Only studies that administered the SDQ Parent-Report version were included in the summary below. Given the large number of studies that involved the SDQ, not all were reviewed. We focused the review on the use of the measure with trauma-exposed and diverse populations. Although studies are grouped by headings (e.g., “USE WITH TRAUMA-EXPOSED POPULATIONS” and “USE IN OTHER COUNTRIES”), there is a lot of overlap between categories.

The SDQ Parent version, along with other SDQ versions, has been used in many studies of conduct disorder and behavior problems including twin studies examining genetic and environmental influences (e.g., Button, Scourfield, Martin, Purcell, & McGuffin, 2005; Kuntsi, Rijsdijk, Ronald, Asherson, & Plomin, 2005; Saudino, Ronald, & Plomin, 2005; Scourfield, Martin, Eley, McGuffin, & Cherny, 2004; Scourfield, Van den Bree, Martin, & McGuffin, 2004).

CORRELATIONS WITH OTHER MEASURES/OUTCOMES
The Parent Report on SDQ was found to correlate significantly with Parent Report on the Rutter (Rutter Total Deviance and SDQ Total Difficulties: r=.88; Goodman, 1997).

Multiple studies (including those cited below under “USE IN OTHER COUNTRIES”) have examined the relation between the SDQ and Child Behavior
Checklist (CBCL). Goodman & Scott (1999) studied the predictive validity of SDQ and CBCL using ROC curves. Both measures discriminated well between high- and low-risk samples, with no significant differences between the measures in terms of predictive validity, assessed using area under the curve.

They found high correlations between the CBCL and SDQ: Total (r=.87),
Externalizing to Conduct (r=.84), Hyperactivity/Inattention (r=.71),
Internalizing/Emotional (r=.74), and Social/Peer (r=.59).

They also examined correlations between these measures and the Parent Account of Child Symptoms, a semi-structured interview about a child’s
emotional and behavioral symptoms. The measures correlated similarly, except that the SDQ Hyperactivity Scale correlated significantly better with the Interview scale than with the CBCL Inattention scale.

1. Parent SDQ scores were significantly correlated with teacher’s ratings of Anxiety, Aggression, and Hyperactivity/Inattention in the expected direction and with ratings of symptom severity of clinician made Axis I diagnoses (Hawes & Dadds, 2004).

2. Diagnoses made using the SDQ (Parent and Teacher reports combined) were compared to clinicians’ diagnoses of DSM-IV disorders (Mathai, Anderson, & Bourne, 2004). Significant correlations were found between clinical diagnoses and SDQ prediction (Hyperactivity Disorder Kendall’s tau-b = .44, p<.001; Conduct Disorder = .56, p<.001; Emotional Disorder = .39, p<.001).

3. Indicators of risk, assessed by high SDQ symptom scores, parents’ perception of definite or severe difficulties, or high symptoms plus impairment
were related to service contact/use in a large American nationwide sample (Bourdon et al., 2005).

4. McDermott (2005) reported a week correlation between the SDQ Total and PTSD-RI (r=.22, p<.01) in a sample of children who attended a school in a “wildfire affected” area. Exposure to threat from the fire was related to SDQ scores, particularly on the Total and Emotional scales.

5. The Burden scale from the impact supplement has been found to correlate significantly (r=.74) with the Parent Burden Interview (Goodman, 1999).

DISCRIMINANT VALIDITY
1. The SDQ (Parent Report) has been found to discriminate between groups of children. ROC analyses showed that the SDQ and Rutter questionnaires have equivalent predictive validity, with respect to their ability to discriminate between psychiatric and dental clinic samples (Goodman, 1997).

2. Another study found that compared to non-institutionalized children, institutionalized children scored higher on the SDQ total score and all problem subscales and lower on Prosocial Behavior (Muris & Maas, 2004).

3. Children diagnosed with an externalizing disorder have also been found to score higher on SDQ Peer Problems, Hyperactivity/Inattention, and Conduct Problem and lower on Prosocial Behaviors than did nonanxious and nonclinical groups. The anxious children scored higher than the nonclinical group did (Lyneham & Rapee, 2005).

4. A study of 40 preschool children classified as “hard to manage” based on maternal SDQ scores and 40 comparison children (mean age=52 months), showed that these children differed on observational data gathered from interactions with peers. The “hard to manage” group showed higher rates of Antisocial Behavior and Negative Emotions, and lower rates of Empathic/Prosocial responses (Hughes, White, Sharpen, & Dunn, 2000).

5. Examination of the validity of the Impact Supplement showed that Perceived Difficulties, Impact Score, and Burden ratings (Parent, Teacher, and Child) were significantly different for community versus clinic samples. Impact scores were better at discriminating between clinic and community samples than were Symptom scores (Goodman, 1999).

TREATMENT OUTCOME
Multiple studies have used the Parent SDQ as an outcome measure, with the SDQ showing sensitivity, at least on some scales, to change in studies of children with behavioral problems (e.g., Anderson, Vostanis, & O’Reilly, 2005; Callaghan, Young, Pace, & Vostanis, 2004; Costin, Lichte, Hill-Smith, Vance, & Luk, 2004; Mathai, Anderson, & Bourne, 2003).

1. In a small sample of Chinese Cantonese-speaking parents in Sydney, Australia, significant changes were found on the SDQ Prosocial Behavior
postintervention (Crisante & Ng, 2003).

2. Comparable change was also found on the on the SDQ and the CBCL for an evaluation of a parenting program with Irish families (Behan, Fizpatrick, Sharry, Carr, & Waldron, 2001).

3. Significant change has also been found for inpatient samples (e.g., Gavida- Payne, Littlefield, Hallgren, Jenkins, & Coventry, 2003).

4. Mathai et al. (2003) also found significant change using the Impact Supplement with significant reduction in parents’ perception of Overall
Difficulties, Level of Burden Placed by the Child on the Family, and Impact scores.

5. While the majority of these studies have been conducted using pre- to posttest comparisons, the SDQ Parent Report also appears to be sensitive to change in randomized trials. For example, a study of Cantonese-speaking parents randomly assigned to the Triple P Parenting Program or a wait list group showed significant effects on SDQ Conduct Problem, Hyperactivity, Peer Problems, and Emotional Symptoms (Leung, Sanders, Leung, Mak, & Lau, 2003).

FACTOR ANALYSIS
As described below, under “USE IN OTHER COUNTRIES” a number of factor analytic studies have been conducted with somewhat divergent results.
1. Goodman (2001) reported that, as hypothesized, a 5-factor solution was found for Parent, Teacher, and Child Self-Report using eigen values>1 to
determine the number of factors.

2. Studies in Australia (Hawes & Dadds, 2004), Germany (Becker, Woerner, Hasselhorn, Banaschewski, and Rothenberger, 2004; Woerner, Becker, &
Rothenberger, 2004), Sweden (Smedje, Broman, Hetta, & von Knorring, 1999), and the Netherlands (Muris, Meesters, & van den Berg, 2003) reported a similar factor structure as that reported by Goodman (2001).

3. However, studies in Gaza (Thabet, Stretch, & Vostanis, 2000) and the United States (Dicky & Wayne, 2004) suggest that the factor structure proposed by Goodman (2001) has a somewhat variable and questionable fit.

4. Dickey & Wayne (2004) examined the factor structure of the SDQ Parent Report using data from 9,577 American children aged 4-17. Principal
components analysis suggested a 5-factor solution.

The first 3 factors (Hyperactivity/Inattention, Emotional Symptoms, and Prosocial Problems) consisted of items that were intended to fit those domains, but the conduct problems and peer problems did not factor as hoped for.

An Exploratory Factor analysis using eigen values>1 and a promax rotation, suggested a 3-factor solution, titled Prosocial Problems, Externalization
Problems, And Internalization Problems. A Confirmatory Factor analysis provided support for the 3-factor model.

The authors suggested that for American children the 3-factor model may be more appropriate than were the 5 factors found typically in British samples.

USE WITH DIVERSE SAMPLES
1. All versions of the SDQ (Child, Parent, and Teacher) were found to have acceptable internal consistency and validity and to be considered a robust
measure for children and adolescents with intellectual disabilities (Emerson, 2005).

2. The use of the SDQ Teacher and Parent versions was examined in a sample of strictly Orthodox preschool children aged 3 to 4 (Lindsey, Frosh, Loewenthal, & Spitzer, 2003).

USE IN OTHER COUNTRIES
The SDQ has been used in many countries to examine rates of psychopathology.
1. For example, the SDQ was used to screen for prevalence of psychopathology in a sample of 448 Russian children. Prevalence rates were generally higher for the Russian sample than for the comparison British sample (Goodman, Slobodskaya, & Knyazev, 2005).

2. Woerner et al. (2004) reported on the use of the SDQ overseas (beyond Europe) in Brazil, Canada, the Middle East, Asia, and Australia. They report that
the data provides support for the psychometric properties of the measure.

BRAZIL
1. Cury & Golfeto (2003) used Brazilian Teacher and Parent versions of the SDQ and suggested that the SDQ may be useful for preliminary screening of possible psychiatric disorders. The article was not reviewed, as it is in Portuguese.

2. The SDQ (Parent, Teacher, and Child versions) was also used in another study that examined child mental health problems in a rural African-Brazilian community (Goodman, dos Santos, Nunes, de Miranda, Fleitlich-Bilyk, & Filho, 2005). The authors report significant agreement between the SDQ and the Development and Well-Being Assessment (DAWBA).

AUSTRALIA
1. The SDQ has been used in multiple studies of Australian children. Principal components analysis of 1,359 Australian children aged 4-9, with an oblimin rotation suggested the presence of 5 factors, consistent with that found by Goodman (2001) (Hawes & Dadds, 2004). Analyses showed that for boys, Hyperactivity accounted for most of the total variance (22.4%) while for girls, Prosocial Problems accounted for the most variance (19.7%).

GAZA
1. A study of Arab children living in the Gaza Strip suggests that the standard factor structure may not be appropriate for these children and that certain items appeared to have different meaning for these participants compared to Western participants (Thabet, Stretch, & Vostanis, 2000).

BANGLADESH
1. Mullick & Goodman (2001) examined the psychometrics of a Bangla version
(translated and backtranslated) with a sample of 99 clinic and 162 community
Bangladeshi children aged 4-16.

They found that SDQ scores distinguish between community and clinic samples,
and between children with different psychiatric diagnoses. Using ROC curves for
each SDQ scale, AUC (Area under curve) = >.80 were found for Total Impact,
Conduct Problems, and Hyperactivity. For Parent, Teacher, and Child reports,
Emotional Symptoms were able to distinguish between clinic cases with and
without an emotional disorder; Conduct Problems were able to distinguish
between clinic cases with and without conduct disorder; and Hyperactivity was
able to distinguish between those with and without a hyperactivity disorder.

SOUTHERN EUROPEAN COUNTRIES
1. Marzocchi, Capron, Di Pietro, Tauleria, Duyme, Frigerio, et al. (2004) described the use of the SDQ in Southern European countries (Italy, Spain,
Portugal, Croatia, France).

SPANISH SDQ
The Spanish version of the SDQ has been used in a number of studies.
1. García, Goodman, Mazaira, Torres, Rodríguez-Sacristán, Hervas & Fuentes (2000) reported on the initial psychometrics comparing the SDQ with the CBCL and Child Behavior Questionnaire.

2. García Cortázar, Mazaira, & Goodman (2000) examined the psychometrics of the Spanish Parent and Teacher SDQs in a sample of 132 clinic children and 48 pediatric patients in Spain. The abstract of the article suggests that the SDQ discriminated between the two groups and had satisfactory validity.

GREEK
1. Bibou-Nakou, Kiosseoglou, & Stogiannidou (2001) examined the correspondence between Teacher and Parent ratings on the SDQ in a sample of
Greek children. Difficulties scores according to Teacher Report are related to school achievement, and according to Parent Report are related to family
dysfunction.

NORDIC COUNTRIES
A review article on the use of the SDQ in Nordic countries (Obel, Heiervang, Odriguez, Heyerdahl, Smedje, Sourander, et al., 2004)Obel et al., 2004)
suggested that the distributions of the SDQ are similar across countries and suggested collaboration in developing norms for Nordic countries. The authors described the use of the SDQ in Sweden, Finland, Norway, Denmark, and Iceland, detailing studies in each of these countries that had used the SDQ.

Swedish
1. Smedje, Broman, Hetta, & von Knorring (1999) reported on the psychometrics of the Swedish SDQ. Internal consistency for the total scale was
.76, with alphas ranging from .51-.75 for subscales. Principal Components analysis suggested the presence of 5 factors based on eigen values>1 and interpretability. The authors provide means and cutoff scores.

2. The Swedish SDQ has also been found to differentiate between clinical and community samples, and showed good Sensitivity and Specificity (Malmberg, Rydell, & Smedge, 2003).

Finnish
1. Koskelainen, Sourander, & Kaljonen (2000) reported on the psychometrics of the Parent, Teacher, and Child SDQ in a sample of Finnish children aged 7-15 (n=735). They reported on the internal consistency for all three reporters as ranging from alpha=.63-.86. Teachers had the best internal consistency (M=.79) compared to Parents and Child (.67 and .65, respectively). Inter-rater reliability (correlations) ranged from .28-.40 for Child and Parents, .28-.38 for Child and Teachers, and .29-.45 for Parents and Teachers. The validity was supported through correlations with strong correlations with the CBCL and Youth Self- Report. For example, the Total CBCL and Parent SDQ were correlated at r=.75 and the Total Child Self-Report SDQ and YSR Total were correlated at .71.

Dutch
1. The psychometrics of the Dutch version of the SDQ has been reported on in at least two studies (Muris, Meesters, & van den Berg, 2003; van Widenfelt, Goedhart, Treffers, & Goodman, 2003). Muris et al. (2003) examined the psychometrics of the SDQ (Parent and Child versions) in a sample of 562 children aged 9 to 15 (M=12.3).

Factor analysis of the Parent SDQ suggested a 5-factor solution accounting for 47.6% of the variance, with all items loading on the intended factor and only 1 item having a substantial secondary loading.

TEST-RETEST RELIABILITY (average 2-month intraclass correlation):
Total Difficulties (.88); Emotional Symptoms (.76); Conduct Problems (.89);
Hyperactivity/Inattention (.84); Peer Problems (.91); Prosocial Behavior (.75)

INTERNAL CONSISTENCY (alpha):
Total Difficulties (.80); Emotional Symptoms (.70); Conduct Problems (.55);
Hyperactivity/Inattention (.78); Peer Problems (.66); Prosocial Behavior (.68)

INTERRATER RELIABIITY (correlations between Parent and child reports):
Total Difficulties (.46); Emotional Symptoms (.43); Conduct Problems (.31);
Hyperactivity/Inattention (.42); Peer Problems (.43); Prosocial Behavior (.21)

VALIDITY
The Parent SDQ correlated significantly with the CBCL, Child Depression Inventory (CDI), Revised Children’s Manifest Anxiety Scale (RCMAS), and
ADHD Questionnaire (ADHDQ), as expected. Specific correlations are reported in the article.

Germany
1. Woerner, Becker, & Rothenberger (2004) reported on psychometrics of the German SDQ Parent Report version. They report on normative data gathered from 930 children aged 6-16. SDQ scores differentiated between the community sample and a clinical sample.

2. The factor structure also appears to be similar to what has been proposed by Goodman (2001) in English samples, and the SDQ scores correlate with scores on the CBCL and Youth Self-Report, as would be expected (Klasen et al., 2000; Klasen, Woerner, Rothenberger, & Goodman, 2003).

3. Additional data on the validity of the German SDQ are presented in Woerner
et al. (2002).

4. Becker, Woerner, Hasselhorn, Banaschewski, & Rothenberger (2004) examined the validity of the SDQ (Parent and Teacher reports) in a German
clinical sample. They report internal consistencies ranging from .72-.81 for the Parent and .75-.83 for the Teacher.

All correlations between SDQ subscales and corresponding CBCL/TRF scales were significant (p<.001). For example: SDQ Total Difficulties and CBCL Total Problems (r=.83), SDQ Emotional Problems and CBCL Internalizing Problems (r=.77), SDQ Conduct Problems and CBCL Externalizing (r=.82), SDQ Hyperactivity/Inattention and CBCL Attention Problems (r=.75), SDQ Prosocial Behavior and CBCL Social Problems =-.22.

Factor analysis combining additional Parent reports (Total Sample=1,686) resulted in a 5-factor solution explaining 53.9% of the variance matching the intended scale structure. ROC analysis was used to examine the discriminative validity of the SDQ and CBCL/TRF with respect to diagnosis.

SDQ Parent, SDQ Teacher, CBCL, and TRF were equally able to differentiate between patients with a clinical diagnosis and those without.

The SDQ Parent and SDQ Teacher were better at predicting children with ADHD than the Attention Problems Scale of the CBCL or TRF. The Internalizing CBCL Scale was better at detecting children with emotional disorders than was the Emotional Problems Scale of the SDQ.

KENYA
The Parent and Teacher SDQs have been used with a Kenyan, Luo-speaking sample to examine the relation between stress and behavior in children
orphaned because of AIDS.

USE WITH TRAUMA-EXPOSED SAMPLES
The SDQ has also been used in numerous studies of children in the British
foster care system who have most likely experienced traumas.

1. The SDQ has been used in a number of studies examining the effects of traffic accidents (e.g., Bryant, Mayou, Wiggs, Ehlers, & Stores, 2004 ). Bryant et al. (2004) reported that those children who were found to be in the abnormal or borderline range on the Emotional Symptoms Scale were more likely to have Acute Stress Disorder. Emotional symptoms were also associated with PTSD at 3-month follow-up but not at 6 months. At 6 months, Abnormal and Borderline scores on Conduct and Prosocial Behavior were more likely to have PTSD.

2. The SDQ was used to screen children who had experienced a wildfire (McDermott, Lee, Judd, & Gibbon, 2005). Internal consistency was .91 for the
Total sample of 222 children aged 8-18 who attended a school that had been in a “wildfire affected” area.

Internal consistency for different grade levels was as follows: primary school, grades 3-6 (.86); junior high, grades 7-9 (.78); senior high, grades 10-12 (.82). There was a small correlation between SDQ scores and scores on the PTSD Reaction Index (r=.22, p<.01).

The authors suggested that “the SDQ is not on its own a suitable measure for identifying trauma-related symptoms and should not be used in isolation to identify postdisaster trauma in children and adolescents” (p. 141).

3. Correlations between Parent and Child reports on the SDQ were examined in both an at-risk Norwegian sample and a community sample (Waaktaar, Borge, Christie, & Torgersen, 2005). The at-risk sample included children, many of whom had experienced traumas including death of a parent or sibling, war exposure, and refugee experiences.

Correlations for the At-Risk sample:
Total Problems (.43), Disruptive Difficulties (.52), Emotional Difficulties (.22),
Prosocial Behavior (.19)
Correlations for the Community sample:
Youth-Mother: Total Problems (.30), Disruptive Difficulties (.34), Emotional
Difficulties (.30), Prosocial Behavior (.16)
Youth-Father: Total Problems (.33), Disruptive Difficulties (.37), Emotional
Difficulties (.21), Prosocial Behavior (.16)
4. Parent and Child SDQs were collected on a sample of 43 children with end
stage renal disease. Means and SD are presented and compared to a
normative sample (Madden, Hastings, & Hoff, 2002).

5. The SDQ was used in a treatment outcome study of Spanish-speaking mothers and children exposed to domestic violence (Martín & Pérez, 2002). The original article could not be obtained for review.

6. The SDQ was used in a study of homeless children, some of whom had experienced domestic violence and others of whom had experienced neighborhood violence, with results showing high psychiatric comorbidity for both groups (Vostanis, Tischler, Cumella, & Bellerby, 2001).

7. The SDQ was used in a study of Greek children exposed to the Athens earthquake in September 1999 (Kolaitis, 2003). The authors found no significant differences between earthquake exposed and non-exposed children, although they did find differences between immigrant and non-immigrant children. It should be noted that earthquake-exposed children did differ from non-exposed children on the Child’s Report of Depression (CDI) and Anxiety (SCARED).