more than half of the parents in this study have reported use of specific rules for their child’s behaviour and had reviewed or modified their rules to adapt to their child’s level of maturity and responsibility. Given that the child psychopathology literature has underscored that a sense of diminished control and influence on events and outcomes in their surroundings can increase risk for anxiety disorders in children (Chorpita & Barlow, 1998; Barber, 2002), these findings suggest that it may be important to help parents develop the skills and confidence to engage their children in setting rules for self-regulation.

It is concerning that less than 16% of parents were concordant with Guidelines in the Managing emotions domain. Specifically, a large proportion of parents endorsed use of non-supportive strategies that may be perceived by the child as minimising, dismissive or controlling, which are known to undermine a child’s ability to regulate their own emotions (Fox & Calkins, 2003). Consistent with evidence about the importance of emotion socialisation in child development (Izard et al., 2002; Yap, Allen & Sheeber, 2007), our findings point to the need for preventive efforts to support parents in developing skills in this area of parenting. Interestingly, a separate study on users of the Guidelines found that this parenting domain of managing their child’s emotions had the highest proportion of parents reporting an attempt to change from just reading the Guidelines document a month ago (Sim et al., 2017). This further suggests that many parents have the motivation and capacity to make changes to their emotion socialisation practices with their children.

Considering the evidence that lifestyle-related parenting behaviours influence their children’s body mass index, physical activity level, dietary habits and screen use (Jansen et al., 2013; Pyper, Harrington & Manson, 2016; Xu et al., 2018), the very low concordance rate for Health habits is notable. Only 15% of parents reported that food treats such as chocolates or soft drinks were not readily accessible to their children in the home. Further, only a third of the parents reported having good health habits themselves, such as regular exercise, healthy diet and sleep hygiene. Similarly, only about a third of the parents reported having set time limits on their child’s screen use. These findings highlight the need to support parents in modelling reasonable health habits and limiting the availability of high dairy/high sugar products and screen-based entertainment at home.

While it may be argued that the current cut-off scores set for parental concordance seem arbitrary, they were deliberately set to require close-to-absolute concordance with the recommendations in the Guidelines, given that the Guidelines were informed by high quality research evidence and supported by international experts. More importantly, the cut-offs for parental concordance at subscale and total scale levels were necessarily high to serve an instructional-design function in preventive programs, such that researchers or clinicians could identify areas where each parent may benefit from further development. Nonetheless, the PaRCADS can also yield continuous scores (subscale scores and total score) that may be useful in monitoring progress or in program evaluation.

Correlates of parental concordance with guidelines

The small but significant correlations between parental concordance and both parent-and child-reported child anxiety and depressive symptoms provide support for the parenting Guidelines informing the development of the PaRCADS, and for the potential utility of the PaRCADS in preventive interventions targeting parents of primary school-aged children.

Comparing the PaRCADS total scores for parents of children above and below the RCADS borderline cut-off scores, we also found greater parental concordance among parents of children below the RCADS borderline cut-off score on the parent report. Further, the negative albeit small correlations between the PaRCADS total score and parent reported child’s current or history of mental health problems suggest that there is a relationship between parental concordance with Guidelines and child’s mental health. Although cross-sectional analyses preclude an assessment of the direction of influence, the present findings are largely consistent with the literature supporting a small but significant association between parenting and child internalising symptoms (McLeod, Weisz & Wood, 2007; McLeod, Wood & Weisz, 2007; Van der Sluis, Van Steensel & Bögels, 2015; Yap & Jorm, 2015; Pinquart, 2017).

The current findings of positive correlations between PaRCADS and the KIDSCREEN subscales provide preliminary evidence that parenting practices are associated with various dimensions of a child’s HRQoL. This is important given burgeoning research demonstrating that children with various mental disorders are consistently rated by their parents to have lower HRQoL (e.g., Dey, Mohler-Kuo & Landolt, 2012). While the mechanisms of transmission are not yet clear, these findings suggest that supporting parents in their efforts to reduce the risk of their children developing clinical anxiety and depression may be fruitful in improving their children’s HRQoL.

With regard to parent’s demographic and mental health characteristics, only parent’s age and parental concerns with their child’s risk of developing depression or anxiety were associated with parental concordance. Specifically, older parents reported higher parental concordance, and parents with higher concordance were less concerned about their child’s risk of having anxiety or depression. That older parents reported themselves as more concordant with the Guidelines is consistent with research that found that older mothers display greater sensitivity and less hostile childrearing practices than younger mothers (Fergusson & Woodward, 1999; Lewin, Mitchell & Ronzio, 2013). Greater maturity may better prepare older parents for the stresses of parenthood and thus evolve more adaptive parenting practices than younger parents. Contrary to studies that found higher parental education level to be associated with increased knowledge about effective parenting and child development, we did not find a significant correlation between parent educational level and parental concordance. As some would contend that parent self-reported knowledge does not necessarily translate into actual practice, the mixed findings may be explained by the fact that our study examined current practices rather than knowledge, while the latter was the focus in studies that reported a correlation with education level (e.g., Morawska, Winter & Sanders, 2009; Bornstein et al., 2010). Alternatively, the relationship with parental concordance may have been attenuated by the narrow range of education level among the participants in the current sample. The finding that parents who rated themselves as more concordant with the Guidelines were less concerned about their child’s risk of anxiety or depression is in line with a similar study that examined parental concordance with guidelines in the context of preventing adolescent depression and anxiety disorders (Cardamone-Breen et al., 2017). Before conclusions can be drawn about the predictive utility of the PaRCADS, further research is required to assess the relationship between parental concordance and their child’s risk of developing anxiety and depression.

Limitations, strengths, implications and future directions

Findings from this study should be considered in light of its limitations. Although only a small proportion of children declined to participate in the study with their parents, the fact that their parents rated them higher on the RCADS total symptoms than parents of participating children suggests that the current findings may not be generalisable to children with clinical levels of anxiety and depression. On the other hand, the homogeneity and composition of the participating parents—self-enrolled, highly educated—suggests that the current findings are likely generalisable to research and programs that typically attract participants with these sociodemographic characteristics, such as web-based preventive parenting programs (Enebrink et al., 2015; Suárez, Byrne & Rodrigo, 2018; Yap et al., 2018). These findings also need to be considered in the context of an over-representation of mothers and the age range of the children (8–11 years) in the sample.

As far as we know, the PaRCADS is the first self-assessment measure that encapsulates the range of parenting risk and protective factors in the child anxiety and depression literature that is also supported by international experts in parenting and child mental health. Further research is needed to examine its measurement precision and to assess its reliability and validity in more diverse samples. Likewise, longitudinal data are required to establish more clearly the associations between PaRCADS scores and child anxiety and depressive symptoms. In addition, some may challenge that children’s perceptions of parenting are more important, given that parent and child reports are known to be only modestly correlated (Martin et al., 2004; De Los Reyes & Kazdin, 2005). Studies that utilised multi-informant reports continue to find that parents’ self-assessment of parenting behaviour explains additional variance beyond the contribution of children’s report and are predictive of child outcomes (e.g., Ratelle, Duchesne & Guay, 2017). Future studies that employ a multi-method (e.g., clinician observation) and multi-informant (e.g., co-parent/caregiver) approach would also provide a more robust assessment of the validity of PaRCADS as a measure of parenting (Podsakoff et al., 2003; De Los Reyes et al., 2013). Another step would be to examine the utility of PaRCADS for other mental health and functioning outcomes (e.g., externalising problems, positive well-being) across different age groups (e.g., 5–7 years) and samples (e.g., clinical, cross-cultural).