Бразил EXAMINING THE FACTOR STRUCTURE OF AN EARLY CHILDHOOD SOCIAL EMOTIONAL SCREENING ASSESSMENT

Вовед Прашалниците за возраст и фази: Социјалноемоционалниот ASQ:SE кај млади деца со социјално-емоционални задоцнувања се препорачуваат за скрининг во педијатриски ординации и други средини за ран детски развој. И додека повеќето психометрички карактеристики на ASQ:SE биле истражени, факторијалната структура на овој инструмент с¢ досега не била детално проучена. Оваа студија ја проучува факторијалната структура на ASQ:SE со цел да ја утврди нејзината конструктивна валидност, вклучувајќи ја и студијата на сите осум интервали на возраст. Метода Вкупно 13,718 парови деца/родители учествуваа во студијата во САД, завршувајќи еден ASQ:SE тест-интервал. ASQ:SE претставува инструмент за скрининг што исклучиво се фокусира на социјално-емоционалната компетенција кај деца на возраст од 3 66 месеци. Introduction The Ages & Stages Questionnaires: SocialEmotional (ASQ:SE) has been recommended for screening young children for socialemotional delays in pediatric practices and other early childhood settings. While many psychometric properties of the ASQ:SE have been examined, the factorial structure of the instrument has not as yet been comprehensively investigated. This study examined the factorial structure of the ASQ:SE in order to establish its construct validity, including study of all eight age intervals. Method A total of 13,718 child/parent dyads participated in the study in the United States, completing one ASQ:SE test interval. ASQ:SE is a screening instrument exclusively focusing on social-emotional competence in 3-to-66 month old children.


Introduction
The Ages & Stages Questionnaires: Social-Emotional (ASQ:SE) has been recommended for screening young children for socialemotional delays in pediatric practices and other early childhood settings.While many psychometric properties of the ASQ:SE have been examined, the factorial structure of the instrument has not as yet been comprehensively investigated.This study examined the factorial structure of the ASQ:SE in order to establish its construct validity, including study of all eight age intervals.Method A total of 13,718 child/parent dyads participated in the study in the United States, completing one ASQ:SE test interval.ASQ:SE is a screening instrument exclusively focusing on social-emotional competence in 3-to-66 month old children.
A series of eight questionnaires at different age intervals (i.e., 6-, 12-, 18-, 24-, 30-, 36-, 48-, and 60-month) make up the content.Exploratory factor analysis was conducted to evaluate the number of factors that should be retained, factorial structures, and item loading on factors.Confirmatory factor analysis was conducted to further examine the goodness-offit of the fit model to data, based on the exploratory factor analysis result.

Results
One-factor and two-factor structure models were suggested by the exploratory factor analysis results depending on the age intervals.Confirmatory factor analysis indicted that a two-factor structure model was a better fit than a one-factor structure for all intervals.

Conclusion
The items representing social-emotional competence in the ASQ:SE can be classified in two highly correlated clusters, labeled as Emotion and Sociality.The findings supported the construct validity of the ASQ:SE, measured the intended underlying constructs.
displaying attention, initiating contact with people, playing with/alongside friends, participating in playing within a group, and beginning of specific friendships with peers.Emotional competence can be considered as a complicated ability to realize self and others' emotion, to negotiate interpersonal exchanges, to control and regulate emotions (33).Children who are deficient in social-emotional competence may experience difficulties in social interactions and have problem behaviors (34).Social-emotional competence is often hypothesized as a multidimensional construct, reflected in the diverse structures of existing instruments (35)(36)(37).The design of the ASQ:SE did not indicate its underlying constructs by applying subscales or domains within the test.Instead, the conceptual framework posited that social competence and emotional competence are highly related but represent different developmental abilities and behavioral process (17).The aim of this study was to examine the consistency between the theory and the mathematical factorial structure of the ASQ:SE in order to investigate the internal construct validity of the eight ASQ:SE intervals.(21,(44)(45)(46).

Data Analysis
This study examined the factor structure on the eight age intervals of the ASQ:SE using a cross-validation procedure (exploratory and In this study, all parameters of the model were estimated using AMOS (59).

Exploratory Factor Analysis
According to the EFA results, one-factor and two-factor structure models were suggested to be retained depending on the age intervals and methods.Specifically, Parallel Analysis suggested a two-factor structure for all eight intervals; Minimum Average Partial test suggested a one-factor structure for 6-, 12-, and 18-month ASQ:SE, while a two-factor structure for the remaining intervals; Hull method suggested a one-factor structure for 18-, 30-36-, 48-, and 60-month ASQ:SE, while a twofactor structure for the remaining intervals.

Two-factor Structure Model
As a result of discrepancy found among these test methods, a two-factor model was constructed for each interval to compare with the one-factor model (i.e., included all items).To construct a two-factor model, EFA was conducted again for each interval by limiting number of factors to two to obtain the factor loading.Factor loadings and the theoretical framework underlying young children's social-emotional competence were considered when assigning an item to a specific factor.Some ASQ:SE items are the same or similar across different testing intervals (e.g., "Can your child stay with activities he enjoys for at least 3/5/10/15 minutes?" appears on 30m, 36m, 48m, 60m).These items were expected to be categorized in the same factor.Therefore, the mean of these cross-interval items was calculated for making a decision on factor placement.For example, "Does your child destroy or damage things on purpose?"appeared on four intervals (i.e., 30m, 36m, 48m and 60m).Its loading on Factor 1 was: 0.53, 0.56, 0.57, and 0.59 (M = 0.56); the loading on Factor 2 was: 0.13, 0.10, 0.07, and 0,56); вчитувањето на Фактор 2 беше: 0,13, 0,10, 0,07 и 0,07 (М=0,10 0.07 (M = 0.10).Items were categorized to the specific factor with higher mean of the factor loadings.Furthermore, based on the content of items in these two factors, authors labeled these factors as Sociality and Emotion.This proposed twofactor structure model was then evaluated by two psychologists who independently verified the placing of these items and overall factor labeling.A total of 83% (173 out of 211) items were agreed upon by both psychologists.Disagreements on items were then discussed and reassigned.For example, "Do you and your child enjoy mealtimes together?"was reassigned to Sociality.Finally, one of the ASQ:SE developers was consulted to evaluate the current mapping of items.The final version of item mapping can be seen in Table 2. Table 3
approach.For GFI, the one-factor model did not reach the criterion in each interval (> 0.90), while four intervals met this criterion, when two-factor models were applied (i.e., 6m, 12m, 18m, and 24m).For the AGFI criterion, five intervals (i.e., 6m, 12m, 18m, 24m, and 30m) matched our expected standard (> 0.80) when applying the one-factor models, while each of the eight intervals reached the standard for the two-factor model.All intervals reached the criteria of SRMR < 0.08, and RMSEA < 0.08 for both models, except RMSEA for one-factor model in 18m and 36m intervals.
Comparative fit indices were applied to compare the goodness-of-fit between two models.First, comparing the chi-square values between nested models, the two-factor model presented a better fit than the one-factor model with a significantly decrease of the chi-square values for each of the eight intervals.Secondly, all intervals presented lower values in AIC and ECVI when the two-factor models were used, showing better fit indices for the two-factor model in each interval.Factor models reflected a better fit in the younger intervals (i.e., 6-, 12-, 18-, and 24month) than in the older intervals.It is possible that the behaviors representing Sociality or Emotion are easier to define when children are young, which was also reflected with lower correlations between constructs in the younger intervals.For example, the content of items in younger intervals reflects more observational information and less parental interpretation.For example, "Does your child have trouble falling asleep at naptime or at night?" (Emotion) and "Does your child look for you when a stranger approaches?" (Sociality).In contrast, items in the older intervals such as "Does your child cling to you more than you expect?" (Emotion) and "Does your child stay away from dangerous things?"(Sociality) involved more parental judgment and explanation of situational contexts for caregivers when rating the behaviors.

Implications
The current findings provide evidence for the construct validity of the ASQ:SE.The twofactor structure model was supported according to our analyses.These findings may improve further understanding of the measurement and nature of social-emotional competence in infants and young children.Furthermore, the findings support use of an overall score because of the high correlations between constructs, while the two-factor construct implies that ASQ:SE is able to detect particularly social or emotional concern by applying the current factorial classification.

Limitations
Since the ASQ:SE items have been studied for years and proven to be effective in screening for children's social-emotional problems, our efforts attempted to retain all items into the analyses.However, this might have resulted in lower values on the indices of model fit.In addition, each ASQ:SE item has a response option asking if the target behavior was of concern to parents.This question was excluded from our analysis because it was not based on a Likert scale but included a separate response option.However, we might have lost some information regarding parents' concerns for severe social-emotional problems.Furthermore, our sample is rather skewed regarding gender (58% boys), which may limit generalizing the results to other populations.Lastly, the factorial structures we suggested were based on data for a U.S. sample, which may or may not generalize to other cultures and translations.
The results of this study suggested that a two-factor structure model (i.e., Sociality and Emotion) fits the eight intervals of the ASQ:SE better than a one-factor structure, supporting the construct validity of the ASQ:SE.Future studies can focus on the invariance testing for the current two-factor structures using different ethic/cultural populations.

Acknowledgement
We acknowledge the many families and children that participated in our research studies.

Table 3 :
The EFA item loadings of the 6-month interval of the ASQ:SE