Examining the psychometric properties of the Norwegian version of the Social Aptitudes Scale in two clinical samples

Participants

Two clinical samples were included in the study. The first one is a clinical neuropediatric sample of 257 children and adolescents that were referred for a developmental and neurological assessment to the neuropediatric outpatient clinics at the University Hospital of North Norway (UNN; n = 216) and the Finnmark Hospital Trust (FIN; n = 41). Children under four years were excluded from the study because of a lack of suitability of at least one of the instruments used in the study. Another exclusion criterion was a lack of parental fluency in Norwegian. The most frequent neurodevelopmental disorders in the sample were (a) specific developmental disorders (33.1%), (b) intellectual disability (ID; 18.7%, none with severe IDs), (c) other diseases of the nervous system such as epilepsy and cerebral palsy (17.9%), (d) ASD (13.6%), (e) ADHD (13.6%), and (f) congenital malformations and chromosomal abnormalities (10.5%). Specific developmental disorder was operationalized according to ICD-10. This included F80 specific developmental disorders of speech and language, F81 specific developmental disorders of scholastic skills, F82 specific developmental disorder of motor function, and F 83 mixed specific developmental disorders. A given subject could have more than one diagnosis. For further description of the design and samples, see Halvorsen, Aman [13], Halvorsen, Mathiassen [14], or Halvorsen, Mathiassen [15].

The second clinical sample includes 804 patients from the Child and Adolescents Mental Health Services at UNN. UNN serves as a specialist health care hospital a population of 190,726 residents of the county municipalities of Troms and the northern part of Nordland. The health care trust is covering an area of approximately 31,300 km². Annually, UNN provides mental health services to about 5% (2100/42,000) of the population aged 0–18 years. About 60% of the treated patients are new referrals from general practitioners and the child protection services. The CAMHS at UNN consist of six outpatient and one inpatient clinic. All of them include the online version of the DAWBA [6] in the routine clinical assessment. The questionnaires SAS and SDQ are an integrated part of DAWBA. The main diagnostic groupings based on the DAWBA in the sample were, in descending order: (a) emotional disorders (57.6%), (b) behavioral disorders (47.3%), (c) hyperkinesis (28.1%), and (d) ASD (2.6%). All DAWBA data at UNN are stored in a de-identified local CAMHS quality register. For further description of the design and sample, see Fernández de la Cruz, Vidal-Ribas [16]. The data protection officer at UNN has approved the use of data from the quality register for research purposes.

The main demographic and clinical characteristics of the neuropediatric- and the CAMHS sample can be found in Table 1. The mean age of the neuropediatric sample was 10.54 years (SD = 3.46; range 5 to 18 years) and 11.88 years (SD = 3.52; range 4 to 19 years) in the CAMHS sample.

Measures

The Norwegian SAS [5] was used to assess social skills of the children and adolescents and the form was completed by their parents. The SAS is a ten-item questionnaire (e.g., “Aware of what is and is not socially appropriate”) and each item is rated on a on a five-point scale from 0 (A lot worse than average) through 4 (A lot better than average), with lower scores indicating worse function. Detailed information about SAS can be found elsewhere (http://dawba.info/SAS/). Both the neuropediatric- and the CAMHS sample completed the SAS in the DAWBA.

The Norwegian Social Responsiveness Scale [17] is a parent-completed screening questionnaire often used to measure ASD severity. It is composed of 65 items within the five subdomains Social Awareness, Social Cognition, Social Communication, Social Motivation, and Autistics Mannerisms in addition to an overall total score. Parents respond to how often their child displays a given behavior on a four-point Likert scale from 0 (not true) through 3 (almost always true) in the past six months, with higher scores indicating worse function. The manual recommends the use of the SRS raw scores in research. The SRS has been validated in different cultures, with results indicating good psychometric properties (e.g., [18]), and scores on the SRS are strongly correlated with Autism Diagnostic Interview – Revised domain scores (r = 0.65–0.77; [19]). Only the neuropediatric sample completed the SRS and Cronbach’s alpha ranged from 0.69 (Social Awareness) to 0.89 (Social Communication). SRS scores were missing for 14 children.

The Norwegian SDQ parent version [8] was used in the current study. The SDQ is a 25-item mental health questionnaire covering four problem areas (emotional, hyperactivity-inattention, conduct, and peer problems), one area of strength (prosocial behavior), and additional questions related to distress and functional impairment. Each item is rated on a three-point scale from 0 (not true) through 2 (certainly true). The SDQ has been validated in different cultures, with results indicating good psychometric properties [20, 21]. Both the neuropediatric- and the CAMHS samples completed the SDQ in the DAWBA. In the neuropediatric sample Cronbach’s alpha ranged from 0.69 (conduct) to 0.80 (prosocial behavior) for the parent version. In the CAMHS sample Cronbach’s alpha ranged from 0.65 (peer problem) to 0.81 (hyperactivity-inattention) for the parent version.

The Norwegian version of the DAWBA [6] was used to establish diagnoses in the CAMHS sample based on DSM-IV diagnostic criteria [22]. The DAWBA is a detailed diagnostic tool completed by parents (takes approximately 30 min), and youths (takes approximately 10 min), with a briefer questionnaire for teachers (takes approximately 10 min) In the current study, the DAWBA was completed as an online package of questionnaires on admission to the clinics. In this paper, we group mental disorders into emotional disorder (including anxiety and depressive disorders); behavioral disorders (including oppositional defiant and conduct disorders); ADHD; and ASD (including autism and Asperger’s syndrome). Participants were assigned a positive diagnosis if they scored 3 or higher in the relevant DAWBA bands [23], as previously described [16]. The DAWBA has shown good discriminative ability in both population-based samples and clinical samples, as well as across different categories of diagnoses [6]. Both in Norway and Great Britain, the DAWBA generates realistic estimates of prevalence for psychiatric illnesses as well as high predictive validity when used in public health services [24, 25]. As a SAS score of 12 or less are included as part of the diagnostic process towards an ASD diagnosis in the DAWBA, we limited the examination of SAS psychometric properties in the CAMHS sample to factor structure and correlations of SAS and SDQ scores, and not relation to an ASD diagnosis, to avoid circularity. Detailed information about SAS can be found elsewhere (www.dawba.info). We report DAWBA data for the CAMHS sample.

VABS-II [26], a semi-structured interview, was used to establish the child’s adaptive level of functioning and includes the following four domains with related subdomains: Communication (receptive, expressive, and written), Daily Living Skills (personal, domestic, and community), Socialization (interpersonal relationships, play and leisure time, and coping skills), and Motor Skills (gross and fine). The neuropediatric sample completed the VABS-II and the VABS-II total, the Communication, and the Socialization scores were used. VABS-II scores were missing for 15 children.

Procedure

Neuropediatric sample: The children underwent an interdisciplinary, neurodevelopmental/neurological assessment over two days, where they were assessed for the presence of a neurological/ neurodevelopmental disorder. In addition, the examinations included MRI Caput, EEG and/or genetic testing. Clinical psychologist/ neuropsychologist administered the VABS-II and a standardized intelligence scale. Diagnoses were based on ICD-10 criteria [27, 28]. A score below 70 on both the standardized intelligence test and the VABS-II was used to diagnose the presence of an ID. Furthermore, in the current study, ASD diagnoses for the neuropediatric sample were not based on the computer-predictions from the DAWBA, but on clinic diagnoses where the results from the ADI-R and the Autism Diagnostic Observation Schedule often were included in the assessments.

CAMHS sample: Parents, children and their teachers complete the online version of the DAWBA [6] in the routine clinical assessment. The questionnaires SAS and SDQ are an integrated part of DAWBA. The data protection officer at UNN allowed to analyze the two datasets for the neuropediatric- and the CAMHS sample separately, but did not agree to merge the two files to one data file.

Statistical analysis

The statistical analyses were conducted with SPSS and included the calculation of Pearson’ correlations (r), McDonald’s Omega, and McDonald’s Omega if item deleted. The guidelines from the European Federation of Psychologists’ Association (EFPA; [29]) for the evaluation of the psychometric properties were used. To evaluate congruent validity correlations of r < 0.55 are considered inadequate, r between 0.55 and 0.64 as adequate, r between 0.65 and 0.74 as good, and r of 0.75 or bigger as excellent. To evaluate McDonald’s Omega coefficients smaller than 0.70 are considered inadequate, coefficients between 0.70 and 0.79 as adequate, coefficients between 0.80 and 0.89 as good, and coefficients of 0.90 or bigger as excellent [29].

A CFA-model with one general latent factor with ten indicators from the SAS was tested separately for each sample, respectively, using Mplus. The weighted least square mean and variance adjusted (WLSMV) estimator was used. Different fit indices were used to evaluate model fit: The χ²-statistic, the χ²/degrees of freedom ratio (χ²/df) with a threshold level of 3.00 or 2.00, the Comparative Fit Index (CFI) and the Tucker Lewis Index (TLI), where greater values than 0.90 or 0.95 indicate good model fit, and the Root Mean Square Error of Approximation (RMSEA), where smaller values than 0.07 or 0.06 indicate good model fit [30].