Psychometric Properties Of The Social Skills
Attention Deficit Hyperactivity Disorder (ADHD) is one of the most common neurodevelopmental disorders in children (Danielson et al., 2018). The essential features of ADHD are inattention, hyperactivity, and impulsivity, which are associated with both social and behavioral risk factors in school-aged children (Danielson et al., 2018). Also, children with ADHD are more likely to develop deficits in social skills that can lead to academic underachievement, peer rejection, and maladjustment (Barkley, 2006). Many of the difficulties children with ADHD have with academic performance have been linked to ADHD-related deficits in social skills (Loe & Feldman, 2007). “Social skills” refers to the emotional, social, intellectual, and behavioral skills that are required to positively interact with others, for example, cooperation with peers, sharing belongings, taking turns, and self-regulation (Thurber, Heller, & Hinshaw, 2002). Children with ADHD are more likely to be rated by their peers as argumentative, intrusive, and awkward in social interactions (Carpenter Rich et al., 2009). Moreover, teachers have reported students with ADHD as interruptive, less compliant, and more aggressive than other students (Barry et al., 2016).
Teachers may be the first observers of behavioral issues associated with ADHD in a classroom setting. According to Snider, Frankenberger, and Aspenson (2000), teachers play a significant role in providing the initial albeit informal screening for ADHD and are often involved in making referrals to healthcare providers. Furthermore, teachers’ ratings serve as an invaluable summary of children’s behaviors and skills observed in the school setting (Barry et al., 2016; Jensen & Cooper, 2002). Therefore, surveying teachers about the social skills of children with ADHD will become a significant factor in identifying whether the children will be diagnosed with social skills deficits.
Various instruments have been developed to screen and assess children with social skills issues. As an example, the Social Skills Rating System (SSRS) is a multi-rater assessment developed by Gresham and Elliot (1990). The SSRS is a standardized instrument used to assess children’s social behavior at home and at school setting. It is comprised of three rating forms: one for parents, one for teachers, and one for the student. The SSRS is a norm-referenced instrument that is appropriate for use with U.S. populations of different ethnicities and an age range of 3 to 18 years. Also, the items number of the SSRS range from 34 to57 items; the range varies based on which the three forms will be distributed (Gresham & Elliot, 1990). The SSRS is used to assess students’ behaviors that can impact peer-student relations, teacher-student relations, and academic performance. The SSRS has three scales: (1) the Social Skills Scale, that includes the cooperation, assertion, and self-control subscales; (2) the Problem Behaviors Scale, that includes the externalizing and internalizing subscales; and (3) the Academic Competence Scale (Gresham et al, 2010).
The teacher-specific version of the SSRS (SSRS-T) has been used as a screening tool of social behaviors for children with ADHD and Oppositional Defiant Disorder (ODD) (Frankel & Feinberg, 2002), as well as an assessment measure of social skills training outcomes for children with ADHD (Antshel, 2003; Pfiffner & McBurnett, 1997). While the psychometric properties of the SSRS-T have been evaluated in various populations and settings, such as when used by teachers to assess social skills for children with learning disabilities (Bramlett & Smith, 1994; Gresham, Elliott, & Black, 1987), there are few studies in which the psychometric properties of the SSRS-T have been examined when used to assess children with ADHD (Antshel, 2003; Pfiffner & McBurnett, 1997; Van der Oord et al., 2005). The purpose of this study was to evaluate the psychometric properties of the SSRS-T when used to assess social behaviors of children with ADHD. The specific aims were to: (1) evaluate the internal consistency reliability of the SSRS-T and (2) examine the construct validity of the SSRS-T by performing an exploratory factor analysis.
Methods
This was a secondary analysis of existing longitudinal data from the Fragile Families and Child Well-Being Study (FFCWS). The purpose of the FFCWS was to examine the social, economic, and health conditions of unmarried parents and their children (Reichman et al., 2001). The study included six waves; parents were interviewed shortly after their child’s birth (Wave 1), when their children were a year old (Wave 2), 3 years old (Wave 3), 5 years of age (Wave 4), 9 years old (Wave 5), and 15 years old (Wave 6). The study included 4,898 births at 75 hospitals across large U.S. large cities. Refer to Reichman et al., (2001) for a more detailed description of the study sample, design, and protocol.
Sample Description
For this study, data collected in wave 5, when children were 9 years of age, were included. Only children with ADHD whose had teachers had completed the SSRS-T questionnaires were included in this analysis.
Procedures
For the parent study, research interviewers collected informed consent from each child’s primary caregiver, assent consent from the child to contact the teacher, and school and address information for the appropriate teacher. Then, research staff members mailed teachers copies of the consent forms and survey instrument to complete and mail back to the research staffers. Ethical approval for the parent study was obtained per protocol from the Institutional Review Boards (IRB) or human subjects committees in each of the 20 U.S. hospitals that served as settings for data collection. For this secondary analysis, ethical approval was obtained from the IRB at the University of Kentucky.
Instrument Description
The full SSRS-T has 37 items. This version was designed for teachers to use to evaluate the social behaviors of children ages 3 to 18 years. The SSRS-T also includes two additional separate scales: the Social Skills scale and Problem Behaviors Scale. The Social Skills Scale has 25 items across three subscales: (1) the Cooperation subscale that has 10 items; the Assertion subscale that has 5 items; and the Self-Control subscale that has 10 items. The total score for the Social Skills Scale ranges from 0 to 75, and scores on each subscale are as follows; Cooperation subscale, 0 to 30; Assertion subscale, 0 to 15; and Self-Control subscale, 0 to 30. The Problem Behaviors Scale has 12 items and consists of two subscales: the Internalizing subscale has 6 items and the Externalizing subscale has 6 items. The overall score of the Problem Behaviors Scale ranges from 0 to 36, and scores on each subscale are as follows: Internalizing subscale, 0 to 18; and Externalizing subscale, 0 to 18. The SSRS is rated on a 4-point Likert scale ranging from 1 (never) to 4 (very often). Higher scores on each of the SSRS subscales indicate better social skills and lower scores indicate social skills deficits. The full SSRS-T score (the Problem Behaviors subscale and the Social Skills subscale) ranges from 0 to 111.
Statistical Analysis
The Statistical Package for Social Science (SPSS) version 26 (Cronk, 2019) was used to analyze the data. Descriptive statistics were used to describe the characteristics of the participants using frequencies and percentages for categorical variables, and means and standard deviations for continuous variables. Cronbach’s alpha value was used to assess the internal consistency reliability of the total SSRS-T scale and subscales (Social Skills and Problem Behaviors). Item analyses were used including item-item intercorrelations, item-to-subscale correlations, and Cronbach’s alpha value changes with item deletion. Exploratory factor analysis was used to assess the construct validity of the full SSRS scale.
Established criteria were used to examine the factorability of correlation among the 37 SSRS items. The sampling adequacy was assessed using Kaiser-Meyer-Olkin which was .93, above the recommended value of .6, and Bartlett’s test of sphericity was significant (χ2 = 5497.6, p < .05). The off-diagonal elements of the anti-image correlation matrix have values more than .3. Lastly, communalities were all above .3, indicating that each item shared some common variance with other items. Taken together, factor analysis was considered to be suitable for the 37 items.
Principle components factor analyses with orthogonal rotation (varimax) were performed to determine the factor structure. Three criteria were used to examine the factor structure and assign items to factors: eigenvalues > 1; a scree plot that arranges the eigenvalues from the correlation matrix in descending order, and factor loading coefficients > .40 for each item. Then, varimax rotation was used to rotate the components orthogonally. Factor loading coefficients were assessed to identify the placement of items on components, and coefficients > 0.4 were considered acceptable to retain in an item grouping.
Results
Demographic Characteristics
Of the 428 9-year-olds included in wave 5, 194 met our inclusion criteria (had a diagnosis of ADHD and had teacher-completed evaluations using the SSRS-T) and were included in this analysis. Of these, 69.6% were male and lived in households whose mean income was $29,152.98 ± $29,006.72. More than half of the parents of the children were married/cohabitated (56.7%), and the majority of the children’s fathers and mothers were Black, non-Hispanic, 48.5% and 46.4% respectively. The majority of both the mothers and the fathers had a high school education or less. The mean for the overall SSRS-T was 45.9 (SD = 10.6). The means for the Social Skills scale and the Problem Behaviors scale were 34.2 (SD= 13.6) and 11.7 (SD= 6.9), respectively. Sample characteristics are shown in Table 1.
Reliability and Item Analysis
In this sample of children with ADHD, Cronbach’s alpha for the total scale of the SSRS was .80. Cronbach’s alpha for the Social Skills subscale was .95, indicating high internal consistency reliability, and did not improve with the deletion of any item. Cronbach’s alpha for the Problems Behaviors subscale was .89, also indicating high internal consistency reliability. Cronbach’s alpha did not improve with the deletion of any item.
Item analysis showed that values on the inter-item correlation of the Social Skills subscale ranged from .12 to .78. The minimum range was below the ideal value (.20 to .80) (Priest, 1995). This was due to the low inter-correlations of item 15 (child finishes class assignments with time limits) and item 28 (child compromises in conflict by changing own ideas). However, the corrected item-total correlation of item 15 was .55, and the Cronbach’s alpha for the Social Skills subscale would not change if item 15 were deleted (.95). The inter-item correlations for items on the Problem Behaviors subscale ranged from .01 to .80. The minimum range was also below the ideal value (.20 to .80). This was due to the low inter-correlations of item 5 (child threatens or bullies others) with item 34 (child likes to be alone). However, the corrected item-total correlation of item 5 was .32, and the Cronbach’s alpha for the Problem Behaviors subscale would increase slightly from .888 to .892 if item 5 were deleted.
Exploratory Factor Analyses
Exploratory factor analysis resulted in five factors with eigenvalues >1 that explained 67% of the variance in the measure. The scree plot showed the curve change direction and became horizontal at the point between four and five, suggesting four factors could be retained. The solution derived using varimax rotation is shown in Table 2. The first component included 14 items, and it could be labeled as a Self-Control scale. The second component was composed of 10 items, and it could be labeled as a Cooperation scale. The third component consisted of 7 items, and it could be labeled as an Assertion scale. The fourth component included six items, and it could be labeled as an Internalizing Behavior scale. The fifth component included one item, and it could be labeled as an Externalizing Behavior scale. Cross loading was noted in items 11, 12, 14, 28, 29, 31, and 37. However, each of these items loaded most strongly on a single component with which the item seemed to be conceptually related to other items that also strongly loaded on that factor.
Discussion
The purpose of this secondary analysis was to evaluate the psychometrics of the SSRS-T when used by teachers to evaluate school-aged children with ADHD. The findings of this study support the reliability and the validity of the SSRS-T for this purpose. The internal consistency reliability estimates were acceptable for both the overall SSRS-T scale and the subscales, Social Skills and Problem Behaviors. These results were comparable to the internal consistency reliability of the overall SSRS-T scale and the subscales in the original version of Gresham and Elliot (1990), where the Cronbach’s alpha was > .80.
With regard to validity, the factors analysis of the SSRS-T in children with ADHD sample was satisfied. In this study, factor analysis of the SSRS-T yielded the same factor structure as that of the original version, for example, five factors (Self-control, Cooperation, Assertion, Internalize Behaviors, and Externalize Behaviors) of the SSRS-T (Gresham & Elliot, 1990). However, items that loaded most strongly on each of these factors loaded differently than the items identified for each of the original subscales. For example, 6 items negatively loaded on factor 1(Self-Control), that in the original version loaded more strongly on factor 5 (Externalizing Behavior). The reason for these differences could be negatively wording in some of the items which may cause negative factor loadings. Therefore, the negative sign indicates a high score on a factor with a low score on the item and vice versa. These results were comparable to those reported by Rich et al. (2008). However, only one item, number 37, loaded on the Externalizing Behavior component. Inspection of factor loading of the Externalize Behavior component in this sample has only one loading item, which was .691.
Limitations, Future Studies
Several limitations of this psychometric evaluation should be noted. This was a secondary analysis of cross-sectional data that only covered one age cohort. Therefore, the results cannot be generalized to younger or older students. Another limitation with this study was that it did not include multiple measures of the same construct to perform convergent or divergent validity testing for the SSRS-T. While this study supported the original Social Skills and Problem Behavior subscales, there were insufficient data to assess the subscales of responsibility and academic competence. Future studies are needed to evaluate all the factors structure of the SSRS-T to ensure the appropriate use of all SSRS-T subscales, and to assess with children across a spectrum of ages.
Conclusion and Implications
The results of this study support that the SSRS-T has satisfactory reliability and validity when used by teachers to evaluate school-aged children with ADHD. This suggests that academicians can use the SSRS-T as a tool to determine the social skills and problem behaviors of students with ADHD. Researchers interested in measuring the social behavior of children may use the assessment results of the SSRS-T to form interventions to improve social skills.
