Fluent Essays

Summarize the Results section of the article Critique/Evaluate the Results issues: What are the major results/findings of the study? How do these results answer the original research question(s)? How confident are you with the findings? Summarize the Discussion/Conclusion of the article Critique/Evaluate the Implications of the Findings: What conclusions did the authors reach? Are these conclusion(s) appropriate? What did the researchers report as the implications of the study? In your opinion, what is the significance of the findings for your area of interest? NO WORD FOR WORD. NO IN-TEXT CITATIONS. ANSWER ALL QUESTIONS RESEARCH ARTICLE Different associations between intelligence and social cognition in children with and without autism spectrum disorders Tetsu HirosawaID 1,2*, Keiko Kontani1,2, Mina Fukai1, Masafumi Kameya1, Daiki Soma1, Shoryoku Hino 3 , Tatsuru Kitamura 3 , Chiaki Hasegawa 2 , Kyung-min An 2 , Tetsuya Takahashi 2 , Yuko Yoshimura 2,4 , Mitsuru Kikuchi 1,2 1 Department of Psychiatry and Neurobiology, Graduate School of Medical Science, Kanazawa University, Kanazawa, Japan, 2 Research Center for Child Mental Development, Kanazawa University, Kanazawa, Japan, 3 Department of Neuropsychiatry, Ishikawa Prefectural Takamatsu Hospital, Ishikawa, Japan, 4 Faculty of Education, Institute of Human and Social Sciences, Kanazawa University, Kanazawa, Japan * [email protected] Abstract Autism spectrum disorders (ASD) are characterized by impaired social cognition and com- munication. In addition to social impairment, individuals with ASD often have intellectual dis- ability. Intelligence is known to influence the phenotypic presentation of ASD. Nevertheless, the relation between intelligence and social reciprocity in people with ASD remains unclear, especially in childhood. To elucidate this relation, we analyzed 56 typically developing chil- dren (35 male, 21 female, aged 60–91 months) and 46 children with ASD (35 male, 11 female, aged 60–98 months) from university and affiliated hospitals. Their cognitive function was evaluated using the Kaufman Assessment Battery for Children. Their social cognition was assessed using the Social Responsiveness Scale. We used linear regression models to ascertain whether the associations between intelligence and social cognition of typically developing children and children with ASD are significantly different. Among the children with ASD, scores on the Kaufman Assessment Battery for Children correlated significantly with social cognition, indicating that higher intelligence is associated with better social cogni- tion. For typically developing children, however, no significant correlation was found. One explanation might be that children with ASD fully use general intelligence for successful learning in social cognition, although extensive use of intelligence might not be necessary for TD children. Alternatively, autistic impairment in social cognition can be compensated by intelligence despite a persistent deficit in social cognition. In either case, when using the SRS as a quantitative phenotype measure for ASD, the influence of intelligence must be considered. Introduction Numerous and diverse difficulties driven by social impairment are experienced by individuals with autism spectrum disorder (ASD), a neurodevelopmental disorder associated with PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0235380 August 21, 2020 1 / 18 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Hirosawa T, Kontani K, Fukai M, Kameya M, Soma D, Hino S, et al. (2020) Different associations between intelligence and social cognition in children with and without autism spectrum disorders. PLoS ONE 15(8): e0235380. https://doi.org/10.1371/journal.pone.0235380 Editor: John Richey, Virginia Polytechnic Institute and State University, UNITED STATES Received: August 24, 2019 Accepted: June 7, 2020 Published: August 21, 2020 Copyright: © 2020 Hirosawa et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the manuscript and its Supporting Information files. Funding: The authors received no specific funding for this work. Competing interests: The authors have declared that no competing interests exist. http://orcid.org/0000-0001-8710-5638 https://doi.org/10.1371/journal.pone.0235380 http://crossmark.crossref.org/dialog/?doi=10.1371/journal.pone.0235380&domain=pdf&date_stamp=2020-08-21 http://crossmark.crossref.org/dialog/?doi=10.1371/journal.pone.0235380&domain=pdf&date_stamp=2020-08-21 http://crossmark.crossref.org/dialog/?doi=10.1371/journal.pone.0235380&domain=pdf&date_stamp=2020-08-21 http://crossmark.crossref.org/dialog/?doi=10.1371/journal.pone.0235380&domain=pdf&date_stamp=2020-08-21 http://crossmark.crossref.org/dialog/?doi=10.1371/journal.pone.0235380&domain=pdf&date_stamp=2020-08-21 http://crossmark.crossref.org/dialog/?doi=10.1371/journal.pone.0235380&domain=pdf&date_stamp=2020-08-21 https://doi.org/10.1371/journal.pone.0235380 http://creativecommons.org/licenses/by/4.0/ repetitive behaviors and characterized by impaired social cognition and communication [1]. For example, social impairment leads people with ASD to experience difficulties in education, employment, and in severe cases, independent living [2–5]. Moreover, it is noteworthy that severe social impairment in childhood strongly predicts those factors later in their adulthood [6], which emphasizes the importance of understanding factors associated with childhood social impairment. Many individuals with ASD have intellectual disability (ID) in addition to social impairment [7]. Although shared etiologies are largely unknown, some researchers have inferred shared genetic risk factors [8] and prenatal risk factors such as exposure to infection [9], medications [10], and preterm delivery [11]. Furthermore, earlier reports describe that intelligence and autistic symptomatology might not be mutually independent. For instance, repetitive and restrictive behaviors (RRBs) have been shown to correlate with a lower nonver- bal developmental quotient [12–16], with the sole exception of a study by Joseph et al. which found no significant relation [17]. As another illustrative instance, Bishop et al. explored the relation between nonverbal IQ and RRBs in 830 children with ASD and reported that the prev- alence of most RRBs (e.g. repetitive use of objects, and hand and finger mannerisms) is nega- tively associated with nonverbal IQ. Conversely, the association was positive for RBBs of certain types (e.g., circumscribed interests) [16]. In addition, the relation between social cognition and intelligence in ASD or PDDNOS has been debated since the 1990s [18–20]. For instance, Buitelaar et al. examined 60 children: 20 children with autism; 20 with pervasive developmental disorder, not otherwise specified (PDDNOS); and 20 children under psychiatric control. Performance IQ and verbal memory were found to be strong predictors of performance on emotion recognition tasks. A significant positive correlation between performance IQ and second-order ToM was also reported, although the effect was not found to be significant for first-order ToM. They reported relation- ships throughout the sample, but also reported that the coefficient patterns did not differ markedly across the three diagnostic groups [18]. Subsequent studies of ASD and PDDNOS populations using various measures of social cognition almost invariably found significant association between better social cognition and higher intelligence [21–29]. For instance, Liv- ingston et al. measured ASD symptoms using ADOS-G and compared children with ASD hav- ing milder ASD symptoms despite poor ToM performance, with children who had severer ASD symptoms and poor ToM performance. They reported the former group as having higher intelligence [29]. One exception is a study conducted by Muller et al. They evaluated partici- pants’ social cognition using a Movie for the Assessment of Social Cognition (MASC), in which the participants inferred film characters’ mental states including emotions, thoughts, and intentions. Based on the MASC results, Muller et al. reported non-significant association between intelligence and social cognition [30]. Some studies have explored the relation between social cognition and intelligence in TD populations [22, 25, 28, 30–33]. Unlike those in ASD populations, some reports have described results showing non-significant correlation [22, 30–32]. Others have reported significant cor- relation [25, 28, 33]. Mohn et al. examined 250 healthy adults and reported non-significant association between IQ and social cognition, as measured using the Mayer–Salovey–Caruso Emotional Intelligence Test (MSCEIT) [31]. Farrelly et al. reported significant association between social cognition measured using the same battery (MSCEIT) and crystallized intelli- gence, as measured using the Gf/Gc Quickie Test Battery [33]. More generally, some other studies examined how autistic social impairment and intelli- gence are related. A few studies have examined the association between intelligence and social communication or motivation. Those studies examined ASD populations and found signifi- cant correlation between intelligence and social communication [34–36]. Qualls et al. PLOS ONE Intelligence and social cognition in autism spectrum disorders PLOS ONE | https://doi.org/10.1371/journal.pone.0235380 August 21, 2020 2 / 18 https://doi.org/10.1371/journal.pone.0235380 examined adolescents with ASD and reported positive association between scores of the Social Communication Questionnaire (SCQ; [37]) and IQ [34]. To assess a general population, one study conducted with more than 5000 participants [38] revealed that higher verbal IQ has a significant association with better social communication, as measured by Social Communica- tion Disorders Checklist and Children’s Communication Checklist, only in female partici- pants. For social motivation, two studies have been conducted. Dawson et al. reported that lower IQ is related to lower social motivation, as measured by Broader Phenotype Autism Symptom Scale, in an ASD population [39]. Dubey et al. assessed social motivation by their “choose-a-movie paradigm measures” and reported that correlation was not significant for a combined population including both TD and ASD [40]. The Social Responsiveness Scale (SRS) [41] is a widely used rating scale for autistic symp- tomatology. It measures the severity of autistic symptomatology as a quantitative trait among children with ASD and among TD children. Actually, SRS comprises five sub-scales of the autistic mannerism sub-scale and four components of social reciprocity: receptive (social awareness sub-scale), cognitive (social cognition sub-scale), expressive (social communication sub-scale), and motivational (social motivation sub-scale) subdomains [41]. Those autistic traits and symptoms measured using SRS are distributed continuously, in ASD populations and in the general population [42]. Therefore, SRS is particularly useful for characterizing milder social impairment such as that which lies at the boundary between the normal popula- tion distribution and clinical-level affectation. In this sense, SRS might be suitable for examin- ing the association between potentially milder autistic symptoms and intelligence. According to this framework, four studies have been conducted [43–46]. Hus et al. recruited 2,368 indi- viduals with ASD and 1,913 unaffected siblings. Subsequently, they assessed factors influencing the SRS raw scores. They reported the respective associations of higher SRS scores with greater non-ASD behavior difficulties, higher age, more impaired language, and lower non-verbal IQ in an ASD population. It is particularly interesting that non-verbal IQ effects on the SRS raw score were not found to be significant for unaffected siblings [45]. Three other recent studies specifically examined the sub-scales of SRS. The results are inconsistent. Rodgers et al. examined the association in a sample of children with ASD using the SRS-Second Edition (SRS-2; [47]) and the Wechsler Intelligence Scale for Children-4th Edition (WISC-IV; [48]). They reported that lower IQ is significantly associated with higher scores in the social cognition sub-scale (i.e., greater difficulties with social cognition) for girls only. No significant correlation was found for any other sub-scale [43]. Torske et al. assessed children with ASD or PDDNOS using the SRS [41]), age-appropriate Wechsler tests of intelligence, and Behavior Rating Inventory of Execu- tive Function (BRIEF). They reported that IQ did not correlate to any sub-scale [44]. Chouinard et al. examined adolescent participants with and without ASD. They reported that higher IQ was correlated with a better social communication sub-scale of SRS [46]. Overall, studies of ASD subjects have tended to find significant association between intelli- gence and subdomains of social reciprocity. Studies of TD subjects have tended to find non- significant association. The relation in ASD children might differ from those in TD children. Actually, only one report describes a study examining this subject [22]. For that study, Kim et al. examined adolescent participants with and without ASD using a facial emotion recogni- tion task in which participants identify six emotions expressed by avatars: happiness, fear, anger, disgust, sadness, and surprise. They reported that full scale IQ was positively correlated with accuracy of happy expression identification only in the ASD sample. Moreover, the group difference of the correlations was significant. The study described herein was conducted to extend findings reported by Kim et al. [22] and those we described above to assess differences in the relations between intelligence and subdomains of social reciprocity in children with ASD and those in TD children. We narrowed PLOS ONE Intelligence and social cognition in autism spectrum disorders PLOS ONE | https://doi.org/10.1371/journal.pone.0235380 August 21, 2020 3 / 18 https://doi.org/10.1371/journal.pone.0235380 the age range of the participants because social ability in childhood might influence the devel- opment of intelligence [49]. Consequently, the relation found for age-inhomogeneous partici- pants might be unreliable. We hypothesize that the relations between intelligence and social reciprocity differ in children with ASD and in TD children. More specifically, higher intelli- gence is associated with better social awareness, social cognition, social communication, and social motivation only in children with ASD. Materials and methods Participants From Kanazawa University and affiliated hospitals, we recruited 56 TD children (35 male, 21 female, aged 60–91 months) and 50 children with ASD (38 male, 12 female, aged 60–98 months). Four children with ASD were unable to complete the psychometric evaluation because of their severe psychomotor agitation. For those children, the experimenter decided to halt the evaluation. Their data were excluded from statistical analyses. Consequently, we ana- lyzed 56 TD children and 46 children with ASD (35 male, 11 female, aged 60–98 months). The ASD diagnosis was made according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth edition (DSM-IV) [1] using the Diagnostic Interview for Social and Com- munication Disorders (DISCO) [50] or the Autism Diagnostic Observation Schedule–Generic (ADOS-G) [51]. The exclusion criteria were (1) blindness, (2) deafness, (3) any other neuro- psychiatric disorder including epilepsy, and (4) ongoing medication. Written informed con- sent was obtained from parents before participation by the children. The Ethics Committee of Kanazawa University Hospital approved the methods and procedures, all of which were con- ducted in accordance with the Declaration of Helsinki. We are continually recruiting participants as part of a single large project (Bambi plan, http://bambiplan.w3.kanazawa-u.ac.jp/pdf/jusen_english.pdf). Some participants overlap with those of our earlier study [52]. However, the results do not. In addition, the emphases of that earlier study differed from those of the present study. Assessment of intelligence For this study, we assessed the intelligence of the participants using the Japanese version of the Kaufman Assessment Battery for Children (K-ABC) [53]. In K-ABC, the set of skills for prob- lem-solving abilities is interpreted as intelligence. Knowledge of facts is defined as achieve- ment. In this sense, K-ABC was developed to distinguish intelligence from knowledge [53, 54]. In K-ABC, intelligence is measured on the Sequential Processing Scale and the Simultaneous Processing Scale. The Mental Processing Scale (MPS), a unification of those two scales, is intended as a measure of total intelligence in the assessment battery [54]. On the Sequential Processing Scale, the child must solve problems in a serial, stepwise manner. The subtests included in this scale are Hand Movements, Number Recall, and Word Order. For the Simulta- neous Processing Scale, the child must integrate many stimuli simultaneously to solve prob- lems. The subtests included in this scale are Magic Window, Face Recognition, Gestalt Closure, Triangles, Matrix Analogies, Spatial Memory, and Photo Series. The Achievement Scale (ACH) provides an estimate of earlier learning. It reflects the effectiveness by which a child applies intelligence in real-life situations [55]. The subtests for ACH are Expressive Vocabulary, Faces and Places, Arithmetic, Riddles, Reading/Decoding, and Reading/Understanding. Kamphaus et al. reported that aCH includes measures of what have traditionally been identified as verbal intelligence (i.e., verbal concept formation and vocabulary) and acquired general information or school skills (arithmetic, letter and word reading, and word and sentence comprehension). We described details of the K-ABC sub-scales in S1 File. Some subtests are given at all ages. PLOS ONE Intelligence and social cognition in autism spectrum disorders PLOS ONE | https://doi.org/10.1371/journal.pone.0235380 August 21, 2020 4 / 18 http://bambiplan.w3.kanazawa-u.ac.jp/pdf/jusen_english.pdf https://doi.org/10.1371/journal.pone.0235380 Others are given to selected age groups. Scores are provided as age-adjusted standardized scores, normalized to have mean of 100 and standard deviation of 15. Haddad and Naglieri asserted that K-ABC differs from the Wechsler Intelligence Scale for Children–Revised (WISC-R; [56]): WISC splits intelligence by verbal vs. non-verbal, but K-ABC measures intelligence based on the processing style necessary to solve tasks; moreover, it separates that assessment from achievement. Reported correlations of MPS with Full Scale IQ measured using the WISC-R tend to be moderate (0.62–0.76) [53]. Those results imply that intelligence measured using the K-ABC resembles that measured using the WISC-R, but they are not identical. In fact, they have their own unique characteristics. The unique characteristics of K-ABC might be a result derived from numerous factors including the following: (i) Limited verbalization is required of the child. In fact, K-ABC uses a minimal amount of verbal involve- ment in measuring intellectual processing [54]. (ii) The use of “teaching” items helps to ensure that all children understand what is expected of them for each task. (iii) The inclusion of a wider variety of tasks in the K-ABC than the variety found in WISC (e.g., Gestalt Closure, Face Recognition, Word Order). K-ABC is expected to be a good test battery for Children with ASD because they might have good intelligence irrespective of social or language impairment. Assessment of social reciprocity We assessed participants’ social reciprocity using SRS [41]: a 65-item rating scale that measures sociality and autistic mannerisms as a quantitative trait for TD children and for children who are clinically affected by autism spectrum conditions. It measures the social awareness sub- scale, social cognition sub-scale, social communication sub-scale, social motivation sub-scale, and autistic mannerisms sub-scale to generate a single measure. In both groups, a parent of each participant filled out the SRS. We used gender-normed T scores (SRS-T) of each sub- scale [57]. Higher scores represent greater difficulties with social cognition. Although the validity of self-ratings of children is still under study, the SRS can be com- pleted by a parent, a teacher, or another adult informant. In this way, it involves ratings of chil- dren in their natural social contexts and reflects what has been observed consistently over weeks or months of time rather than merely reflecting results of a single clinical or laboratory observation [41]. Therefore, it capitalizes on both direct observation and on the accumulated history of behaviors observed by the informant over time. By virtue of this characteristic, SRS reflects social abilities appearing not in only one-to-one real-world communication but also in one-to-many or many-to-many real-world communication, rather than those appearing in one-to-one communications or virtual communications in an examination room. For that rea- son, good agreement has been reported for comparison with other parents or teacher reported ASD-directed behavior assessments (e.g., SCQ [37, 58, 59]), Children’s Communication Checklist ([58–60]), and Social and Communication Disorders Checklist [61]). The SRS scores are also known to exhibit high inter-rater reliability [62] and are known to be distributed con- tinuously in a general population [42]. Statistical analysis We tested differences in age and scores in K-ABC and SRS between TD and ASD using Stu- dent t-tests. Sex difference was tested using chi-square tests. Statistical analysis–Regression analysis. Before we applied linear regression, we verified that our data meet the assumptions for regression analysis. Specifically, we used standard methods to verify linearity, normality, homogeneity of variance, model specifications, influ- ence, and collinearity. In some cases, the assumption of homogeneity was violated. Therefore, we used heteroscedasticity-robust standard errors [63]. PLOS ONE Intelligence and social cognition in autism spectrum disorders PLOS ONE | https://doi.org/10.1371/journal.pone.0235380 August 21, 2020 5 / 18 https://doi.org/10.1371/journal.pone.0235380 First, to elucidate different relations between intelligence and social reciprocity in children with ASD from those in TD children, we applied linear regression models to predict the four SRS sub-scales (i.e., social awareness sub-scale, social cognition sub-scale, social communica- tion sub-scale, social motivation sub-scale), respectively, based on the participants’ condition (i.e., ASD or TD), K-ABC sub-scales (i.e., MPS or ACH), age, and sex as well as their interac- tion. We incorporated age and sex in the models because of their possible influences on SRS scores [57, 64]. We did not examine relations between cognitive performance and the diagnos- tic group for the autistic mannerism sub-scale because our emphasis was to ascertain different relations between social reciprocity and intelligence. Second, if a significant interaction effect was found, then we applied post-hoc analysis to elucidate the relation between SRS sub-scales and MPS or ACH further. We predicted the sub- scale based on MPS or ACH controlling for age and sex, respectively, in each condition (i.e., ASD or TD) using linear regression models. Statistical analysis–Coarsened exact matching. The results of simple regression analysis alone can be misleading if patient characteristics are very different between the conditions. In such cases, two options are available to address the imbalances: multiple regression and match- ing. For the detection of class effects, a regression model is often a more powerful tool than matching [65]. However, matching might be preferred in some cases (e.g., when the linearity assumption fails). To investigate the relations between social cognition and general intelligence or achievement further, we matched the groups in terms of MPS and ACH scores, age, and sex. We improved the balance by coarsened exact matching (CEM). Then we performed adjusted regression analysis. In the CEM algorithm, we temporarily coarsen (or categorize) each variable based on its distribution or on natural or intuitive divisions. Each participant is then assigned to one of a specified set of strata in which the participant characteristics are exactly matched on a set of coarsened, or categorized variables. A weighting variable (CEM- weight) is generated to equalize the number of treated and control cases in one stratum. It is used for subsequent regression analysis [66]. We matched ASD and TD groups on MPS and ACH, age, and sex. As a binning algorithm, we used Scott’s rule [66]. This report describes the degree of imbalance between the two data- sets before and after matching by measuring the multivariate L1 distance. The L1 distance, which is a value between zero and one, represents how two groups are balanced in terms of matched variables. Smaller values signify better balance. Larger values signify imbalance. In subsequent regression analysis, we predict social cognition sub-scale based on the condi- tion, and MPS or ACH as well as their interaction with CEM-weight for weighting. If a signifi- cant interaction effect was found, then we predicted social cognition sub-scale based respectively on MPS or ACH, in each condition. For all statistical analyses, we inferred significance for P < .05. All statistical analyses were conducted using software (Stata ver. 15.0; Stata Corp., College Station, TX, USA). Results We found significant differences in age [t(100) = -2.96, p< .01], Mental Processing Scale [t(100) = 3.85, p < .01], Achievement Scale [t(100) = 5.05, p < .01, and SRS total score [t(100) = -11.0, p< .01], social awareness sub-scale [t(100) = -7.64, p< .01], social cognition sub-scale [t(100) = -9.33, p< .01], social communication sub-scale [t(100) = -9.71, p< .01], social motivation sub-scale [t(100) = -5.77, p< .01] between TD and ASD groups. Table 1 pres- ents results. Children with ASD are older. They also had lower K-ABC and higher SRS sub- scale scores. PLOS ONE Intelligence and social cognition in autism spectrum disorders PLOS ONE | https://doi.org/10.1371/journal.pone.0235380 August 21, 2020 6 / 18 https://doi.org/10.1371/journal.pone.0235380 Group difference in associations between K-ABC sub-scales and SRS sub- scales–regression analysis Significant interaction between condition and K-ABC sub-scales was found only in the model for the social cognition sub-scale. For the social cognition sub-scale, in the model using MPS, the interaction effect was found to be significant [t(96) = -2.33, p = .02]. The main effect of the condition [t(96) = 3.57, p < .01] was also found to be significant. No other factor was found to be significant. Similarly, in the model using ACH, the interaction effect was significant [t(96) = -2.19, p = .04]. A main effect of the condition [t(96) = 3.54, p < .01] was also significant. The results are presented in Table 2. The results in the models for other SRS sub-scales are pre- sented in S1 Table in S1 File. Table 1. Characteristics of the participants. TD ASD χ2 t p N 56 46 Sex (\% Male) † 63\% 76\% 2.17 .141 Age in months ‡ 68.5 (6.8) 73.1 (10.3) -2.96 < .001 K-ABC scores Mental Processing scale ‡ 107.4 (13.9) 91.5 (17.7) 5.05 < .001 Achievement scale ‡ 103.9 (14.2) 92.2 (16.3) 3.85 < .001 SRS-T scores Total ‡ 46.9 (6.9) 68.9 (12.9) -11 < .001 social awareness sub-scale ‡ 47.4 (9.4) 61.9 (9.8) -7.64 < .001 social cognition sub-scale ‡ 49.7 (8.2) 69.6 (13.2) -9.33 < .001 social communication sub-scale ‡ 45.7 (6.8) 66.1 (13.8) -9.71 < .001 social motivation sub-scale ‡ 50.3 (6.8) 63.0 (14.5) -5.77 < .001 Numbers are mean (standard deviation) or counts. ‡ Student t-test. † Chi-square test. K-ABC, Kaufman Assessment Battery for Children. https://doi.org/10.1371/journal.pone.0235380.t001 Table 2. Association between social cognition sub-scale and K-ABC sub-scales controlling for age and sex. vs. social cognition sub-scale Coeff. SE β t P>t 95\% CI F p R2 MPS 0.05 0.10 0.06 0.48 0.63 -0.15 0.24 22.16 < .001 .53 Condition (ASD, 1; TD, 0) 49.61 13.91 1.70 3.57 <0.01 22.00 77.21 Condition × MPS -0.32 0.14 -1.04 -2.33 0.02 -0.60 -0.05 Sex -2.56 2.36 -0.08 -1.08 0.28 -7.25 2.13 Age (months) 0.21 0.14 0.13 1.50 0.14 -0.07 0.49 vs. social cognition sub-scale Coeff. SE β t P>t 95\% CI F p R2 ACH -0.02 0.08 -0.02 -0.20 0.84 -0.17 0.14 23.17 < .001 .53 Condition (ASD, 1; TD, 0) 47.52 13.42 1.62 3.54 <0.01 20.88 74.17 Condition × ACH -0.31 0.14 -0.99 -2.19 0.03 -0.58 -0.03 Sex -2.47 2.29 -0.08 -1.08 0.28 -7.00 2.07 Age (months) 0.15 0.14 0.09 1.07 0.29 -0.12 0.41 MPS, mental processing scale; ACH, achievement scale; Coeff, regression coefficient; SE, robust standard error; CI, confidence interval; ASD, autism spectrum disorder; TD, typically developed controls. https://doi.org/10.1371/journal.pone.0235380.t002 PLOS ONE Intelligence and social cognition in autism spectrum disorders PLOS ONE | https://doi.org/10.1371/journal.pone.0235380 August 21, 2020 7 / 18 https://doi.org/10.1371/journal.pone.0235380.t001 https://doi.org/10.1371/journal.pone.0235380.t002 … 1 Running Head: ARTICLE CRITIQUE Article Critique – Task Stream Assignment EDF 5481 Florida International University ARTICLE CRITIQUE 2 Article Critique – Task Stream Assignment Introduction Summary The article I critiqued is called “Interactions Among Online Learners: A Quantitative Interdisciplinary Study” by Pawan Jain, Sachin Jain, and Smita Jain, 2011. The study focuses on the amount of interaction students have with each other and with their professors to identify if interaction levels differ among disciplines. Research Problem The major research problem identified in the study is that there is not enough research concerning the matter of online interaction. The majority of the completed research is very discipline specific and cannot be generalized. The author’s justification for conducting the study was to remedy the lack of prior research on this increasingly important topic. The purpose is “to fill the gap and try to understand the relationship between the interaction and differences in discipline;” however, the authors also noted that this was “one” of the purposes of the study and failed to mention any other purpose. While the authors did not specifically discuss their decision to utilize a quantitative approach in this study, it was clearly justified by their need to examine the relationship between discipline and study interaction. The theoretical basis that is used as the basis for this study is that increased interaction within an online course will ultimately lead to a better designed course and better outcomes for students grades. Further, the authors attest that much of the literature is inadequate for their study due to the specificity of the studies, disallowing them to be universally applied. ARTICLE CRITIQUE 3 The research question is “do the differences in the discipline area impact the overall interaction among learners as defined by the number of learner-learner interactions?” The hypothesis that was tested is “there is no significant relationship between the number of learner- learner interactions and discipline area.” The only noted relationship that could be inferred between the theory and the research question/hypotheses is that there are not any studies that have been conducted that measure said information in a way that can be used for their purposes. They note the importance of learner-learner interaction as a major pedagogical design; however, they continued to note the lack of available research. METHODS SUMMARY Measurement The primary concepts/variables in this study were the observations of the amount of discussion posts by students in 39 different courses across 4 disciplines (College of Education, College of Business, College of Arts and Sciences and the College of Health Sciences). The identified independent variable was the 4 categories of discipline; the dependent variable was the overall interactions per student per week. The conceptual or operational definition of these variables was not mentioned beyond the above listed information. The authors did not address whether the measures were valid or reliable. Research Design Although the authors in this study did not explicitly state the specific research design, one can infer the researchers used a Randomized-to Groups Pretest-Posttest Design. Maturation could be a threat to the internal validity because of the time elapsed between the pretest and between posttest 1 and 2. Experimenter effects could also be a threat to internal validity because ARTICLE CRITIQUE 4 the study does not assure the reader that experimenter bias has not influenced the results. Subject effects could also be a threat to internal validity. Students may have responded differently knowing they were part of a study. The authors did not address whether or not the children knew they were being videotaped or used for a research study, which may have affected their behavior and responses. Because this study only included a sample from the majority one race/ethnic group (White), and was not specific to the SES of the participants, there is a clear threat to external validity. The study may not be generalized appropriately to a larger population. Sampling The population that the authors wanted to study was interdisciplinary students enrolled in online courses. The population from which the sample was taken does not represent all of the appropriate people. The sample was taken from Graduate online courses at a major university in the Rocky mountain region from four different disciplines. The sample is limited in that it excludes all students besides graduate students. There was no mention of a specific sampling technique that was used. I presume that it was a convenience sample. Additionally, there was not any address of the implications of the sampling technique for the findings. There may be issues with external validity based on the limited population of the sample. Since they only sampled graduate courses, it cannot be generalized that the results would be similar for other class levels. The difference in intensity for graduate courses compared to undergraduate courses is not noted as it should be. Data Collection The researchers collected data based on observations of the number of discussion posts per student per week in each course across the four disciplines. The advantage of this data collection is that the information from the courses was archived and researchers needed only to ARTICLE CRITIQUE 5 count the amount of discussion posts. The disadvantages to this use of data collection include human error in counting, as well as the lack of distinction between quality and quantity of the posts. Data Analysis Critique The data was organized using SPSS 15.0 statistical software. Descriptive statistics were utilized to summarize, organize and simplify the data collected for the study Additionally, a one- way analysis of variance was used to find the relationship between the dependent variable and the nominal independent variable, discipline. Follow-up tests were conducted to analyze the pairwise differences among the mean and Scheffe’s post hoc comparison test was employed for this purpose. Results Critique The major findings of the study were that “the mean number of interactions per student per week was 4.76; the standard deviation for this variable was 3.89” (Jain et al, 2011, p. 541). The use of a one way analysis of variance was conducted and concluded that “differences in discipline accounted for 22\% of the variance of the dependent variable” (Jain et al, 2011, p. 542). The results showed that the overall interaction in Arts & Science courses was significantly different than the overall interaction in Health Science courses. The interaction in Business courses was significantly different than both, the interaction in Health Science courses and the interaction in education courses. But no significant differences were found between the interactions in Arts & Sciences courses and interaction in Business courses and interaction in health Science courses and between interaction in education courses and the interaction in Arts & Sciences courses. Overall, the health sciences courses had a higher interaction rate than the rest, leading the authors to conclude that the “interactivity in an online class depends on the discipline ARTICLE CRITIQUE 6 it belongs to” (Jain et al, 2011, p. 543). I am not highly confident in the results because of the nature of the data collection as well as the limited sample used. Additionally, they do not provide for a practical use. Implications of the Findings Critique The conclusions that the authors reached were that interactivity was different based on the discipline of the course; these conclusions are appropriate but the author did not provide practical implications of the study. The only implications the author noted was that additional research would need to be conducted. The authors did not provide any practical information on how the conclusions of this study could be used to further develop courses. In my opinion, the significance of findings was minimal for my area of interest. Online education design and implementation is my area of interest and I did not find this study to be particularly helpful or beneficial. My Contribution Overall the study did not provide any practical information. I would suggest adjusting the data collection method to include additional information about the particular students from whom data was being collected (Are they full time or part time? Are they married? Do they have children? Do they work? If yes, do they work part time or full time?). The information gathered as to which students provided higher levels of interaction would be far more beneficial in designing online courses. I would keep the data collection of the graduate students, but also expand it to include undergraduate as well. I feel it is important to note the difference in interaction between graduate level business courses and undergraduate level business courses. ARTICLE CRITIQUE 7 Simply knowing that there is a difference among interactions does not provide any beneficial information in developing courses and activities designed to increase interaction. The authors noted on more than one occasion the lack of research already in this field; however, I felt that their contribution should have been more significant. They conducted this study, which in my opinion, still leads to a lack of necessary research. Additionally, the actual layout of the information in their paper is not easy to navigate through. There was key information that was left out such as the sampling technique used, as well as any external factors that were not controllable. I would suggest changing the layout of the article and address key information in a clear and easy to follow way. The independent and dependent variables were not clearly addressed until the Results section of the article, when they should have been listed at the beginning of the Methods section. The Theoretical Framework section was well laid out and the authors did a good job of defining any new terms that they used; however, they lacked a clear connection between the question and the theoretical framework. They were not clear as to specifically what the theoretical framework had to do with the questions at hand. Further, the purpose of the study was mentioned in the theoretical framework section as opposed to the opening portion of the study. The paper also noted in the theoretical framework that said purpose was “one of the purposes” (2011, p. 541), but another purpose was never clearly identified. I think it is important to lay out a clear purpose toward the beginning of the article so that the reader can easily identify what they will get out of the article. As mentioned above, there was a significant lack of practical implementations. The results failed to provide information and suggestions on how to move forward. They simply suggest that interaction is good based on their theoretical framework and that in fact there is a ARTICLE CRITIQUE 8 difference in interaction amongst disciplines. I would change the procedures of the study to be able to provide practical uses, such as what can be done to increase interaction for students who have additional responsibilities such as full time jobs, spouses, and children? Information on when the discussion posts were completed as well as the timeframe as to how long the students were given to complete each post, would also be beneficial. I also feel that information should be provided on the length of posts, to determine if the students in each discipline are contributing quality comments and discussion or are they simply replying with “yes, I agree”. In conclusion, I thought that the study lacked focus, and implications. I felt as though the authors took the easy way out in stating that “more research was necessary”, which is generally true of most studies. I felt as though the authors could have completed additional data collection in relation to the particular students in each course as well as expanded the study across undergraduate courses as well. There was no mention in the article about specifically studying graduate students. I was disappointed in the article; I did not feel as though I gained any information from reading it, other than a conclusion that I could have inferred on my own. 9 ARTICLE CRITIQUE References Jain, P., Jain, S., Jain, S., (2011). Interactions among online learners: a quantitative interdisciplinary study. Education, 131, 3, p. 538-544. References Walker, D. D., Stephens, R., Roffman, R., DeMarce, J., Lozano, B., Towe, S., & Berg, B. (2011). Randomized controlled trial of motivational enhancement therapy with nontreatment-seeking adolescent cannabis users: A further test of the teen marijuana check- up. Psychology of Addictive Behaviors, 25(3), 474-484. doi:10.1037/a0024076 10 ARTICLE CRITIQUE