The Strengths and Difficulties Questionnaire: Factor Structure of the Father-Report and Parent Agreement in 2-Year-Old Children

Abstract

There is limited research on the preschool version of the Strengths and Difficulties Questionnaire (SDQ), and comparisons between mothers and fathers as informants and whether the factor structure shows measurement invariance across parents is lacking. Our study involved mothers (n = 6,246) and fathers (n = 3,759) of 2-year-old children from the Growing Up in New Zealand birth cohort. Confirmatory factor analysis was used to evaluate the factor structure of the SDQ and test for measurement invariance across mothers and fathers. For fathers, we found support for a modified five-factor model that accounts for a positive construal method effect. Internal consistency was good for measures except peer problems. Full measurement invariance of this modified model was found across mothers and fathers, and parents showed moderate agreement in their SDQ ratings (0.34 ≤ r ≤ 0.44). More research is needed on whether mother- and father-reports differ in sensitivity when screening for early childhood psychiatric disorders.

Keywords

child behavior SDQ cohort longitudinal measurement invariance psychometric

Introduction

When identifying children with behavioral and emotional difficulties, parents are often used as informants on screening questionnaires such as the Strengths and Difficulties Questionnaire (SDQ; Goodman, 1997). The early use of such questionnaires is vital, as psychiatric disorders can occur from early in childhood (Egger & Angold, 2006). Young children with psychosocial difficulties can also show a continuity in symptoms and are at greater risk of negative health outcomes later in life (Bardone et al., 1998; Bittner et al., 2007; Caspi, Moffitt, Newman, & Silva, 1996; Costello, Mustillo, Erkanli, Keeler, & Angold, 2003; Fergusson, Horwood, & Ridder, 2005; Fergusson & Lynskey, 1998; Hofstra, van der Ende, & Verhulst, 2002; Odgers et al., 2007). As such, screening questionnaires need to be psychometrically sound. Importantly, in instances where parents are used as informants, it is necessary to consider whether measures used show consistent validity and reliability across both mothers and fathers.

The SDQ is a 25-item behavior screening questionnaire for children and adolescents. It consists of four difficulties subscales (emotional symptoms, peer problems, hyperactivity–inattention and conduct problems), and a prosocial behavior subscale. A measure of total difficulties can also be calculated by summing all difficulties subscales. The questionnaire includes 10 positively worded items (all items for the prosocial behavior subscales and five reverse-scored items from the difficulties subscales), with all other items worded negatively. The standard parent-report version can be used with children aged 4 to 17 years, with a modified preschool version also available for use with children aged 2 to 4 years (Youth in Mind, 2014). There is extensive research on the psychometric properties of the standard parent-rated SDQ (e.g., Becker, Woerner, Hasselhorn, Banaschewski, & Rothenberger, 2004; Dickey & Blumberg, 2004; Du, Kou, & Coghill, 2008; Goodman, 2001; Hawes & Dadds, 2004; He, Burstein, Schmitz, & Merikangas, 2013; Hill & Hughes, 2007; Matsuishi et al., 2008; Mellor & Stokes, 2007; Smits, Theunissen, Reijneveld, Nauta, & Timmerman, 2016; Van Leeuwen, Meerschaert, Bosmans, De Medts, & Braet, 2006; Van Roy, Veenstra, & Clench-Aas, 2008; Woerner, Becker, & Rothenberger, 2004), and to a lesser extend the preschool version (e.g., Croft, Stride, Maughan, & Rowe, 2015; Doi, Ishihara, & Uchiyama, 2014; D’Souza, Waldie, Peterson, Underwood, & Morton, 2017; Ezpeleta, Granero, de la Osa, Penelo, & Domènech, 2013; Klein, Otto, Fuchs, Zenger, & von Klitzing, 2013; Theunissen, Vogels, de Wolff, & Reijneveld, 2013). Teacher- and self-report versions have also been developed for use in appropriate age groups.

Studies investigating the agreement between mothers and fathers on the SDQ have typically evaluated correlations and differences in children’s scores between parents. Using Pearson or Spearman’s correlation coefficients, moderate to large correlations between parents’ SDQ scores have been observed for children aged 4 to 9 years (r range: 0.37-0.64; Borg, Kaukonen, Salmelin, Joukamaa, & Tamminen, 2012; Davé, Nazareth, Senior, & Sherr, 2008; Mellor, Wong, & Xu, 2011). Similar results have been found in older children and adolescents (Griffith, Hastings, & Petalas, 2014).

When looking at differences in children’s SDQ scores between mothers’ and fathers’ ratings, mixed results have been found. In siblings of individuals with autism spectrum disorder, Griffith et al. (2014) found no differences in the scores from mothers and fathers on all difficulties measures, though mothers reported greater prosocial behavior in their children than fathers. Similar results were also found when looking at mothers’ and fathers’ ratings of SDQ measures for Chinese children (Mellor et al., 2011). However, in a sample of 4- to 6-year-olds, higher scores were reported by fathers on the hyperactivity–inattention and conduct problems subscales, as well the total difficulties score (Davé et al., 2008).

The mixed findings in the studies reported above could be explained by the age range of the studies’ samples. While the sample used by Griffith et al. (2014) included younger children, the age range of participants was broad and extended to late adolescence. Additionally, Mellor et al.’s (2011) sample included older primary school-aged children. It is possible that mothers of younger children, such as those in the study by Davé et al. (2008), rate their child as having fewer difficulties than fathers. They may be more used to difficult behaviors in their children as a result of spending more time with their child in the early years than fathers (Craig, 2006). However, time spent with their child tends to decrease in both mothers and fathers as children get older (Åman-Back & Björkqvist, 2004). This may result in similar scores between parents, as mothers may be less exposed to their children’s difficult behaviors. Indeed, there does appear to be evidence that the agreement between parents’ scores increases as child’s age increases (Duhig, Renk, Epstein, & Phares, 2000). Therefore, differences may be expected in mothers’ and fathers’ scores of their child, particularly on measures of behavioral difficulties, if their child is of preschool age or younger.

When evaluating agreement between mothers’ and fathers’ ratings on the SDQ, it is also important to consider measurement invariance of the questionnaire’s factor structure across both parents. The factor structure of the parent-report version of the SDQ has been thoroughly investigated in 2- to 19-year-olds, generally using SDQ responses from an unspecified parent or just mothers (e.g., Becker et al., 2004; Croft et al., 2015; D’Souza et al., 2017; Du et al., 2008; Ezpeleta et al., 2013; Goodman, 2001; Hawes & Dadds, 2004; He et al., 2013; Hill & Hughes, 2007; Klein et al., 2013; Stone et al., 2013; Van Leeuwen et al., 2006; Van Roy et al., 2008). Several of these studies have tested Goodman’s (1997) five-factor structure, though some studies (e.g., Croft et al., 2015; Dickey & Blumberg, 2004; Klein et al., 2013; Riso et al., 2010; Van Leeuwen et al., 2006) have also investigated a three-factor model comprising of an internalizing factor consisting of some items relating to peer problems and emotional symptoms; an externalizing factor consisting of items relating to hyperactivity–inattention and conduct problems; and a prosocial factor, which consists of the prosocial items and some of the positively worded reverse-scored items. In general, studies that compared the two models showed that the three-factor model was a poorer fit than the five-factor model (Croft et al., 2015; Klein et al., 2013; Van Leeuwen et al., 2006).

However, similar to the three-factor model, there is also evidence from exploratory factor analysis that the positively worded reverse-scored items cross-load onto the prosocial subscale with the five-factor model (e.g., Du et al., 2008; Goodman, 2001; Hawes & Dadds, 2004), indicating that there may be a positive construal method effect present; that is, that this “positive construal factor” may be reflecting informants’ willingness to attribute positive qualities (Goodman, 2001). Further support for the presence of a positive construal factor has come from studies that have used confirmatory factor analysis (CFA) to compare Goodman’s (1997) original five-factor model with a model that accounts for this positive construal response style. In an earlier evaluation of the mother-report of the preschool SDQ in 2-year-old New Zealanders, we found unsatisfactory model fit for the original five-factor model, but improved and satisfactory fit in a model that allowed positively worded items to cross-load onto the prosocial factor (D’Souza et al., 2017). Model fit was also improved when accounting for a positive construal factor in a sample of older children and adolescents (Van Roy et al., 2008), and with the self-report version of the SDQ (van de Looij-Jansen, Goedhart, de Wilde, & Treffers, 2011).

Studies that use SDQ scores from an unspecified parent or compare scores from both parents are assuming that the SDQ factor structure shows measurement invariance across both mothers and fathers; that is, that mothers and fathers are interpreting the SDQ in the same way, and that factors are measuring the same underlying constructs across both parents. However, only one study thus far has investigated the factor structure in fathers’ report of the SDQ, and evaluated measurement invariance of the factor structure across mothers and fathers (Chiorri, Hall, Casely-Hayford, & Malmberg, 2016). Using exploratory structural equation modelling, Chiorri et al. (2016) found support for the five-factor model with both mothers’ and fathers’ reports of the SDQ in 51-month-old children. However, similar to studies using exploratory factor analyses (e.g., Du et al., 2008; Goodman, 2001; Hawes & Dadds, 2004), the study found that most of the reverse-scored items had substantial cross-loadings onto the prosocial behavior subscale, indicating that these items may reflect more than one latent construct. Chiorri et al. (2016) also demonstrated measurement invariance of this model across both mothers and fathers.

To our knowledge, no study has yet evaluated the factor structure of a preschool SDQ rated by fathers, and measurement invariance of this preschool version across mothers and fathers have not yet been assessed; as such, the current study addressed this gap in the literature. Mother- and father-rated SDQ data were obtained for a sample of 2-year-old New Zealand (NZ) children. As noted above, we have previously evaluated the psychometric properties, including the factor structure, of the mother-rated preschool SDQ for this cohort and found support for a modified model that accounted for a positive construal factor (D’Souza et al., 2017). Thus, our aim was to investigate whether these results were replicated using the father-rated SDQ data.

Given that past research with both mother- and self-rated SDQ data (D’Souza et al., 2017; van de Looij-Jansen et al., 2011) has found superior model fit with a modified model that accounts for a positive construal factor, we hypothesize that this modified model will also show better fit than Goodman’s (1997) original model when using the father-rated SDQ. If our hypothesis is supported, measurement invariance of the modified model will be assessed across both mothers and fathers. We will also evaluate the agreement between mothers’ and fathers’ SDQ scores of their child. If measurement invariance is satisfactorily established, mean differences in the scores from mothers and fathers will be investigated.

Method

Participants

Participants were mothers and fathers of children from the longitudinal Growing Up in New Zealand study. The prospective study follows a socioeconomically and ethnically diverse cohort of NZ children that were recruited via 6,822 pregnant mothers who had expected delivery dates between April 25, 2009 and March 25, 2010. A detailed description of the study’s design and recruitment is provided elsewhere (Morton et al., 2013, 2014). Briefly, pregnant mothers were recruited from the geographical area covered by three contiguous District Health Board regions, which contains approximately one third of the NZ’s birth population (Morton et al., 2013). In general, statistics for mothers on key parameters such as mother’s age, ethnicity, parity, and socioeconomic status were comparable with those of all NZ parents (Morton et al., 2013), and children within the cohort are broadly generalizable to current NZ births (Morton et al., 2015). Recruited pregnant women were asked to provide contact information for a partner, defined as someone whom they were in a “significant social relationship with” (Pryor, Morton, Bandara, Robinson, & Grant, 2014). A total of 4,401 partners consented to participate in the study. So as to remain consistent with the literature, as past research has generally focused on fathers (as opposed to mothers’ partners in general), the current study will only include data from fathers of the children. In this article, a father is defined as someone who identifies themselves as male and is either the biological, adoptive, or stepfather of the child. At the 9-month data collection wave, information on partners’ gender and relationship to the child was obtained. Overall, 4,074 partners were identified as fathers, with 4,062 being biological fathers and 12 being either adoptive or stepfathers. Partners that were excluded from the analyses included 13 partners who identified as female, 1 partner who identified as another family member, and 4 who identified as “other.” Several partners (n = 309) did not respond at the 9-month data collection wave, and were therefore also excluded from analyses as the nature of their relationship to the child was unknown.

Major data collection waves for the study have occurred during the late antenatal period, at 9 months and at 2 years, using computer-assisted personal interviews. Mothers and partners were interviewed separately, with 64% of the 2-year interviews being follow-ons (i.e., the partner’s interview followed immediately after the mother or vice versa). Each data collection wave focuses on gathering information relating to six interconnected domains of child development: health and well-being, family and whanau (extended family), culture and identity, education, psychosocial and cognitive development, and neighborhoods and societal context. Ethical approval was granted from the Ministry of Health Northern Y Regional Ethics Committee (NTY/08/06/055), with written informed consent provided by both mothers and partners.

At 2 years, 6,246 mothers and 3,759 fathers completed the preschool SDQ. Children without father-rated SDQ data were more likely to come from high deprivation areas and have a mother of Māori or Pacific ethnicity, ps < .001. These children were also more likely to have higher scores at 2 years on all difficulties measures from the mother-rated SDQ when compared with children with father-rated data, ps < .001. Fathers who did not provide SDQ data were also more likely to identify as Māori, Pacific, or Asian ethnicity, p < .001.

Measures

Strengths and Difficulties Questionnaire

Mothers and fathers independently answered the English (U.K.) version of the parent-rated preschool SDQ, which was developed for children between the ages of 2 and 4 years (Goodman, 1997; Youth in Mind, 2014). The 25-item questionnaire is designed to measure five subscales (emotional symptoms, peer problems, hyperactivity–inattention, conduct problems, and prosocial behavior). All subscales except prosocial behavior can be summed to give a measure of total difficulties. Three items have been modified in the preschool version of the questionnaire, relative to the standard version, so that it is more appropriate for preschool children. On the conduct problems subscale, the items “often lies and cheats” and “steals from home, school or elsewhere” have been modified to “often argumentative with adults” and “can be spiteful to others,” respectively; on the hyperactivity subscale, the item “can stop and think things out before acting” has been changed to “thinks things out before acting.” A 3-point Likert-type scale was used to rate each item, with response scores of 0 for not true, 1 for somewhat true, and 2 for certainly true. Positively worded items from the difficulties subscales were reverse-coded. Higher scores reflect greater difficulties or greater prosocial behavior. Previous research with the mother-report has shown generally good internal consistency for the original subscales and total difficulties score within this cohort (D’Souza et al., 2017).

Data Analysis

CFA was used to test two a priori models with the father-rated SDQ: the original five-factor model proposed by Goodman (1997), and the modified five-factor structure identified as the best-fitting model by van de Looij-Jansen et al. (2011) and D’Souza et al. (2017). This modified model accounted for a positive construal response style by allowing cross-loadings of positively worded, reverse-scored items onto the prosocial behavior subscale. SDQ items were treated as ordinal variables, due to being measured on a 3-point Likert-type scale.

Robust diagonal weighted least squares (DWLS) estimation was used with the CFA. This is because DWLS estimation is more appropriate for ordinal data than maximum likelihood estimation, as the latter assumes that the observed variables follow a multivariate normal distribution (Li, 2016). With robust DWLS estimation in the current study, thresholds and polychoric correlations are estimated using maximum likelihood (Li, 2016). Robust standard errors and a mean- and variance-adjusted test statistic are then calculated using the estimated asymptotic covariance matrix of the thresholds and polychoric correlations in a weight matrix, and model parameters are obtained using the diagonal weight matrix (Li, 2016; Rosseel, 2012).

Scales were set for latent variables using the marker variable strategy. The comparative fit index (CFI), Tucker–Lewis index (TLI), gamma hat, and root mean square error of approximation (RMSEA) were used to evaluate model fit. CFI and TLI values of around .90 or greater, gamma hat values of approximately .95 or greater and RMSEA values less than .06 indicate acceptable model fit (Bentler, 1990; Hu & Bentler, 1999).

It was planned that measurement invariance of the model would be assessed if the best-fitting model for the father-rated SDQ was consistent with what has been found with the mother-report within this sample (i.e., the modified five-factor mode; D’Souza et al., 2017). To test for measurement invariance, latent factors were modelled separately for mothers and fathers and nested within the child. Latent factors were correlated between mothers and fathers, and corresponding items’ residuals were correlated (e.g., the residual for Item 1 from the mother-report was correlated with the residual for Item 1 from the father-report). Configural, metric, and scalar invariance was tested. First, factor loadings and thresholds were free to vary across mothers and fathers (Model 1). This configural invariance test examines whether the same configuration holds across groups. Following this, factor loadings were constrained to be equal across groups, but thresholds were free to vary across mothers and fathers (Model 2). This metric or weak invariance test examines whether respondents from the two groups attribute the same meaning to the latent SDQ factors. Finally, factor loadings were also constrained to be equal across mothers and fathers (Model 3). This scalar or strong invariance test examines whether the interpretation of the latent factor (factor loadings) and the level of endorsement of the items (intercepts) are equal in both groups and hence whether both groups can be compared with their latent factor scores. Researchers have typically used ΔCFI as an index of difference in fit between two samples with increasing levels of model constraint. Models are considered to not significantly differ, and demonstrate invariance, if ΔCFI is equal to or less than .01 (Cheung & Rensvold, 2002).

Cronbach’s alpha was used to evaluate the internal consistency of the father-rated SDQ subscales and total difficulties score. Mean inter-item correlation (MIC) coefficients were also used to assess internal consistency, as Cronbach’s alpha calculations can be influenced by short subscale length (Nunnally & Bernstein, 1994; Streiner, 2003). Alpha coefficients of .60 and above are considered acceptable in SDQ literature (Hawes & Dadds, 2004; Matsuishi et al., 2008; van de Looij-Jansen et al., 2011; Van Leeuwen et al., 2006; Woerner et al., 2004). MIC coefficients of .20 and above are recommended (Nunnally & Bernstein, 1994; Piedmont, 2014). Due to the ordinal nature of SDQ items, calculations of alpha and inter-item coefficients were based on polychoric correlations.

Similar to previous studies (Borg et al., 2012; Davé et al., 2008; Griffith et al., 2014; Mellor et al., 2011), Pearson and Spearman’s correlation coefficients were used to evaluate the agreement between scores reported by mothers and fathers. If measurement invariance of the modified five-factor model is found across mothers and fathers, mean differences in SDQ subscale and total difficulties scores will be investigated using paired sample t tests. Effect sizes for mean differences were determined using Cohen’s (1988) d.

CFA, measurement invariance analyses, and Cronbach’s alpha calculations were conducted in R, using the statistical packages lavaan (Rosseel, 2012) and semTools (Pornprasertmanit, Miller, Schoemann, & Rosseel, 2013). All other analyses were performed using IBM SPSS Statistics version 22. Statistical significance was set at an alpha level of .05.

Results

Confirmatory Factor Analysis

Model fit indices did not meet recommended criteria for the original five-factor model, χ²₍₂₆₅₎ = 4215.90; CFI = .755; TLI = .723; gamma hat = .915; RMSEA = .068. Improved and acceptable model fit was found for the modified five-factor model, χ²₍₂₆₀₎ = 1891.66; CFI = .899; TLI = .883; gamma hat = .966; RMSEA = .044. Standardized factor loadings for the original and modified five-factor models are presented in Table 1. All factor loadings were significant for the original five-factor model, ps < .05. With the modified model, near zero loadings were observed for Item 11 (good friend) on the peer problems subscale, and Item 21 (reflective) on the hyperactivity–inattention subscale. The loading for Item 21 also did not reach significance. All other items had significant factor loadings, ps < .05.

Table 1.

Standardized Factor Loadings for the Original and Modified Five-Factor Models of the Father-Rated SDQ.

Item	Original model					Modified model
Item	Emotional symptoms	Peer problems	Hyperactivity–inattention	Conduct problems	Prosocial behavior	Emotional symptoms	Peer problems	Hyperactivity–inattention	Conduct problems	Positive construal
3. Somatic	.59					.60
8. Worries	.63					.64
13. Unhappy	.75					.73
16. Clingy	.43					.43
24. Fears	.54					.54
6. Solitary		.35					.43
11. Good friend^a		.31					.06			−.34
14. Popular^a		.52					.18			−.48
19. Bullied		.54					.68
23. Best with adults		.36					.46
2. Restless			.60					.62
10. Fidgety			.69					.74
15. Distractible			.65					.69
21. Reflective^a			.39					.04*		−.50
25. Persistent^a			.48					.23		−.38
5. Tempers				.60					.62
7. Obedient^a				.50					.23	−.45
12. Fights				.67					.69
18. Argues				.61					.64
22. Spiteful				.57					.59
1. Considerate					.72					.68
4. Shares					.54					.54
9. Caring					.60					.59
17. Kind to kids					.62					.61
20. Helps out					.48					.52

Note. SDQ = Strengths and Difficulties Questionnaire.

Reverse-scored items.

p > .05.

Measurement Invariance

As the modified five-factor model showed superior model fit with the father-report of the SDQ, replicating what has been found previously with the mother-report (D’Souza et al., 2017), measurement invariance of this model across mothers and fathers was tested. Model fit indices, including CFI and ΔCFI, for Models 1, 2, and 3 are provided in Table 2. Configural, metric, and scalar invariance were supported, as constraining factor loadings and then factor thresholds to be equal across mothers and fathers resulted in ΔCFI of less than .01 when compared with less constrained models. This suggests that the interpretation of the SDQ factors (the factor loadings), and the level of endorsement of the underlying items are equal in both mothers and fathers and hence, the groups can be compared on their SDQ latent scores.

Table 2.

Fit Indices for Invariance Analyses for the Modified Five-Factor Model Across Mothers and Fathers.

	χ²	df	RMSEA	Gamma hat	TLI	CFI	ΔCFI^a
1. Configural invariance	4276.56	1095	.031	.957	.894	.905
2. Metric invariance	4146.94	1116	.030	.956	.901	.909	.004
3. Scalar invariance	4346.51	1161	.030	.954	.899	.905	.004

Note. df = degrees of freedom; RMSEA = root mean square error approximation; TLI = Tucker–Lewis index; CFI = comparative fit index.

CFI of constrained model—CFI of less constrained model. Only the absolute value is given.

Internal Consistency

Table 3 presents Cronbach’s alpha and MIC values for father-rated SDQ subscales (including both the original prosocial behavior subscale and the modified positive construal factor) and total difficulties. Table 3 also provides alpha and MIC coefficients for mother-rated SDQ data, as reported by us in D’Souza et al. (2017). D’Souza et al. (2017) did not calculate alpha and MIC coefficients for the positive construal factor, so this has also been provided. Father-rated peer problems had the lowest Cronbach’s alpha and MIC values, α = .50; MIC = 0.17. Cronbach’s alpha and MIC coefficients were satisfactory for all other father-rated SDQ measures, α > .60; MIC > 0.20. Father-rated total difficulties obtained an alpha value of .82 and MIC coefficient of 0.28.

Table 3.

Internal Consistency, Mother–Partner Agreement, and Mean Scores for Each SDQ Measure.

Scale	Cronbach’s alpha (MIC)		Pearson r (Spearman’s r)	Mean scores (SD)		t
Scale	Mother^a	Father	Pearson r (Spearman’s r)	Mother^c	Father	t
Emotional Symptoms	0.75 (0.37)	0.71 (0.34)	0.39 (0.38)**	1.61 (1.47)	1.73 (1.48)	4.60**
Peer Problems	0.54 (0.20)	0.50 (0.17)	0.36 (0.35)**	2.01 (1.62)	2.00 (1.55)	0.29
Hyperactivity–Inattention	0.71 (0.33)	0.68 (0.29)	0.35 (0.35)**	4.17 (2.13)	4.26 (2.00)	2.30*
Conduct Problems	0.74 (0.36)	0.71 (0.33)	0.43 (0.41)**	2.86 (1.86)	2.86 (1.81)	0.12
Original Prosocial Subscale^b	0.71 (0.33)	0.73 (0.35)	0.34 (0.34)**	7.11 (1.85)	6.99 (1.86)	3.34**
Positive Construal Factor	0.79 (0.27)	0.79 (0.27)	0.34 (0.33)**	13.88 (3.03)	13.80 (3.03)	1.54
Total Difficulties	0.84 (0.31)	0.82 (0.28)	0.44 (0.43)**	10.65 (4.80)	10.85 (4.64)	2.39*

Note: MIC = mean interitem correlation; SDQ = Strengths and Difficulties Questionnaire.

Cronbach’s alpha and MIC coefficients for all mother-reported SDQ measures, excluding the positive construal factor, were obtained from D’Souza et al. (2017). ^bResults are presented for the prosocial behavior subscale to remain consistent with the literature. ^cMean scores for mother-rated SDQ measures are based on the scores of children with father-rated SDQ scores (n = 3,759).

p < .05. **p < .001.

Agreement Between Mothers and Fathers

Correlations between SDQ scores reported by mothers and fathers are presented in Table 3. In general, significant moderate correlations were found for all SDQ measures, Pearson r > 0.30, ps < .001 for all. Among the subscales, conduct problems had the strongest correlation, Pearson r = 0.43. Total difficulties obtained a Pearson correlation coefficient of 0.44.

Mean Differences Between Mother- and Father-Rated Scores

Mean scores for both mother- and father-rated SDQ subscales and the total difficulties score are presented in Table 3. As comparisons were conducted between mother- and father-ratings, mean scores provided for mother-rated SDQ measures are based on children with father-rated SDQ scores. Paired sample t tests revealed that fathers reported significantly higher emotional symptoms, hyperactivity–inattention and total difficulties than mothers, ps < .05 for all. Mothers reported significantly higher scores on the original prosocial subscale when compared with fathers, p < .05. No other differences were significant, and all significant mean differences were negligible in size, 0.04 ≤ d ≤ 0.07.

Discussion

The current study evaluated the factor structure and internal consistency of a father-rated preschool SDQ in 2-year-old NZ children. The factor structure and internal consistency of the preschool SDQ, rated by the children’s mothers, was assessed previously by D’Souza et al. (2017). Measurement invariance of the factor structure across mothers and fathers was also evaluated, as well as the agreement and differences between mothers’ and fathers’ ratings for each SDQ measure. This study is an important contribution to the literature, as it not only adds to limited research on the preschool version of the SDQ, it also addresses significant gaps in the literature by evaluating the psychometric properties of the father-rated preschool SDQ and its agreement with the mother-rated version.

Similar to our procedure used earlier (D’Souza et al., 2017), here we evaluated Goodman’s (1997) original five-factor model, as well as a modified five-factor model, which included a positive construal factor. This modified model accounted for a possible positive construal method effect by allowing positively worded reverse-scored items to cross-load onto what was originally the prosocial factor. The results supported our hypothesis, and replicated what we found with the mother-rated SDQ. Specifically, we found poor model fit for the original five-factor model, but improved and acceptable model fit with the modified five-factor model. Similar findings have also been shown with the self-report version of the SDQ in an older sample (van de Looij-Jansen et al., 2011). These results indicate that more investigation is needed with the SDQ, and future studies should assess both the original and modified models to evaluate whether an improved fit with the modified model is a consistent finding.

It is important to note that with the modified five-factor model, reverse-scored items have primary loadings onto the positive construal factor and smaller loadings onto their original factor. Near zero loadings were also found for the items “good friend” and “reflective” on the peer problems and hyperactivity–inattention subscales, respectively. Similar results were observed with this sample when evaluating the mother-report of the SDQ (D’Souza et al., 2017). These findings suggest that while some reverse-scored items are indicators of difficulties in this cohort, they appear to be predominantly reflecting the positive construal factor. The age of the cohort could explain why trivial and/or nonsignificant loadings were observed for the same two items on their original difficulties subscales, with both mother- and father-rated data. Strong friendships and advanced cognitive abilities such as reflective thinking are unlikely to have developed in children as young as 2 years; therefore, the characteristics that these items are measuring likely do not reflect “difficulties” within this age group. Modification of these items may be necessary if low loadings are found in other samples of 2-year-old children.

While our findings seem to support the presence of a positive construal method effect, there is a possibility that this positive construal factor can still be used as a measure of prosocial behavior. As noted by D’Souza et al. (2017), the reverse-scored items still reflect socially desirable behaviors that are important for prosociality, and van de Looij-Jansen et al. (2011) have suggested that the positive construal factor may reflect agreeableness and dutifulness with the self-report version of the SDQ. Recent research with the Growing Up in New Zealand cohort (Peterson, Dando, D’Souza, Waldie, Mohal, & Morton, 2016) has investigated the association between temperament at 9 months and the 2-year mother-report of the positive construal factor. Temperamental factors such as high orienting and regulatory control, high positive affect/surgency, high affiliation and low negative emotionality, which have been associated with prosocial behavior in other studies (Eisenberg et al., 1996; Rothbart, 2011; Rothbart, Ahadi, & Hershey, 1994), were found to be significantly associated with the mothers’ report of the positive construal factor at age 2. However, evaluations of the relationship between the positive construal factor and direct measures of prosocial behavior are still needed.

As the modified five-factor model was found to be the better fitting model for both the mother-report (D’Souza et al., 2017) and the father-report of the SDQ, measurement invariance analyses were performed to evaluate whether mothers and fathers were interpreting and responding to the questionnaire in the same way. Our results showed that configural, metric, and scalar invariance was found across mothers and fathers, supporting measurement invariance of the modified five-factor model. Previous research with the mother-report has also found full measurement invariance of this model across child’s gender and socioeconomic status, as well as partial but satisfactory measurement invariance across mother’s ethnicity (D’Souza et al., 2017).

Internal consistency was calculated for the father-rated SDQ measures, specifically for all original subscales, the positive construal factor and the total difficulties score. The internal consistency of the mother-rated SDQ subscales and total difficulties score was previously evaluated by D’Souza et al. (2017). We calculated the internal consistency of the original prosocial factor to remain consistent with the literature, though caution should be used when interpreting this finding as our CFA does not support its use. Our results were generally consistent with what was found with the mother-rated SDQ, specifically that four of the original five SDQ subscales (emotional problems, hyperactivity–inattention, conduct problems, prosocial behavior) and the total difficulties score obtained acceptable Cronbach’s alpha and MIC coefficients. We also calculated the internal consistency values of the mother- and father-reported positive construal factors, and found that both factors showed good internal consistency.

The peer problems subscale was the only measure to show poor internal consistency, with low alpha and MIC values of 0.50 and 0.17, respectively. A possible reason for this could be the near-zero loading of Item 11, but removal of this item decreased the alpha coefficient to .47. Low alpha coefficients have been regularly reported for the peer problems subscale in past research (Doi et al., 2014; Du et al., 2008; Goodman, 2001; Hawes & Dadds, 2004; Matsuishi et al., 2008; Theunissen et al., 2013; Van Leeuwen et al., 2006; Woerner et al., 2004), though it was unclear from these studies whether this was due to the interpretation of the underlying construct or short subscale length (Nunnally & Bernstein, 1994; Streiner, 2003). In our earlier report (D’Souza et al., 2017), we found a low alpha value but acceptable MIC coefficient for the mother-rated peer problems subscale, indicating that the low Cronbach’s alpha may have just been due to short subscale length. However, as the MIC coefficient obtained for the father-rated peer problems subscale was less than the acceptable value of 0.20 (Nunnally & Bernstein, 1994; Piedmont, 2014), our results suggest that when fathers are used as raters, the peer problems subscale appears to show genuinely poor internal consistency, regardless of subscale length.

We found moderate correlations between mothers’ and fathers’ ratings on all SDQ measures. Moderate to large correlations between mothers’ and fathers’ ratings have been found in previous research with the SDQ, with larger correlations typically observed for externalizing measures such as conduct problems and hyperactivity–inattention (Borg et al., 2012; Davé et al., 2008; Griffith et al., 2014; Mellor et al., 2011). In our sample, the strongest correlations were found for total difficulties and conduct problems, though the agreement for the hyperactivity–inattention subscale was comparable to that of the other subscales. However, unlike past research, large correlations were not observed for any of the SDQ measures. This is likely due to the young age of the cohort, as the studies noted above have only been conducted with children aged 4 years or older. Indeed, a meta-analysis of studies assessing interparental agreement on measures of child behavior found that agreement between parents gets stronger as children get older, with low to moderate agreement found for preschool children and moderate to strong correlations observed for those in middle childhood and adolescence (Duhig et al., 2000).

As expected, mothers and fathers differed in their child’s SDQ ratings. Mothers generally gave their children lower scores on the emotional symptoms, hyperactivity–inattention, and total difficulties measures when compared with fathers. These differences could be due to mothers being more habituated to difficult behaviors in their children than fathers, likely as a result of spending more time with their children at this age (Craig, 2006). Nevertheless, these differences were notably small in magnitude and may therefore not be of any meaningful significance. Mothers also gave their children higher scores on the original prosocial behavior subscale relative to fathers. However, as our factor analyses do not support the use of this subscale, we caution against interpreting this finding.

There are some limitations with the current study. We have not evaluated the sensitivity and specificity of the difficulties measures against formal clinical diagnoses, and whether this differs across mother- and father-reports, as very few children in the cohort have received diagnoses for disorders at such a young age. Additionally, although mothers and children from the Growing Up in New Zealand cohort were broadly generalizable to the current NZ population, the sample of fathers recruited is biased (Morton et al., 2010, 2013). In the current study, children without father-rated SDQ data were more likely to live in high deprivation areas, have a mother of Māori or Pacific ethnicity, and a father of Māori, Pacific, or Asian ethnicity. Consequently, these particular groups are not adequately represented in our study. These children were also more likely to have higher mother-rated SDQ scores on all difficulties subscales; therefore, the lack of father-rated data for these individuals may have skewed the current study’s results, particularly when considering the agreement and differences between mothers’ and fathers’ ratings. Furthermore, we were only able to make comparisons for children with mother- and father-rated data, which reduced our sample size.

It is also important to note that while mothers and fathers generally completed their interviews independently, 36% were not follow-on interviews (i.e., there was a gap in time between when mothers and fathers were interviewed), either due to parents not living within the same household or time constraints. Therefore, we cannot rule out the possibility that parents may have discussed the contents of the interview within that time. However, given that a considerable amount of information was collected during the interviews, we believe that it is unlikely that one parent would have remembered enough details about the SDQ to strongly influence the other parent’s answers. Nevertheless, this is a limitation to be aware of.

In spite of these limitations, our results suggest that agreement was still satisfactory and consistent with what we would expect based on past research, and differences between mothers’ and fathers’ ratings were fairly small. Furthermore, few longitudinal studies collect behavioral data from both parents, or adequately consider the agreement between parents’ ratings and measurement invariance of the tools used. Additionally, our study adds to rather limited research on the preschool version of the SDQ and, to our knowledge, is the first study to exclusively evaluate father-ratings of a preschool SDQ.

This article examined the factor structure and internal consistency of the preschool SDQ, measured in 2-year-old children and completed by the children’s fathers. Similar to the mother-rated preschool SDQ, we found support for a modified five-factor model, which accounts for a positive construal factor. Aside from the peer problems subscale, SDQ measures also showed good internal consistency. Importantly, we found measurement invariance of the modified five-factor model across mothers and fathers, suggesting that both informants are reporting the same types of strengths and difficulties. We also found moderate agreement in the ratings of mothers and fathers, and differences in parents’ ratings were small. Therefore, our results suggest that clinicians and researchers can be justified in using just one parent or caregiver as an informant on the preschool SDQ, though the peer problems subscale from either parent should be used with caution. Nevertheless, the use of multiple informants is encouraged in general and, more specifically, more research is needed to determine whether the sensitivity and specificity of these measures in screening for psychiatric illnesses differ across mothers and fathers.

Footnotes

Acknowledgements

The authors acknowledge the contributions of the original study investigators: Susan M. B. Morton, Polly E. Atatoa Carr, Cameron C. Grant, Arier C. Lee, Dinusha K. Bandara, Jatender Mohal, Jennifer M. Kinloch, Johanna M. Schmidt, Mary R. Hedges, Vivienne C. Ivory, Te Kani R. Kingi, Renee Liang, Lana M. Perese, Elizabeth Peterson, Jan E. Pryor, Elaine Reese, Elizabeth M. Robinson, Karen E. Waldie, Clare R. Wall.

Authors’ Note

The study has been designed and conducted by the Growing Up in New Zealand study team, led by the University of Auckland. The views reported in this article are those of the authors and do not necessarily represent the views of the Growing Up in New Zealand investigators.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Growing Up in New Zealand has been funded by the New Zealand Ministries of Social Development, Health, Education, Justice and Pacific Island Affairs; the former Ministry of Science Innovation and the former Department of Labour (now both part of the Ministry of Business, Innovation and Employment); the former Ministry of Women’s Affairs (now the Ministry for Women); the Department of Corrections; the Families Commission (now known as the Social Policy Evaluation and Research Unit); Te Puni Kokiri; New Zealand Police; Sport New Zealand; the Housing New Zealand Corporation; and the former Mental Health Commission, The University of Auckland, and Auckland UniServices Limited. Other support for the study has been provided by the NZ Health Research Council, Statistics New Zealand, the Office of the Children’s Commissioner and the Office of Ethnic Affairs.

References

Åman-Back

Björkqvist

(2004). Parents’ assessments of how much time they spend with their children at different ages. Psychological Reports, 94, 1025-1030. doi:10.2466/pr0.94.3.1025-1030

Bardone

A. M.

Moffitt

T. E.

Caspi

Dickson

Stanton

W. R.

Silva

P. A.

(1998). Adult physical health outcomes of adolescent girls with conduct disorder, depression, and anxiety. Journal of the American Academy of Child & Adolescent Psychiatry, 37, 594-601. doi:10.1097/00004583-199806000-00009

Becker

Woerner

Hasselhorn

Banaschewski

Rothenberger

(2004). Validation of the parent and teacher SDQ in a clinical sample. European Child & Adolescent Psychiatry, 13(Suppl. 2), ii11-ii16. doi:10.1007/s00787-004-2003-5

Bentler

P. M.

(1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238-246.

Bittner

Egger

H. L.

Erkanli

Jane Costello

Foley

D. L.

Angold

(2007). What do childhood anxiety disorders predict? Journal of Child Psychology and Psychiatry, 48, 1174-1183. doi:10.1111/j.1469-7610.2007.01812.x

Borg

A.-M.

Kaukonen

Salmelin

Joukamaa

Tamminen

(2012). Reliability of the Strengths and Difficulties Questionnaire among Finnish 4–9-year-old children. Nordic Journal of Psychiatry, 66, 403-413.

Caspi

Moffitt

T. E.

Newman

D. L.

Silva

P. A.

(1996). Behavioral observations at age 3 years predict adult psychiatric disorders: Longitudinal evidence from a birth cohort. Archives of General Psychiatry, 53, 1033-1039.

Cheung

G. W.

Rensvold

R. B.

(2002). Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling, 9, 233-255.

Chiorri

Hall

Casely-Hayford

Malmberg

L.-E.

(2016). Evaluating measurement invariance between parents using the Strengths and Difficulties Questionnaire (SDQ). Assessment, 23, 63-74.

10.

Cohen

(1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.

11.

Costello

E. J.

Mustillo

Erkanli

Keeler

Angold

(2003). Prevalence and development of psychiatric disorders in childhood and adolescence. Archives of General Psychiatry, 60, 837-844.

12.

Craig

(2006). Does father care mean fathers share? A comparison of how mothers and fathers in intact families spend time with children. Gender & Society, 20, 259-281. doi:10.1177/0891243205285212

13.

Croft

Stride

Maughan

Rowe

(2015). Validity of the Strengths and Difficulties Questionnaire in preschool-aged children. Pediatrics, 135, e1210-e1219. doi:10.1542/peds.2014-2920

14.

Davé

Nazareth

Senior

Sherr

(2008). A comparison of father and mother report of child behaviour on the Strengths and Difficulties Questionnaire. Child Psychiatry and Human Development, 39, 399-413.

15.

Dickey

W. C.

Blumberg

S. J.

(2004). Revisiting the factor structure of the Strengths and Difficulties Questionnaire: United States, 2001. Journal of the American Academy of Child & Adolescent Psychiatry, 43, 1159-1167. doi:10.1097/01.chi.0000132808.36708.a9

16.

Doi

Ishihara

Uchiyama

(2014). Reliability of the Strengths and Difficulties Questionnaire in Japanese preschool children aged 4-6 years. Journal of Epidemiology, 24, 514-518. doi:10.2188/jea.JE20140050

17.

D’Souza

Waldie

K. E.

Peterson

E. R.

Underwood

Morton

S. M.

(2017). Psychometric properties and normative data for the preschool Strengths and Difficulties Questionnaire in two-year-old children. Journal of Abnormal Child Psychology, 45, 345-357. doi:10.1007/s10802-016-0176-2

18.

Kou

Coghill

. (2008). The validity, reliability and normative scores of the parent, teacher and self report versions of the Strengths and Difficulties Questionnaire in China. Child and Adolescent Psychiatry and Mental Health, 2, 8. doi:10.1186/1753-2000-2-8

19.

Duhig

A. M.

Renk

Epstein

M. K.

Phares

(2000). Interparental agreement on internalizing, externalizing, and total behavior problems: A meta-analysis. Clinical Psychology: Science and Practice, 7, 435-453.

20.

Egger

H. L.

Angold

(2006). Common emotional and behavioral disorders in preschool children: Presentation, nosology, and epidemiology. Journal of Child Psychology and Psychiatry, 47, 313-337. doi:10.1111/j.1469-7610.2006.01618.x

21.

Eisenberg

Fabes

R. A.

Karbon

Murphy

B. C.

Wosinski

Polazzi

. . . Juhnke

(1996). The relations of children’s dispositional prosocial behavior to emotionality, regulation, and social functioning. Child Development, 67, 974-992.

22.

Ezpeleta

Granero

de la Osa

Penelo

Domènech

J. M.

(2013). Psychometric properties of the Strengths and Difficulties Questionnaire^3–4 in 3-year-old preschoolers. Comprehensive Psychiatry, 54, 282-291. doi:10.1016/j.comppsych.2012.07.009

23.

Fergusson

D. M.

Horwood

L. J.

Ridder

E. M.

(2005). Show me the child at seven: The consequences of conduct problems in childhood for psychosocial functioning in adulthood. Journal of Child Psychology and Psychiatry, 46, 837-849. doi:10.1111/j.1469-7610.2004.00387.x

24.

Fergusson

D. M.

Lynskey

M. T.

(1998). Conduct problems in childhood and psychosocial outcomes in young adulthood: A prospective study. Journal of Emotional and Behavioral Disorders, 6, 2-18. doi:10.1177/106342669800600101

25.

Goodman

(1997). The Strengths and Difficulties Questionnaire: A research note. Journal of Child Psychology and Psychiatry, 38, 581-586.

26.

Goodman

(2001). Psychometric properties of the Strengths and Difficulties Questionnaire. Journal of the American Academy of Child & Adolescent Psychiatry, 40, 1337-1345.

27.

Griffith

G. M.

Hastings

R. P.

Petalas

M. A.

(2014). Brief report: Fathers’ and mothers’ ratings of behavioral and emotional problems in siblings of children with autism spectrum disorder. Journal of Autism and Developmental Disorders, 44, 1230-1235.

28.

Hawes

D. J.

Dadds

M. R.

(2004). Australian data and psychometric properties of the Strengths and Difficulties Questionnaire. Australian and New Zealand Journal of Psychiatry, 38, 644-651.

29.

J. P.

Burstein

Schmitz

Merikangas

K. R.

(2013). The Strengths and Difficulties Questionnaire (SDQ): The factor structure and scale validation in U.S. adolescents. Journal of Abnormal Child Psychology, 41, 583-595. doi:10.1007/s10802-012-9696-6

30.

Hill

C. R.

Hughes

J. N.

(2007). An examination of the convergent and discriminant validity of the Strengths and Difficulties Questionnaire. School Psychology Quarterly, 22, 380-406. doi:10.1037/1045-3830.22.3.380

31.

Hofstra

M. B.

van der Ende

Verhulst

F. C.

(2002). Child and adolescent problems predict DSM-IV disorders in adulthood: A 14-year follow-up of a Dutch epidemiological sample. Journal of the American Academy of Child & Adolescent Psychiatry, 41, 182-189. doi:10.1097/00004583-200202000-00012

32.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1-55.

33.

Klein

A. M.

Otto

Fuchs

Zenger

von Klitzing

(2013). Psychometric properties of the parent-rated SDQ in preschoolers. European Journal of Psychological Assessment, 29, 96-104. doi:10.1027/1015-5759/a000129

34.

C.-H.

(2016). Confirmatory factor analysis with ordinal data: Comparing robust maximum likelihood and diagonally weighted least squares. Behavior Research Methods, 48, 936-949. doi:10.3758/s13428-015-0619-7

35.

Matsuishi

Nagano

Araki

Tanaka

Iwasaki

Yamashita

. . . Kakuma

(2008). Scale properties of the Japanese version of the Strengths and Difficulties Questionnaire (SDQ): A study of infant and school children in community samples. Brain & Development, 30, 410-415. doi:10.1016/j.braindev.2007.12.003

36.

Mellor

Stokes

(2007). The factor structure of the Strengths and Difficulties Questionnaire. European Journal of Psychological Assessment, 23, 105-112. doi:10.1027/1015-5759.23.2.105

37.

Mellor

Wong

(2011). Interparent agreement on the strengths and difficulties questionnaire: A Chinese study. Journal of Clinical Child & Adolescent Psychology, 40, 890-896.

38.

Morton

S. M. B.

Atatoa Carr

P. E.

Bandara

D. K.

Grant

C. C.

Ivory

V. C.

Kingi

T. R.

. . . Waldie

K. E.

(2010, November). Growing Up in New Zealand: A longitudinal study of New Zealand children and their families (Report 1: Before we are born). Retrieved from https://cdn.auckland.ac.nz/assets/growingup/research-findings-impact/report01.pdf

39.

Morton

S. M. B.

Carr

P. E. A.

Grant

C. C.

Robinson

E. M.

Bandara

D. K.

Bird

. . . Wall

(2013). Cohort profile: Growing up in New Zealand. International Journal of Epidemiology, 42, 65-75. doi:10.1093/ije/dyr206

40.

Morton

S. M. B.

Grant

C. C.

Carr

P. E. A.

Robinson

E. M.

Kinloch

J. M.

Fleming

C. J.

. . . Liang

(2014). How do you recruit and retain a prebirth cohort? Lessons learnt from Growing Up in New Zealand. Evaluation & the Health Professions, 37, 411-433. doi:10.1177/0163278712462717

41.

Morton

S. M. B.

Ramke

Kinloch

Grant

C. C.

Carr

P. A.

Leeson

. . . Robinson

(2015). Growing Up in New Zealand cohort alignment with all New Zealand births. Australian and New Zealand Journal of Public Health, 39, 82-87. doi:10.1111/1753-6405.12220

42.

Nunnally

J. C.

Bernstein

I. H.

(1994). Psychological theory. New York, NY: McGraw-Hill.

43.

Odgers

C. L.

Caspi

Broadbent

J. M.

Dickson

Hancox

R. J.

Harrington

. . . Moffitt

T. E.

(2007). Prediction of differential adult health burden by conduct problem subtypes in males. Archives of General Psychiatry, 64, 476-484. doi:10.1001/archpsyc.64.4.476

44.

Peterson

E. R.

Dando

D’Souza

Waldie

K. E.

Mohal

Morton

(2016). Can the new five factor model of infant temperament be used to predict which kids will be the terrible twos? Manuscript under submission.

45.

Piedmont

R. L.

(2014). Inter-item correlations. In Michalos

A. C.

(Ed.), Encyclopedia of quality of life and well-being research (pp. 3303-3304). Dordrecht, Netherlands: Springer.

46.

Pornprasertmanit

Miller

Schoemann

Rosseel

(2013). semTools: Useful tools for structural equation modeling (R Package version 0.3-0). Retrieved from https://www.rdocumentation.org/packages/semTools/versions/0.3-0

47.

Pryor

Morton

Bandara

Robinson

Grant

(2014). Pregnant partners: Fathers of the Growing Up in New Zealand children. Journal of Family Studies, 20, 5-18. doi:10.5172/jfs.2014.20.1.5

48.

Riso

D. D.

Salcuni

Chessa

Raudino

Lis

Altoè

(2010). The Strengths and Difficulties Questionnaire (SDQ). Early evidence of its reliability and validity in a community sample of Italian children. Personality and Individual Differences, 49, 570-575. doi:10.1016/j.paid.2010.05.005

49.

Rosseel

(2012). lavaan: An R package for structural equation modeling. Journal of Statistical Software, 48(2), 1-36.

50.

Rothbart

M. K.

(2011). Becoming who we are: Temperament and personality in development. New York, NY: Guilford Press.

51.

Rothbart

M. K.

Ahadi

S. A.

Hershey

K. L.

(1994). Temperament and social behavior in childhood. Merrill-Palmer Quarterly, 40, 21-39.

52.

Smits

I. A. M.

Theunissen

M. H. C.

Reijneveld

S. A.

Nauta

M. H.

Timmerman

M. E.

(2016). Measurement invariance of the parent version of the Strengths and Difficulties Questionnaire (SDQ) across community and clinical populations. European Journal of Psychological Assessment. Advance online publication. doi:10.1027/1015-5759/a000339

53.

Stone

L. L.

Otten

Ringlever

Hiemstra

Engels

R. C. M. E.

Vermulst

A. A.

Janssens

J. M. A. M

. (2013). The parent version of the Strengths and Difficulties Questionnaire: Omega as an alternative to alpha and a test for measurement invariance. European Journal of Psychological Assessment, 29, 44-50. doi:10.1027/1015-5759/a000119

54.

Streiner

D. L.

(2003). Starting at the beginning: An introduction to coefficient alpha and internal consistency. Journal of Personality Assessment, 80, 99-103.

55.

Theunissen

M. H.

Vogels

A. G.

de Wolff

M. S.

Reijneveld

S. A.

(2013). Characteristics of the strengths and difficulties questionnaire in preschool children. Pediatrics, 131, e446-e454.

56.

van de Looij-Jansen

P. M.

Goedhart

A. W.

de Wilde

E. J.

Treffers

P. D.

(2011). Confirmatory factor analysis and factorial invariance analysis of the adolescent self-report Strengths and Difficulties Questionnaire: How important are method effects and minor factors? British Journal of Clinical Psychology, 50, 127-144.

57.

Van Leeuwen

Meerschaert

Bosmans

De Medts

Braet

. (2006). The Strengths and Difficulties Questionnaire in a community sample of young children in Flanders. European Journal of Psychological Assessment, 22, 189-197. doi:10.1027/1015-5759.22.3.189

58.

Van Roy

Veenstra

Clench-Aas

. (2008). Construct validity of the five-factor Strengths and Difficulties Questionnaire (SDQ) in pre-, early, and late adolescence. Journal of Child Psychology and Psychiatry, 49, 1304-1312. doi:10.1111/j.1469-7610.2008.01942.x

59.

Woerner

Becker

Rothenberger

(2004). Normative data and scale properties of the German parent SDQ. European Child & Adolescent Psychiatry, 13(Suppl. 2), ii3-ii10. doi:10.1007/s00787-004-2002-6

60.

Youth in Mind. (2014). SDQ: Information for researchers and professionals about the Strengths and Difficulties Questionnaires. Retrieved from http://www.sdqinfo.com/