Emotional and Behavioral Difficulties, Communication, Intelligence, and Reasoning in Children with Hearing Difficulties

education sciences Article Emotional and Behavioral Difficulties, Communication, Intelligence, and Reasoning in Children with Hearing Difficulties Olga María Alegre de la Rosa 1,* and Luis Miguel Villar Angulo 2 1Department of Didactics and Educational Research, University of La Laguna, 38200 San Cristóbal de La Laguna, Santa Cruz de Tenerife, Canary Islands, Spain 2Department of Didactics and Educational Organization, University of Seville, 41004 Sevilla, Spain *Correspondence: [email protected]; Tel.: +34-92-231-9026 (ext. 38204) Received: 14 July 2019; Accepted: 29 August 2019; Published: 4 September 2019   Abstract: This study aims to investigate whether emotional and behavioral difficulties (EBD) differ between children with cochlear implants (CIs) or hearing aids (HAs), according to multi-informant ratings. Methods: A battery of psychological measures (e.g., Strengths and Difficulties Questionnaire (SDQ), Illinois Test of Psycholinguistic Abilities (ITPA), Peabody Image Vocabulary Test (PPVT), and Raven Progressive Matrices Test (RPM), was administered to children with CIs or HAs. The study involved 187 children with CIs, 113 children with HAs, 176 fathers and mothers, and 300 schoolteachers. Results: Significant disagreements were found between children, parents, and schoolteachers with regard to SDQ ratings. Total SDQ difficulties score correlated poorly with other outcomes (e.g., total ITPA, final PPVT, and final RPM). Regression linear analyses showed six socio-demographic and linguistic covariates that significantly predicted children’s total SDQ difficulties score. Conclusion: Several independent variables were found to be associated with children’s total EBD scores. This article highlights the importance of providing professionals with sufficient training so that they are knowledgeable about the procedures they are to use to support children with EBD (e.g., developing and implementing instructional approaches for school students with EBD). Keywords: cochlear implant; hearing aid; family; teacher; SDQ 1. Introduction This study attempts to detect individual differences between children who use two types of hearing assistive devices. The sensory information perceived by children with hearing assistive devices is essential to overcome the delay they exhibit in the quality of spoken language. So, children using hearing assistive devices have psychological difficulties related to their speech perception (Henner, Novogrodsky, Reis, & Hoffmeister [ 1 ]. From an educational perspective, the administration of psychological tests concerns researchers because children with insufficient linguistic development need to thoroughly understand test instructions. In fact, class teachers and speech-language pathologists are not the only educational agents that remedy the linguistic weaknesses of children with hearing assistive devices. Families also provide fluency in the primary oral communication, which sustains the basic perceptual scheme of the child’s future cognitive development. As Luft [ 2 ] (p. 2) has written: “All families function as the de facto language development interventionists for their children.” The mental health of children with hearing loss may be impacted by difficulties in communication (Stevenson, Kreppner, Pimperton, Worsfold, & Kennedy [ 3 ]. Moreover, Edwards, Hill, and Mahon [ 4 ] reported that the concept of communication and independence in children is problematic. Overall, the Educ. Sci. 2019,9, 234; doi:10.3390/educsci9030234 www.mdpi.com/journal/education Educ. Sci. 2019,9, 234 2 of 15 notion denotes students’ perceptual difficulties, or ascribes knowledge dependency and disability to children who use cochlear implants (CIs) or hearing aids (HAs). The equality or difference of children with CIs or HAs affects their perceptions of educational achievement, personal health, emotional well-being, and emotional and behavioral difficulties (EBD). To date, there has been no attempt to give a quantitative assessment of the magnitude of differences in the rates of EBD in general in children with CIs or HAs in the Canary Islands. 1.1. Measurement of Individual Differences This paper focuses specifically on the Strengths and Difficulties Questionnaire (SDQ) because it provides “well-validated sub-scale scores, with equivalent ratings based on parent, teacher and self-report which can be analyzed separately but in parallel” [ 3 ] (p. 478). Spanish analyses also have demonstrated the structure and reliability of SDQ [5,6]. The questionnaire SDQ is unique in that it derives essential understandings of students’ interpersonal relations. The SDQ is conceived as an instrument to perform the screening of positive socialization behavior and the psychopathology of 3–16-year-old children. Many health clinics now use SDQ as part of the initial assessment of children with hearing loss getting parents, schoolteachers, and students to complete questionnaires prior to the first clinical assessment. As such, SDQ measures are useful for investigators to demonstrate, scientifically, the impact of EBD of children using hearing assistive devices on their instructional needs. In general, children with CIs are perceptually affected, and consequently, parents and schoolteachers are obliged to acquire and demonstrate new communication and language competencies. In fact, Huber and Kipman [ 7 ] (p. 152) used SDQ to assess the mental health condition of 32 adolescents with CIs and 212 typical hearing peers. From their study, an important conclusion was drawn: “Teachers rated those with CIs as having more peer problems, and more ‘total’ problems” (p. 152). On the contrary, Wong et al. [ 8 ] (p. 15) found that “children with CIs or HAs did not show significant emotional or behavioral difficulties compared with norms (SDQ).” Furthermore, educators’ challenges have been to develop strategies to activate the social skills of children with CIs. Similarly, Wong et al. [ 9 ] compared families’ and educators’ perceptions of the social behavior of children with CIs. The existing literature comprises conflicting reports concerning the cognitive capacity of children after implantation of hearing assistive devices. The specific evaluation of all or some of the psycholinguistic skills of children with CIs or HAs attracted the interest of some researchers when the Illinois Test of Psycholinguistic Abilities (ITPA) was used. Specifically, ITPA is a criterion measure to determine the psycholinguistic training of educators, providing a wide variety of diagnosed children’s cognitive abilities [ 10 ]. For example, Alegre, Rodr í guez, Villar, and P é rez [ 11 ] administered the ITPA and the Peabody Image Vocabulary Test (PPVT) and found that the children’s age at the time of early cochlear implantation was a good predictor of adequate linguistic development. Although children with CIs or HAs are able to solve complex instructional issues cognitively, the PPVT establishes how children’s age at the time of cochlear implantation is an essential predictor of speech and language outcomes in students with CIs [ 12 ]. Later studies have examined the extent to which children with unilateral and mild bilateral hearing loss are at risk of social skills difficulties, compared with typical hearing children and children with moderate to severe hearing loss [ 13 ]. After all, PPVT test reports predictors of language development in children with CIs [14,15]. Children with CIs—from the school curriculum perspective—experience reduced rates in school learning processes. They also display slower understandings of spoken language. Nevertheless, the identification of students with hearing loss who are academically proficient is an issue of great concern in educational theory and research. The use of the Raven Progressive Matrices Test (RPM) estimates the abstract reasoning fluency of children with CIs or HAs. Even though the test has a large number of items, researchers have reduced Raven’s test size [ 16 ] without losing their predictive validity, demonstrating “that items are largely measurement invariant” [ 17 ]. In retrospect, other scholars have reported on the limitations of the linguistic and cognitive processing of children with CIs compared Educ. Sci. 2019,9, 234 3 of 15 to their peers who hear complex semantic tasks [ 18 ]. For example, reports have demonstrated that children with CIs have an altered perception of music due to their auditory difficulties. However, Shirvani et al. [ 19 ] (pp. 475-476) used the Iranian version of 32 Raven elements to compare three groups of children (normal hearing, bimodal adjustment, and profound hearing loss and unilateral CIs) and found that “a weak perception of musical emotions in children with unilateral CIs can lead to weakness in the communicative message of music” (pp. 475–476). 1.2. Significance and Purposes of the Current Study Parents and schoolteachers of children with CIs or HAs deal with many challenges in making decisions regarding diagnostic teaching, creating informed opinions about children’s abstract reasoning, and promoting the development of spoken or sign language in children with hearing difficulties [ 9 , 20 ]. The aims of this study were: 1. To explore multiple inter-rater (i.e., class teachers–speech-language pathologists, children–school professionals, children–parents, and class teachers–parents) concordance correlation coefficients for SDQ data. Authors calculated agreement reports between students, parents, and schoolteachers on EBD of children with CIs or HAs to give homogeneity and consistency to scores [21]. 2. To uncover the functional relationships of EBD and psycholinguistic abilities, image vocabulary, and abstract reasoning of children with CIs or HAs [22]. To predict the children’s total SDQ score using the children’s psychological variables (e.g., ITPA, PPVT, and RPM). Furthermore, we aimed to predict the children’s total SDQ score by influences from other aggregated multi-informants’ socio-demographic predictors: Student-level covariates (e.g., gender; hearing device; individualized curriculum adaptation (ICA) instructional program; chronological and implantation age), school-level covariates (e.g., grade, school speech therapy (number of hours), tutor, and speech-language pathologist experience (number of years), family-level covariates (e.g., gender, studies, parents’ age (father and mother), and home communication), and geographical location-level covariate (e.g., administrative province of residence) [15] (Sarchet et al., 2014). 2. Methods Participants All eligible children in the Canary Islands were recruited to the study. The study cohort consisted of children with CIs (n=187) and HAs (n=113). They had access to the same hearing service provider (Canary Health Service) before they turned three years of age. One hundred and thirty-seven children were recruited from the island of Tenerife and 163 from the island of Gran Canary. The chronological ages of the children were between 6–9 years (n=104), 10–12 years (n=88), and 13–16 years (n=108). They attended inclusive school centers. One hundred and nineteen followed the ICA instructional program, while 181 enrolled in the regular curriculum of their typical peers. Etiological causes were excluded from the study because of lacking information. All parents and proxy families provided informed written consent for administering questionnaires and tests to each child with CIs or HAs. Further, ethical and legal disposition approval for the present study was attained from the Clinical Research Ethics Committee (C.E.I.C.) of the Canary Insular Maternal and Child University Hospital. The role of family-center care was essential for children’s communication [ 23 ] (Ching et al., 2018). Therefore, authors included parental education and family labor status as latent variables that might predict children’s strengths and psycho-educational problems [ 24 ] (Porter, Sladen, Ampah, Rothpletz, & Bess, 2013). Thus, fathers (n=87) and mothers (n=89) had primary education degrees, and they were working (parents n=230, mothers n=213). Fathers contributed data in the cycle of 41–50 years old (n=143), between 31–40 years old (n=95), more than 50 years (n=55), and between 20–30 years (n=7). Mothers’ age contributed data between 41–50 years old (n=143), between 31–40 years old (n=95), more than 50 years old (n=55), and between 20–30 years old (n=15). Educ. Sci. 2019,9, 234 4 of 15 On one hand, other authors have substantiated fathers’ active role in oral communication with their children, noting parents who utilized verbal communication (n=215), compared to those who employed only signs (n=28) or were bilingual (n=57). On the other hand, there was ambivalent data about the role played by school professionals who had been in contact with children with CIs or HAs. While class teachers display adequate attitudes towards the inclusion of children with hearing loss in regular school classrooms, school professionals need to prepare the teachers to deal with cochlear implant technology and other auditory issues [25,26]. There are generally a greater number of female teachers (speech-language pathologists (n=249) and class teachers (n=227)) compared to male teachers (speech-language pathologists (n=51) and class teachers (n=73)). The average age of speech-language pathologists was 50–59 years (n=171), and that of the class teachers was 60 or more (n=150). On average, both types of school educators had professional experience of 10 years or more: Speech-language pathologists (n=193) and class teachers (n=185). 3. Materials 3.1. Strengths and Difficulties Questionnaire (SDQ) Goodman [ 27 , 28 ] designed and developed the SDQ. All versions of the SDQ request about 25 attributes, some positive and some negative. SDQ is comprised of four subscales: Emotional problems, behavioral problems, problems with peers, and hyperactivity, which produce subscale scores that can be summed to produce a “total difficulties” score and a fifth separate prosocial scale score. The SDQ is an instrument to perform the screening of the positive behavior of socialization and the psychopathology of 4–16-year-old children and adolescents. The same 25 items are included in the questionnaires for parents or teachers of children to complete them. According to the developer, the SDQ is available in more than 77 language versions. The results of the factor analysis replicated by researchers from different countries validated the SDQ structure [ 5 , 29 ]. Cronbach’s alpha values are different in this research according to the groups studied: Children with CIs or HAs (0.789), speech-language pathologists (0.725), class teachers (0.616), and families (fathers and mothers) (0.759), which represented an adequate reliability value [30]. 3.2. Illinois Test of Psycholinguistic Abilities (ITPA) Before the translated version of the ITPA into Spanish, 68 studies had adopted this instrument between 1968 and 1974 to investigate children’s abilities [ 10 , 31 ]. It is composed of 10 subtests, for each of the following functions: Listening comprehension, visual comprehension, visual-motor sequential memory, auditory association, auditory sequential memory, visual association, visual integration, verbal expression, and grammatical integration. It allows the detection, prediction, and diagnosis of possible errors or difficulties in the child’s communication process. Some researchers have applied the whole instrument to children with CIs [ 32 ] or specific subtests for predicting individual differences in peer acceptance and friendship formation among children with specific language impairments [ 22 ]. Cronbach’s alpha value for internal consistency for this study is 0.852, which represents a reasonable reliability coefficient [30]. 3.3. Peabody. Image Vocabulary Test (PPVT) This interactive test [ 33 ] has two alternative forms (A and B) of a receptive vocabulary assessment. Each type includes 228 elements, equally distributed across 19 item-sets. Each item-set contains 12 elements of increasing difficulty. The administration time is limited to 10–15-min by applying the sets according to the child’s age or skill level. Children point to one out of four images that best represents the meaning of a verbally presented stimulus word. The average reliability coefficient for this test, based on the normative sample, was 0.89 [ 14 ]. However, the reported Cronbach’s alpha coefficient for this study was 0.641, which is considered a questionable coefficient [30]. Educ. Sci. 2019,9, 234 5 of 15 3.4. Raven Progressive Matrices Test (RPM) This instrument [ 34 , 35 ] has been distributed to deaf or hard of hearing children for the last 50 years [ 1 ]. It comprises 60 items that measure abstract reasoning, in which children are required to select the missing fragment from an incomplete plan. The reported Cronbach’s alpha coefficient for this study was 0.713, which seems an acceptable coefficient [30]. 4. Procedure Authors obtained written consent for the child’s participation in the study from each father, mother, or both. Parents, class teachers, speech-language pathologists, and children with CIs or HAs completed the SDQ and a socio-demographic background questionnaire either at home or school. The socio-demographic questionnaire for parents and class teachers sought information about gender, the age of hearing loss of children, communication mode, geographical location, and educational setting. Further, the family’s economic situation was measured using the parental employment status. Parents and the Gran Canary Hospital provided data on age of children in the first auditory fitting, the degree of hearing loss, type of hearing device, and age of cochlear implantation. Members of the research team (school psychologists) administered the ITPA, PPVT, and RPM tests in a quiet schoolroom or the child’s clinical setting. Each testing session lasted a morning with certain pauses. Collaboration between class teachers and speech-language pathologists supported children’s communication skills in the ICA instructional program [36]. Statistical Analysis First, descriptive statistics, which included means and standard deviations, were calculated. All statistical analyses selected an alpha level of 0.05. Second, the agreement rates between children and school professionals as well as school professionals and families were computed using Cohen’s kappa coefficient on SDQ subscales. Third, Pearson correlation coefficients examined the strength of the linear relationship between the responses of children with hearing assistive devices on the SDQ subscales, the total ITPA scores, and the final PPVT and RPM scores. Next, a separate Pearson correlation was performed for the class teachers and children with hearing assistive devices on the total SDQ and ITPA scores along with the final PPVT and RPM scores. The size of the effects for this context was d=0.2 which would be considered a “small” effect size, 0.5 a “medium” effect size, and 0.8 a “large” effect size [ 37 ]. Lastly, multiple regression analyses were conducted to examine the predictive power of independent variables in the SDQ total score. Socio-demographic factors and other psychological factors (e.g., total ITPA score, final PPVT and RPM scores, and a combination of SDQ subscale scores of families and class teachers) were the independent variables. All analyses were performed with SPSS statistics version 21. 5. Results 5.1. Examination of Inter-Rater Reliability of the SDQ Subscales with Cohen’s Kappa According to the study’s first aim, a mild agreement was found between speech-language pathologists and class teachers on the SDQ subscales, when they scored on the EBD of children with CIs or HAs. Results also disclosed a moderate agreement between both school professionals on the behavioral problems subscale for HAs (k=−0.405, p<0.001). Furthermore, a mild agreement was observed between school professionals and children on the total SDQ and subscales outcomes of children with HAs (k=0.415, p<0.001). Likewise, a moderate agreement appeared between children and speech-language pathologists, although this was exclusively attributed to the behavioral problems subscale of children with HAs (k=0.415, p<0.001). Similarly, a moderate agreement was revealed on the emotional problems subscale of children with CIs or class teachers (k=0.410, p<0.001). The poor agreements that can be observed in the other SDQ subscales Educ. Sci. 2019,9, 234 6 of 15 have led researchers to think about the class teachers’ limited awareness of the EBD of children with hearing loss. 5.2. Examination the Relationship between Five Subscales SDQ and ITPA, PPVT y RPM We found small associations between children’s self-report SDQ subscale scores (e.g., emotional problems slightly correlated with behavioral problems (r=0.325, p<0.001), which were not statistically correlated with Problems with peers (r=−0.122, p<0.05). (See Table 1). This result implied that when emotional problems increased, problems with peers decreased in a constant proportion. Correlation analyses indicated that behavioral problems demonstrated significant correlations with problems with peers (r=0.391, p<0.001) and hyperactivity (r=0.260, p<0.001), which was not statistically correlated with prosocial behavior (r= − 0.298, p<0.001). This score showed that when behavioral problems increased, prosocial behavior decreased in a constant proportion. Moreover, correlation coefficients indicated a statistically significant relationship between hyperactivity and problems with peers (r=0.445, p<0.001), but hyperactivity was not statistically correlated with prosocial behavior (r= − 0.291, p<0.001). This outcome denoted that when hyperactivity increased, prosocial behavior diminished in a constant proportion. Finally, problems with peers was not statistically correlated with prosocial behavior (r= − 0.432, p<0.001). This result suggested that when problems with peers increased, prosocial behavior declined constantly. Furthermore, a statistically significant, small, and positive correlation was observed between children’s SDQ subscale scores along with the child’s total ITPA score (total ITPA score is the mean of all subscales) (r=0.115, p<0.05) and final RPM score (r=0.115, p<0.05). In the same vein, findings revealed that children’s total ITPA score was small and correlated with children’s final RPM score (r=0.115, p<0.05) and SDQ subscale scores (e.g., behavioral problems, r=0.163, p<0.001, and hyperactivity, r=0.177, p<0.001). Further, correlations between children’s final PPVT score with SDQ subscale scores (e.g., behavioral problems, r=0.118, p<0.05, and prosocial behavior, r= − 0.141, p<0.05) were significant and small. 5.3. Children’s Problems, Psycholinguistic Abilities, and Cognitive Capacities As the second aim of this study claimed, children’s total SDQ scores were small and significantly correlated with speech-language pathologists (r=0.250, p<0.001) and class teachers (r=0.247, p<0.001). Links between children’s problems and receptive vocabulary were accepted. Correlation analyses indicated that speech-language pathologists’ total SDQ score was not statistically correlated with children’s final PPVT score (r= − 0.113, p<0.05). This result indicated that when speech-language pathologists’ total SDQ score increased, final PPVT score decreased constantly. Moreover, correlation analyses demonstrated that speech-language pathologists were significantly correlated with class teachers (r=0.258, p<0.001) on the total SDQ score. As indicated in Table 2, ITPA subtests significantly correlated among themselves (e.g., auditory reception with verbal expression (r=0.616, p<0.001), auditory reception with auditory sequential memory (r=0.632, p<0.001), and auditory reception with grammatic closure (r=0.581, p<0.001). These correlations represented an association between two levels of organization of increasing complexity. Educ. Sci. 2019,9, 234 7 of 15 Table 1. Correlations between the five Strengths and Difficulties Questionnaire (SDQ) subscales and Illinois Test of Psycholinguistic Abilities (ITPA), Peabody Image Vocabulary Test (PPVT), and Raven Progressive Matrices Test (RPM) for children. ITPA PPVT RPM Emotional Problems Behavioral Problems Hyperactivity Problems with Peers Prosocial Behavior ITPA 1.000 PPVT −0.073 1.000 0.206 RPM 0.115 * 0.077 1.000 0.046 0.186 Emotional problems −0.049 −0.053 0.016 1.000 0.394 0.358 0.781 Behavioral problems −0.163 ** 0.118 * −0.061 0.325 ** 1.000 0.005 0.041 0.292 0.000 Hyperactivity −0.177 ** 0.091 −0.011 −0.108 0.260 ** 1.000 0.002 0.116 0.843 0.062 0.000 Problems with peers −0.200 ** 0.108 0.001 −0.122 * 0.391 ** 0.445 ** 1.000 0.001 0.061 0.985 0.034 0.000 0.000 Prosocial behavior 0.094 −0.141 * 0.017 0.098 −0.298 ** −0.291 ** −0.432 ** 1.000 0.103 0.015 0.772 0.090 0.000 0.000 0.000 ** Correlation coefficients significant at the 0.001 level (bilateral), * Correlation coefficients significant at the 0.05 level (bilateral). Educ. Sci. 2019,9, 234 8 of 15 Table 2. Correlations between the SDQ and ITPA, PPVT (Peabody), and RPM (Raven) for teachers and children with hearing aids. SDQ Total. Speech-Language Pathologist SDQ Total. Tutor SDQ Total. Children with Hearing Aids Total ITPA Score. Children with Hearing Aids PPVT Final Score. Children with Hearing Aids RPM Final Score. Children with Hearing Aids SDQ Total. Speech-language pathologist 1.000 SDQ Total. Tutor 0.258 ** 1.000 SDQ Total. Children with hearing aids 0.250 ** 0.247 ** 1.000 Total ITPA Score. Children with hearing aids 0.053 0.037 0.115 * 1.000 PPVT Final Score. Children with hearing aids −0.113 * 0.050 −0.062 −0.073 1.000 RPM Final Score. Children with hearing aids 0.032 0.053 0.053 0.115 * 0.077 1.000 * Correlation coefficients significant at the 0.05 level (bilateral), ** Correlation coefficients significant at the 0.001 level (bilateral). Educ. Sci. 2019,9, 234 9 of 15 5.4. Children’s Problems Related to Children’s Aptitudes and Parents’ and School Covariates This study’s third aim was to explain the relationship between the independent (predictor) and dependent (target) variables. Consequently, a linear regression was analyzed by looking for an R-square as close to 1 that indicated a perfect linear fit. Authors employed the method of multiple linear regression analysis and fulfilled the assumptions of linearity, independence, normality, homoscedasticity, and non-collinearity to guarantee the validity of the predictive value of the selected array of variables. The method of least squares was chosen (Table 3). The Pearson correlation coefficient was 0.751, and the goodness of fit was calculated following the coefficient of determination R-square, whose values ranged between − 1 and 1. Hence, the corrected value of R-square =0.551 predicted 55.1% of the variance (corrected effect of the sample and independent variables). The typical estimation error (square root of the unexplained variance) turned out to be 0.468. The analysis of variance (ANOVA) showed whether the variance explained by the regression was significantly different (and superior) to the unexplained variance. The model was F =42.431, p<0.001, which revealed that the associated probability according to the expectations of the null hypothesis was less than 0.001. This finding led the authors to reject this hypothesis and assume that there was a real effect of these variables on the children’s total SDQ score. Table 3. Regression analysis of the socio-demographic variables as predictors of children’s total SDQ score. Target R R2FExplained Variance Socio-Demographic Predictors βt Children’s total SDQ score 0.751 0.564 42.431 *** 55.10% Hearing device 1.155 10.511 *** Geographical location −0.268 −3.278 *** ICA 0.23 3.019 ** Mothers’ age −0.1504 −2.789 ** Children’s chronological age 0.078 2.006 * *p<0.05, ** p<0.010, *** p<0.001. Finally, the authors reported the regression coefficients for the chosen model, and the constant was expressed first, followed by the slope. The model suggested that the independent variables (e.g., hearing assistive devices, geographical location, ACI, mothers’ age, and children’s chronological age) revealed statistical significance. In this model, hearing assistive device (CIs/HAs) obtained the highest coefficient (t=10.511, p<0.000). The results of the regression analysis—ITPA subscales, PPVT, and RPM dimensions—are shown in Table 4, where ITPA subscales explained 42.6% of the variance of the children’s total SDQ score. The automatic level auditory-vocal had considerable weight in the EBD score of the children’s total SDQ score, specifically, the scales of visual sequential memory, auditory sequential memory, and grammatic closure. The representative level viso-motor in verbal expression scale and the representative level viso-motor in auditory reception scale also had a significant weight. The obtained F-value was 8.594. According to the expectations of the null hypothesis, its associated probability was less than 0.001. Therefore, this hypothesis was rejected. Consequently, these variables could affect the children’s total SDQ score.