You are on page 1of 16

Early Human Development 89 (2013) 561–576

Contents lists available at SciVerse ScienceDirect

Early Human Development
journal homepage: www.elsevier.com/locate/earlhumdev

Psychometric properties of the Brazilian-adapted version of the Ages
and Stages Questionnaire in public child daycare centers
Alberto Filgueiras a, b, Pedro Pires b, c, Silvia Maissonette b, J. Landeira-Fernandez a, b, d,⁎
a
Pontifícia Universidade Católica do Rio de Janeiro, Brazil
b
Instituto Brasileiro de Neuropsicologia e Comportamento, Rio de Janeiro, Brazil
c
Universidade Federal do Rio de Janeiro, Brazil
d
Universidade Estácio de Sá, Brazil

a r t i c l e i n f o a b s t r a c t

Article history: Well-designed screening assessment instruments that can evaluate child development in public daycare
Received 7 March 2012 centers represent an important resource to help improve the quality of these programs, as an early detection
Received in revised form 3 November 2012 method for early developmental delay. The Ages and Stages Questionnaire, 3rd edition (ASQ-3), comprises a
Accepted 12 February 2013
series of 21 questionnaires designed to screen developmental performance in the domains of communication,
gross motor skills, fine motor skills, problem solving, and personal–social ability in children aged 2 to 66 months.
Keywords:
Child development assessment
The purpose of the present work was to translate and adapt all of the ASQ-3 questionnaires for use in Brazilian
Ages and Stages Questionnaire public child daycare centers and to explore their psychometric characteristics with both Classical Test Theory
Brazilian public child daycare centers and Rating Scale analyses from the Rasch model family. A total of 18 Ages & Stages Questionnaires — Brazilian
Classical Test Theory translation (ASQ-BR) questionnaires administered at intervals from 6 to 60 months of age were analyzed based
Item Response Theory on primary caregiver evaluations of 45,640 children distributed in 468 public daycare centers in the city of Rio
de Janeiro. The results indicated that most of the ASQ-BR questionnaires had adequate internal consistency.
Exploratory factor analyses yielded a one-factor solution for each domain of all of the ASQ-BR questionnaires.
The only exception was the personal–social domain in some of the questionnaires. Item Response Theory
based on Rating Scale analysis (infit and outfit mean squares statistics) indicated that only 44 of 540 items
showed misfit problems. In summary, the ASQ-BR questionnaires are psychometrically sound developmental
screening instruments that can be easily administered by primary caregivers.
© 2013 Elsevier Ireland Ltd. All rights reserved.

1. Introduction process that depends, among other factors, on continuous evaluation.
Assessing the development of children enrolled in public daycare centers
Developmental neuroscience research indicates that early life experi- represents one aspect of this evaluation process and might contribute to
ence can have a major impact on cognitive, emotional, behavioral, and so- program enhancement and guide policy decisions [7]. The developmen-
cial development later in life [1,2]. Although most of the brain structures tal assessment of child daycare centers might also help identify children
in early childhood are already present, they are still extremely immature who might need early intervention. For example, Gleason et al. [8]
and thus subjected to a wide range of environmental interactions. These reported that approximately 10% of children between 1 month and
early experiences represent the main underpinning of brain development 5 years of age had some kind of serious psychopathology. Moreover,
that determines the strength and function of several neural circuits [3]. less than 10% of these children were properly identified before they
Accordingly, social deficits, generally associated with poverty and reach school age [9].
environmental degradation, might lead to a disruption of normal brain Most of the early developmental instruments, such as the Bayley
development in children [4]. Scales [10], require specialized training and are time-consuming and
These findings have important implications for political and econom- expensive, which might impose difficulty for a routine examination
ic decisions with regard to public investment in early childhood develop- program in a daycare center. Conversely, some much simpler instru-
ment programs [5,6]. The efficiency of these programs for low-income ments are less expensive and time-consuming and generally designed
families, such as government-funded child daycare centers, is a dynamic to screen for developmental delays. Developmental screening consists
of a brief process of evaluating large numbers of children to identify
those who might be at high risk for developmental delay and for that
⁎ Corresponding author at: Núcleo de Neuropsicologia Clínica e Experimental, Laboratório reason need further evaluation [11].
de Análise de Dados, Departamento de Psicologia, Pontifícia Universidade Católica do Rio de
The Ages and Stages Questionnaire (ASQ) is a screening instrument
Janeiro, Rua Marquês de São Vicente, 225, Rio de Janeiro, RJ 22453-900, Brazil. Tel.: +55 21
3527 2075; fax: +55 21 3527 1187. used for developmental assessment during the first 5 years of life
E-mail address: landeira@puc-rio.br (J. Landeira-Fernandez). [12]. The third edition of the ASQ (ASQ-3) comprises a series of 21

0378-3782/$ – see front matter © 2013 Elsevier Ireland Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.earlhumdev.2013.02.005

The back-translated [21].” Item content points). Statistical analyses used because public child daycare centers only accept children who are older than 4 months. and 60 months of age) designed to screen de.4. 16. 14. Brazilian public to administer these questionnaires. ample. sensitivity. Each translated vals. 42. ceptual correlation among the factors. generally used in the United States were modified to better fit the 12. Spanish [26]. work was to translate all of the ASQ-3 questionnaires into Brazilian Portuguese and explore their psychometric characteristics.24]. Methods to the caregivers who were responsible for a classroom with children within the ASQ-BR age range. child development and education. and a few of the suggestions were incorpo- into Brazilian Portuguese. and the five programs developed by the Early Promotion & Inter. Children occurred between October 12 and December 17. fruit” because children who attend public childcare centers are en- internal consistency. A Cronbach's The Brazilian version of the ASQ-3 adapted for public child daycare alpha equal to or greater than 0. Factors were extracted through principal axis factoring because this item was then evaluated by a multidisciplinary panel of specialists with is the preferable method for factor extraction when employed in an ex- a high level of English fluency and different expertise in psychometrics ploratory manner [38]. Afterward. The ASQ-BR efficients were employed to evaluate the internal consistency of the six items in each domain of the ASQ-BR questionnaires. population and performance of all age intervals in each of the ASQ-BR domains. 24. For example. Therefore. K = 4) because of the likelihood of considerable con- tems. problem solving. public child daycare sys. dimensionality of each of the five domains across the different age inter- fessional experience in English–Portuguese translation. 18. 54. and changes were made when necessary. The multidisciplinary panel was also [39] and parallel [40] analyses were employed to determine the number allowed to make changes to any of the translated items. words which is part of the CFA family [42]. [14–20]. Factor rotation was performed using the oblique and cross-cultural adaptation instruments. each containing approximately 30 children. For ex- Several reports indicate that the ASQ has well-established psycho. gross with “piece of biscuit. so the expression “last skills. Procedure This study was also driven by the need to develop a reliable assess- ment instrument that might be used to help evaluate Brazilian public All of the directors of the 468 public child daycare centers were invit- child daycare programs and allow the screening of children for possible ed to participate in a 1-day.29]. lems of understanding and item comprehension could be detected. Moreover. 22.562 A. method (Promax. each equal to or greater than 0. This is an important issue because almost 18% of Education Secretary of the city of Rio de Janeiro. The project was approved by the PUC-Rio Ethical Descriptive statistics were employed to characterize the study Committee. fine motor skills. Rio de Janeiro. proximately 30 daycare directors. The 2 month questionnaire was not 2. the ASQ is the screening speakers and the multidisciplinary panel. child daycare center typically have several classrooms. The MGFA was performed for each . and Hindi [33]. After- ward. and public pro. of the 20 questionnaires of the ASQ-3 was translated into Brazilian Exploratory factor analysis (EFA) was employed to evaluate the uni- Portuguese by three independent native Portuguese speakers with pro. The daycare director presented the ASQ-BR 2. ASQ-BR and original ASQ-3 were examined by three native American vention Research Consortium [22]. Initially. The unidimensionality of each domain was expected to be found in Conceptual equivalence. Both of these procedures were performed using SPSS syntax ASQ-3 [24] was employed to help solve difficulties that emerged during developed by O'Connor [41]. 4. economics.” and “Cheerio” was replaced velopmental performance in the domains of communication. Korean Comments and suggestions from caregivers were evaluated by the [32]. All of the directors were responsible present work also sought to adapt the ASQ-3 questionnaires to these for taking the ASQ-BR questionnaires to their daycare centers so each public institutions and evaluate whether childcare providers are able child could be evaluated by the daycare caregivers.3. 48. cultural adaptation. 10. However. Norwegian [28. the ASQ has not yet been translated multidisciplinary panel. or “not yet” (0 points) [13]. Only minimal differences instrument that has the most published research in the academic set. French [27]. Indeed. In Rio de Janeiro. / Early Human Development 89 (2013) 561–576 questionnaires for infants (2. in the age range of 4 to 60 months were distributed according to the 20 age intervals defined by the ASQ. 6.1.” Additionally.2. Brazilian context.65 [36] and an item-total correlation centers (ASQ-BR) was based on the original ASQ-3 [13]. For that reason. “sometimes” (5 points). such as test–retest reliability. An invariance assess- syncrasies were considered whenever necessary. Each domain has six items. The Spanish of factors. and 8 months of age). ter. Velicer's minimum average partial grams for low-income families.” “feet” was replaced with “meter. Cronbach's alpha [35] and item-total correlation co- 2. 30. were made to maintain the exact meaning of each item. and pre. in turn. Filgueiras et al. Danish [30]. the caregivers or directors entered the child ASQ-BR data into a Data from the present study were collected from children enrolled in website using a computer located in the daycare center. the ber of the multidisciplinary panel. Participants for administering the ASQ-BR to the children in their classrooms. Each meeting had ap- Brazilian children between 0 and 5 years old attend public daycare cen. toddlers (9. Data collection all of the 468 public daycare centers in the city of Rio de Janeiro. Chinese [31]. The caregivers. 27. the Confirmatory Factor Analysis (CFA) and EFA. and specificity couraged to eat more fruits and vegetables than industrialized food. criterion validity. and 33 months of age). ble for completing the questionnaire with minimal training so prob- such as Portuguese from Portugal [25]. and the 20 ASQ-BR questionnaires ters [34]. a preliminary ASQ-BR version was then assess the impact of Early Head-Start programs and public childcare back-translated into English by a native American English speaker systems. 2010. were detected. such as The Florida Infant Mental Health Pilot Program with high fluency in Brazilian Portuguese.5% of all children within this age range were presented by a person previously trained in the ASQ-BR by a mem- specifically attend public childcare centers [34]. could also be adjusted to the context of public daycare centers. “feed himself a cracker or a cookie” became “feed himself a metric properties in a clinical context. and language idio. Caregivers were responsi- The ASQ has been cross-culturally validated in other languages. Brazilian children do not usually motor skills. Brazil. 2. and personal–social learn their surname until they are 6 years old.3 [37] are considered satisfactory. Dutch [17]. 20. from different public child daycare centers. “inch” was replaced with “centime- schoolers (36. were responsible 2. Although efforts ment was conducted using Multiple-Group Factor Analysis (MGFA). and each item is scored as “yes” (10 name” was replaced with “name of the mother/father. the main purpose of the present rated into the final version of the ASQ-BR questionnaires. ting [23] and appears to be a reliable instrument to measure infant a pilot test was performed with 120 children (six per questionnaire) development in childcare centers [7. 26. this phase. the ASQ has been employed as an instrument to At the end of this phase. 8-hour meeting previously scheduled by the developmental delays.

e. Third. the RS from Andrich [45. Descriptive and reliability analysis Based on this evidence. we used the Test Equating method [48. 1. Both statistics indicate how well items fit the uted across the 24 to 54 month age intervals. Fig. 3. Infit and outfit statistics are ages of the boys and girls were within the estimate range across all expressed as mean square standardized residuals (MNSQ). Chi-squared (χ2) values.. and the percentage of girls ranged from 46.640 ASQ-BR questionnaires were included in the able this analysis. The lower percentage of boys compared with sitive to unexpected responses to items in which performance is far girls across the 18 age intervals was consistent. Finally. These items were the com- best statistical procedure.1.. underlying construct.5. the model by considering which items were common among the scales to should present RMSEA and SRMR b 0.. in turn. and theta's too predictive and thus superfluous for the questionnaire because standard error of the five domains of the ASQ-BR across the 18 age in- the information carried by the item is redundant with the other tervals. Gender differences in mean show that the observed response to an item has excessive variance ASQ-BR scores were compared using Student's t-test. thresholds remain stable was found in the number of questionnaires answered by each age in- throughout the scale.72. we used the One-step Equating method by respecting error variance invariance (to test whether the groups present the same the RS model as recommended by Linacre [48]. A.5 and 1.e. Infit and outfit values greater than 1.4% to a respondent's level (inliers). Results cate the importance of correctly using the PCM or RS analysis to un- derstand the structure of the scale using ordinal items [46. because the ASQ-BR is a among groups. com- item [44. The PCM is conducted when items present unique RS each developmental domain from the ASQ-BR.0%). we excluded difference between the model and empirical data among groups items due to misfit. persons. the Construct Maps of the entire set of items across questionnaires. a unique scale formed by all items across the age cance for χ2 statistics (p value). ASQ-BR scores of the girls tended to be higher than the boys in all swered inconsistently [49]. and polytomous scale. a total of 18 ASQ-BR questionnaires were ana- mation respects the RS and does not affect mean square infit and lyzed.90 [42].000 of the answered ASQ-BR questionnaires (87. the common items were used to build a common Rating Scale (RS) analyses from the Rasch modeling family were scale for each developmental domain. 3. with age intervals as groups. Three Construct Map for each development scale in the ASQ-BR. and Goodness-of-Fit Index (GFI) are reported. Data from the 9 and 10 month questionnaires were ASQ-BR (i. one for Masters [46. measures and fit indices in each developmental domain. This is an indication that the item was an. Infit or out. 10”) for the descriptive statistics and other mea. High variation “0. 1a–e. This analysis is consid. The scores and theta are presented for the entire sample and items of the questionnaire.e. if one has the same instrument with sev. person's av- little variation and thus overfits the model.5%) were distrib- nostic RS parameters.5 indicate that the observed item response has Table 2 depicts the score means. standard deviations. Almost The infit and outfit statistics are the most widely used misfit diag. terval.” and 2 = “yes. Second. then the PCM represents the items. Finally. Item fit of the intervals. we created five calibration matrices the best model for each domain of the ASQ-BR [43]. such as gender. were used to produce structures [46]. Fourth.05). To en. we performed a data transformation by considering analyses.45].5 are considered adequate [48]. a common scale (i.48]. two methods are possible: items from the ASQ-BR. and the infit should be the one with the most power of explanation with no statistical and outfit statistics. we calibrated the entire set of items from the ASQ-BR (i. Four important types of information were Approximation (RMSEA). fit values less than 0. Data from the 4 month questionnaire were excluded be- the traditional “0. The best model tent constructs the item reflects). 5. This type of data transfor. 10” scoring scheme). When Construct Map analysis was performed using all of the calibrated Rasch is used to analyze polytomous data.3 [48]. how many la- (SRMR). whereas the scale presents one factor. Table 1 presents the number of questionnaires answered by outfit statistics. the same.6% ed responses to well-targeted items in which performance is close to (mean = 36.46] and the Partial Credit Model (PCM) from The five common scales that emerged from the Equating. The following linking-equating (p ≥ 0. First. with the exception of the gross motor domain. logit standard error.0.” 1 = cause the sample size was too small to enable statistical inferences “sometimes. 1 for the traditional “0.50] from the Rasch The theta was calibrated to the common scale for each develop- family analyses based on the RS model to link scales and build one mental domain using the linking-equating procedure. This proce- models were proposed to study invariance: configural invariance (to en. The first step was to analyze the fit of the items when calibrated to ance) [42. This was expected because the ASQ-BR questionnaire values between 0. As suggested by Linacre [48]. 2” scoring scheme. we used the RS analysis for the ASQ-BR items by considering the classical m + 1 rating structure [45]. and latent traits [48. in using Winsteps 3. was chosen according to the children's ages. the mean from the respondent's level (outliers). The RS model tested the fit between the empirical data mon and non-common items) to reveal the person's and items' theta and theoretical model of an underlying trait. ent scales among age intervals. metric invari.43]. merged because the items on these two questionnaires are exactly sures for the ASQ-BR in the present study. Rating Scale analyses were performed of the domains. signifi. The outfit statistic. the ered part of the Rasch Measurement Model family [44–46]. “0. thus Conversely. degrees of freedom (df).47]. the full content of the ASQ-BR for each developmental domain. level of measurement error and the other two previous aspects of invari. The percentage of boys in the sample ranged from 25. 2” structure and 5 for the ASQ-BR their respective mean ages and standard deviations. 5. dure allows comparisons of children's performance despite being differ- sure that the groups have the same basic factor structure). sional and share the same polytomous rating structure.. We followed the linking procedures ance (in addition to the same pattern of fixed and freed parameters suggested by Lee and Wright [51]. Consequently.0% to 43. 75% (mean = 64%). A total of 45. from the common used to evaluate the psychometric characteristics of each ASQ-BR scale.51]. Overall. to analyze whether the loadings are equivalent). such as the ASQ-BR [46]. providing valuable information about the relationships between such as ordinal attitude Likert categories. from 0. is more sen. In this case. the item is erage theta (logit) calibrated with the common scale. mon and non-common items calibrated by the common-scale. which boys tended to present higher scores than girls. Finally. The Akaike Information Criterion (AIC) was used to determine procedure was conducted. the and thus underfits the RS. Filgueiras et al. The build the analysis table. The calibration matrices are presented in MGFA was conducted using AMOS 18. eral dichotomous items.e. ranging from 47 (60 months) to 8859 (42 months). Linacre's studies of these two Rasch family models indi. the Standardized Root Mean Square Residual extracted from this common scale: theta or logit (i. in which 0 = “not yet. / Early Human Development 89 (2013) 561–576 563 developmental domain in the ASQ-BR.e. The lower threshold is the same as the higher thresh.5 stratified according to the gender sample. The infit statistic is more sensitive to unexpect.5 to 1. each age interval and the percent distribution of boys and girls and old (i. 1. For example. However.05 and GFI > 0.. RS analysis is performed when items are unidimen.” We did not use the structure of the (n = 10). 40. Therefore. the Root Mean Square Error of intervals of the ASQ-BR). and other polytomous items. The theta for .

564 A. Filgueiras et al. / Early Human Development 89 (2013) 561–576 .

01) were found.45. We conducted an age interval × gender interaction was found (F17. p b 0.605 = 62.18.45. p b 0.45.52. In the gross motor domain. In the problem solving domain. 0. Calibration matrix with all items from the fine motor domain of the ASQ-BR.605 = 246. / Early Human Development 89 (2013) 561–576 565 Fig.45.605 = 100. 1.605 = 231. The communication domain did not (F1. p b 0.45. interval × gender interaction (F7.605 = 171. each person was then calculated with the calibrated common scale In the fine motor domain.605 = 59. With regard to the order of appearance. A.605 = The gross and fine motor skills domains had two of 18 questionnaires 63. enabled an overall analysis by considering all scales from the ASQ-BR p b 0. p b 0. in the personal–social domain.53. present any questionnaires with a Cronbach's alpha less than 0.45.76.605 = 3. Note: The items were numbered according to the order of the first questionnaire (6 months).605 = 73.45. p b 0. p b 0. The p values from the icant age interval × gender interaction (F17. p b 0.01) were found. In the communication domain. Each developmental domain was analyzed separately.01) were found. Calibration matrix with all items from the personal–social domain of the ASQ-BR.01). The numbers in white represent the order of the items from each of the questionnaires by age interval. the subsequent items were numbered according to this criterion.65. p b 0.01) and gender mains across the 18 age intervals.45. Calibration matrix with all items from the gross motor domain of the ASQ-BR. we found a significant age interval × gender with both common and non-common items.56.67. p b 0.33. The subsequent sets were then matched with the previous items if they were the same item in both scales.45.79.45.605 = 57.39.90. ly. Significant main effects of age the theta and respective girls and boys averages were calculated.01) were found. Significant main effects of age interval (F1.33.01) in Table 2. e. Final- possible significant differences between factors. tors. c.65. LSD test that compared gender within each age interval are presented Significant main effects of age interval (F1. Calibration matrix with all items from the communication domain of the ASQ-BR.12).45. The global average of interaction (F17. Fisher's Least Signifi. no significant because we now had comparable theta averages. Filgueiras et al. To identify p b 0.01) and gender (F1. b.65. we found a significant age and gender (F1. p b 0.45.605 = 278.64. Significant Table 2 also presents the Cronbach's alpha for the five ASQ-BR do- main effects of age interval (F1. a. This interval (F1. whereas the problem- 60.605 = 83.45.605 = 83. we found a significant age interval × gender interaction (F17. solving domain had three of 18 questionnaires (17%).45.01) and gender (F1.82.605 = 118. d.01).605 = (11%) with Cronbach's alpha less than 0.01). The same procedure was conducted for the other domains.05. Calibration matrix with all items from the problem solving domain of the ASQ-BR. p b 0.01). p = 18 × 2 two-way ANOVA using age interval (18) and gender (2) as fac.01) were found.01) and gender (F1. The personal– . the average theta presented a signif- cant Difference (LSD) post hoc test was used.45. Significant main effects of age interval (F1. p b 0.

RMSEA = 0. item 2 from acterize a more parsimonious model. The second factor explained 17. χ2/df = 6.51) MGCFA was performed using the six items from each develop- 30 2814 35. Finally.91 (0. item 3 from the 33 month questionnaire. 54.22 (0.9% 19. the configural invariance model. RMSEA = 0.97) 60 47 25.01.08.90 (i. 3.05. df = 216. 45 of the 540 items from all of the question.1% 41. Regarding this index. This fac.60 (1. Multiple-Group Confirmatory Factor Analysis (MGCFA) 24 1454 39.93.34) 26.43 (0. and p b 0. and GFI = 0.0 to char- questionnaire.1%.02. These two RS parameters that presented a two-factor solution for the 10.. Finally.06 (0. from less to more rigor- 48 8528 35.1% 65.95 (0. item 1 from the quate fit indices but also fine parsimony indices to ensure its accep- 24 month questionnaire. p b 0. These authors recommended referring to had 10 of 108 items (9. The NFI and TLI need to be as close as possible to 1. χ2/df = 4.3% 63. The fine motor do- (25%) with item-total correlations less than 0. and GFI = 0. We consid- 42 8859 34.19) 33 3316 34. are summarized in Table 3. However.6% 46. ASQ-BR scales presented good parsimony fit.9% 65.05.56) 53.61 (1.50.01.. were found in only four of the 44 misfit items (9.8% 58. 2. Overall. (b) metric invariance.0% 8. The problem 16 month questionnaire.0% – – the results met the proposed criteria. one overfitting tor incorporated items 3.56) 27 2222 38. [52] stated that a good model needs not only ade- 0.9% 36.90 and NFI = 0.94) 6. presented 4 of 108 items (3. and GFI = 0.65. p b 0.96 (1. Regarding the personal–social domain.92) 24. Based on the parsimony from the 10 month questionnaire.23 (1. any of these models were accepted if Total 45. This factor included items 2.91.10 (1. 8 month questionnaire.4% 57. stantiated the fit of the invariance model.. Filgueiras et al.77) 11. the six items of each domain were consistent across the 18 ASQ-BR SRMR = 0.91. χ2/ alpha less than 0.78 (1.0% 64.84 (0.5% of the variance with an eigenvalue of 2.5% 29.640 36. percentage distributions of boys and girls. and items 2 and 3 from the 36 month ques.03. and GFI = 0. χ2/df = 2. and 4 from the 27 month questionnaire.72. An appropriate model is not only the better fitted but also the item 6 from the 30 month questionnaire). and 6.07.90 (1. SRMR = 0.94) 60 month personal–social domain presented a two-factor solution.. The first factor explained 31. The first factor of the 10 month personal–social domain was ASQ-BR questionnaires showed misfit problems.6% 13. which sub- 20 month questionnaire. Outfit MNSQ problems were . the 6 55 43.83 (1.19. The personal–social domain had TLI = 0. items 1 and 2 from the 42 month questionnaire.01.5% 64.82) 15. The groups were the 18 36 5291 34. The problem-solving domain more parsimonious model. variance as an adequate model for all of the personal–social domain scales when considering fit and parsimony indices.11) explained 19. Overall. and items 1.3%) presented item-total correlations less than 0.7%) with item-total correlations less than However. items 2.4% of the variance Age interval N Percentage Mean age (SD) with an eigenvalue of 1. RMSEA = 0. 2. and 5 from the 42 month questionnaire.04. We report data only from the best model.8% 61. Finally.72 (2. Rating Scale fit statistics Velicer's minimum average partial and parallel analyses indicated that the six items of each domain of the ASQ-BR 18 age intervals presented a Infit and outfit MNSQ statistics were calculated for each item of single factor solution.95) 58. This factor was composed of Number of ASQ-BR questionnaires (N).3 (i. that we did not have sufficient empirical evidence to reject configural in- and 5 from the 54 month questionnaire).91.97 (0.04.87 (1. The 54 month personal–social domain also presented a their respective ages (mean and standard deviation [SD]) across the 18 age intervals. 3.85 (1.18 (0.97) 29.2.49 (1.0% 75.9% of the variance with an eigen- Boys Girls Boys Girls value of 1.89. df = 126.09.77) 9. and 6 were loaded in this factor.9% 60.e.566 A. The fine motor domain showed metric questionnaires.. 5. items 1. RMSEA = 0. item 1 from the 24 month questionnaire.09.3% 56. item 4 from the 22 month measures.04.60. 20 1031 41.7% 53. eight of these questionnaires (44%) df = 4. item 2 from the depicted configural invariance as the best model: χ 2 = 1493.1% 58.8%) with item-total correlations less than 0.3% 64.07 (2. item 1 from the main had TLI = 0.96 and NFI = 0.97 and NFI = 0. The communication do- item 2 from the 54 month questionnaire.89 and NFI = 0. Exploratory factor analysis 3. The communication domain had three of 108 items SRMR = 0.63 (1. 5. p b 0.5% 65.88 in the 24 month questionnaire. df = 108.59) 7.7% 9.02.96 in metric invariance.e. / Early Human Development 89 (2013) 561–576 Table 1 the variance with an eigenvalue of 1.70) contained items 1 and 4.e.48. 2. This domain did not meet any of the established criteria. df = 142.69) first factor was responsible for 48.e.94 and NFI = 0.30. This factor 16 987 39.98) 3. Hair et al.96(0. solving domain had TLI = 0. χ2/ naires (8.2% of the variance with an eigenvalue 12 434 43. The second factor 14 752 42.73) 17. Importantly. The 10 236 36.77 (0. and 60 month ques. questionnaire).06.5% 57.87) 47. the Tucker–Lewis Index (TLI) and Normed Fit Index (NFI) as parsimony item 5 from the 18 month questionnaire.83 (1. item 1 from the 16 month questionnaire. items 1. df = 130. The df = 3. The communication domain presented metric invariance as the social domain had 12 of 18 questionnaires (67%) with a Cronbach's more rigorous accepted model: χ2 = 538.72 (0.17. 44 of 540 items (8.78) 41.56) 8 104 25.3.4.80) 19. and 3 value recommended in the literature [52]). 2. The gross motor domain tionnaire).3 (i.9% 65. This factor was composed of items 3 and 4.95. the personal–social domain had 27 of 108 items had TLI = 0. SRMR = 0.08. respecting the delimited hierarchy.15.78 (1. 4. 5. item 2 from the 27 month questionnaire.2% 17.4% 24. The problem gross motor skills domain presented only one of 108 items (0.68 (0.e.6% 60. item 6 from the 22 month questionnaire. and tance.2% 26.7% 12. all of the the 30 month questionnaire. 4.39 (0.81) 21. which is very close to 0.8. the naire.85.1% 21.92 for error invariance. and (c) error 54 7452 36.99 (0. the personal–social domain (2. The fine motor skills domain GFI = 0. 3.3%) with item-total correlations less than 0. RMSEA = 0.79) 22 955 34.05.91 in configural invari- items 1 and 6 from the 20 month questionnaire.93) of 2.87 (0. and 4 from the 22 month criteria.99) ous: (a) configural invariance. The gross presented a Cronbach's alpha less than 0.10 (3.89 (0. items 2 and 3 from the main had TLI = 0.e. SRMR = 0.5% 74.14) 13.13) invariance variance.21) ment domain in a unidimensional model.98 in metric invariance. item 5 from the 18 month questionnaire.3 (i. invariance with the best results: χ2 = 455. and item 6 from the 36 month questionnaire).74 (1. Items 1.1% 15.73) age intervals divided by the questionnaires of the ASQ-BR. all of the scales from the ASQ-BR substantiated their respective in- questionnaire.35 (0.99.5% 32.4% 6.40) 36.74 (0.78) ered the following hierarchy for the model. items 4 and 6 from the 36 month question.4% of and three underfitting the model).9%) with solving domain presented configural invariance as the best model: item-total correlations less than 0.1%) of the tionnaires.. motor domain depicted error invariance variance as the best model: An item-total correlation test was performed to evaluate whether χ2 = 276.95) 32. we considered items 1. items 1 and 3 from ance. and items 1 and 2.80. and 6.3 (i.01.73) 18 1103 41. 3. The only exception was the personal–social domain the five domains across the 18 age intervals.4% of the variance with an eigenvalue of 1. variance models. 4. two-factor solution. (months) The second factor was responsible for 16.92.7% 47. p b 0. item 1 from the 42 month χ2 = 581.3% 64. Infit MNSQ problems responsible for 34.01.3 (i.

sible exception of the personal–social domain from the 10. suggesting inconsistencies within this domain. and fit greater than 1. Fig. Tsai et al. The results The ASQ personal–social domain involves items related to child inde- confirmed that the ASQ-BR questionnaires consist of a set of easy and pendence and social behavior. and personal–social. the present study represents the first evalu. eight overfitting and 32 caregivers from daycare centers located in Florida were also able to underfitting the model). fine motor. Indeed. For example. Filgueiras et al. 24.5. Cronbach's alphas in the Taiwanese version of the 36 month ASQ Finally. standard error.5. error measurement. item 3 from the 10 month personal– [7].5. . domains across the 18 age intervals are unidimensional. The re- and 4 had outfit less than 0. The results indicated that all five of the ASQ-BR 14 month gross motor skills domain had outfit greater than 1. Importantly. conducted a MGCFA as recommended by Little and Slegers [42].5. Moreover. No items invariance.e. These results are consistent with other studies performed with the Ten domains across the 18 ASQ-BR questionnaires presented two original English version of the ASQ-3 [13] and other versions of items with misfit. the impact of the different types of ASQ-BR invariance The number of common scale items for each domain is the following: models tested in the present study should be emphasized [42. Items 1 and 5 from the 16 month The only exception to these general psychometric characteristics communication domain had outfit greater than 1. Items 2 and 3 from the 20 month problem-solving do. item 3 from the 24 month personal–social domain. we also tested both metric (fac- items for the ASQ-BR scales. including Cronbach's alphas and item-total correlations. This study also found that Taiwanese parents and pre- or more misfit items. Childcare staff knows well the habits and abilities of the children social domain. Moreover. the personal–social domains from eight ASQ-BR Items 1 and 5 from the 18 month problem-solving domain had outfit questionnaires (10. 48. Other studies also detected diffi- the 27 month personal–social domain had outfit greater than 1. 3. These re. Item 3 had both infit and outfit statistics less To further explore the factorial characteristics of the ASQ-BR. 39 items. Item 2 from the the ASQ-BR domains. and items 3 and cause of the low number of items that loaded in each of these factors. such as questions related to gender or strument in daycare centers [7.3.5. results indicated that all of the ASQ-BR domains presented configural problem solving. the gross motor domain also ments with psychometric strength have been adapted for Brazilian public presented satisfactory error invariance. Two of them presented three misfit items. these questionnaires translated into different languages [17.24–33]. A. 54.53]. 42 items. Allen [7] reported that remembering the family name. item 2 from the tionnaires presented internal reliability estimates within expected 30 month problem-solving domain also presented infit statistics greater values. The 6 month communication domain had four 60 month questionnaires. For example. 2. These analyses showed that the communication. on a daily basis when the child is present in the daycare center.5 and outfit greater than 1. For example.5. Items 1 and 4 had outfit less than 0. 42. Items 2 and 3 from the 22 month 54 month personal–social domain presented four of six items with personal–social domain had outfit greater than 1. naires.5 were item 6 from the nient for child development monitoring purposes in a daycare center 8 month communication domain. The Logit.. childcare providers were able to understand and properly administer such as the child's ability to “feed himself with a spoon” or “dress and un- each of the 18 ASQ-BR questionnaires with minimal training.5. meaning that the questionnaires measure the same single presented misfit problems when calibrated together in a common dimension across different age intervals. Items 1. Item 3 from the 14 month gross Exploratory factor analysis examined the dimensionality of each of motor skills domain had outfit less than 0.5. Items 2 and 4 from the 54 month problem-solving domain had outfit [31] reported that the personal–social domain presented the lowest greater than 1. Be- misfit items. Accordingly.” However. and infit and outfit statistics for the common only conclusion that the MGCFA allows us to make is that the ASQ-BR scales of the ASQ-BR using the entire set of items (i. 30. Discussion motor domains depicted adequate metric invariance. such as “toileting skills” or “playing cooper- brief instruments that can be completed in a brief period of time. Finally. Items in the ASQ personal–social domain might also can be properly administered by caregivers as a child developmental in. ilar across all of the age intervals. item 3 from the that might help the screening evaluation process. item 6 from the 20 month personal–social domain. in Table 4.6. dress himself. / Early Human Development 89 (2013) 561–576 567 found in 40 of the 44 misfit items (90. The 60 month gross motor skills domain envisioning possible labels for these factors was impossible. The communication. suggesting that the factor structure and item factor loadings of these domains are sim- Currently. be associated with cognitive skills. 22. 43 items. 6 had outfit greater than 1. main. Items 2 and 3 from item-total correlations less than 0. tion and adaptation of the ASQ-3 for public daycare centers. item and are trained in a wide range of childhood developmental subjects 2 from the 20 month communication domain. no brief and low-cost early developmental screening instru.5. item 1 from 24 month from the have several opportunities to observe the dynamics of child development fine motor skills domain. 45 items. than 1. Items 2 and 6 from was the personal–social domain of some of the ASQ-BR question- the 16 month gross motor skills domain had outfit greater than 1. but we cannot extend this statement to the and non-common items calibrated to the common scale) are presented items' factor loadings or levels of measurement errors. item 1 from the 24 month problem-solving do. Items 1 and 2 from the 10 month gross motor skills do. 43 items. this domain measure the same latent trait with an equivalent level of ation of the psychometric properties of the Brazilian–Portuguese transla. including buttoning and zipping front zipper.” might be re- sults are consistent with previous reports that indicated that the ASQ lated to motor skills. caregivers 24 month communication domain. Public atively with peers. and 5 from the 18 month communication domain had out. 54. 2. 2A–E depicts the five construct maps with the entire set of invariance of all of the ASQ-BR domains. gross motor. only four domains of the ASQ-BR questionnaire had three questionnaire.43].5. school teachers evaluated the same child differently in the personal– Item 4 from the 14 month gross motor skills domain had infit less social domains. than 0. indicating that the items of daycare centers.5. Test Equating and common scales of the ASQ-BR This means that the organization of the item sets based on the factor loadings for each item varies from one questionnaire to another. culty with the ASQ personal–social domain. sults confirmed the information provided by the EFA.5.5.5. which presented a two-factor solution. and item 2 The present study also indicated that most of the 18 ASQ-BR ques- from the 30 month problem-solving domain. main showed outfit greater than 1. gross motor.5. Finally. parents complete a developmental screening instrument is more conve- Items that presented outfit greater than 1.5.5. Item 6 had both infit and outfit greater than 1. with the pos- Items 1. presented Cronbach's alphas less than 0.9%. had five misfit items. successfully administer the ASQ. we than 0. other items that are present in this domain. having caregivers rather than This analysis found nine ASQ-BR domains with only one misfit item. and fine 4. The invariance of the scales was more related to the same basic unidimensional structure. and 60 months of age) greater than 1. tor loading) and error (level of measurement error) invariance. the 27 and main had outfit greater than 1. Given the adequate configural scale.5. both common scales are unidimensional.5.

11) 1.9) 46.45 (0.2) 52.7 (8.2) 48.06)* p b 0.09 (0.85 (0.03)* p b 0.01 0.7 (17.45 1.2 (10.2) 29.67 41.01 0.1 (14.0) 57.01 0.79 (0.8) 30.81 .6 (15.9)* p b 0.19 1.7) p = 0.9) 39.74 (0.03) 1.3 (17.01 1.9 (12.87 (0.01 60 47.56 (0.04) 1.03) 1.36 (0.01 0.79 (0.3 (12.5 (17.20 22 35.1)* p b 0.0) 47.01 0.39 (0.9) 40.2) 46.41 (0.78 (0.06)* p b 0.7 (10.5 (16.15 1.08) 1.2 (13.03) 1.4) 53.6 (13.04) 1.5) 53.3 (12.93 (0.6 (7.3) 32.01 0.01) 1.78 48.0 (8.7) 51.78 (0.03 1.9) 43.01 0.4 (18.2 (11.32 (0.52 (0.2)* p b 0.4) p = 0.0 (3.48 12 36.01 1.4 (10.3) 32.5 (10. Cronbach's alpha (α).3 (18.1) 34.32 (0.0 (10.83 (0.67 (0.8) 46.02 18 33.5) 52.75 43.07)* p b 0.9 (13.34 (0.74 (0.63 (0.12) 1.8) 54.43 (0.19) p = 0.5)* p b 0.02) 1.4) 54.01 33 47.70 33.6) 52.08) 1.9) 49. is also presented.6 (11.0 (10.4 (16. 568 Table 2 Mean and standard deviation (SD) of ASQ-BR raw scores for all of the subjects divided by gender for each of the five ASQ-BR domains across the 18 age intervals.6) 51.16) 1.09)* p b 0.02) 1.02)* p b 0.8 (8.7)* p b 0.06) 1.3 (13.04)* p b 0.04) 1.2)* p b 0.8 (10.06) 1.3 (9.6 (10.05 1.9 (10.3 (10.6) 33.03) 1.0 (10.08)* p b 0.3) 51.01 48 48.14) 0.19 (0.4) 31.29) p = 0.31 (0.15 (0.03 1.5 (9.82 48.88 (0.28) 0.9) 52.6 (9.2 (16.4 (13.8 (7.8)* p b 0.32 (0.7) p = 0.16 (0.5)* p b 0.96 (0.9 (15.08) 0.22) 1.01 1.46 (0.1) p = 0.3) 52.2 (11.06)* p b 0.5 (14.75 47.08) p = 0.3) 47.9 (18.25 0.1)* p b 0.49 (0.73 55. calculated using the One-step Equating method from the Rating Scale model of the ASQ-BR for all the subjects divided by gender for each development domain across age intervals.4)* p b 0.6) 51.5 (17.3 (10.0 (7.2 (13.05) 1.77 52.86 14 31.5 (15.5 (16.33 (0.5 (16.3 (13.9 (10.9 (18.6)* p b 0.27 (0.3 (17.18) 1.9 (11.28 (0.0) 47.32 (0.71 (0.01 1.07) 1.82 1.8) p = 0.19 (0.7 (16.54 1.2) 43.5)* p b 0.04)* p b 0.12 1.24 (0.04)* p b 0.5)* p b 0.41 (0.9) 53.03 36 46.18) 1.95 (0.0) 49.95 (0.3 (16.01 0.41 (0.22) 0.9)* p b 0.34) 0.7 (7.4) 45.2 (13.30 (0.3) 46.10) 1.1) p = 0.01 0.8 (16.55 (0.09) p = 0.03 (0.07) 0.01 1.31 (0.4 (13.1 (18.9 (19.06 (0.4 (11.8 (15.13 0.03) 1.04)* p b 0.01 1.04 0.36 (0.02) 1.03)* p b 0.7) 28.9) 25.02) 1.9 (16.81 52.74 53.1 (14.84 (0.11 (0.04) 1.03) 1.56 (0.84 50.8 (14.6) 29.9 (14.34 (0.3) p = 0.98 0.80 (0.8)* p b 0.01 0.9) 45.98 (0.58 (0.04) 1.66 53.3 (16.16) 1.29) p = 0.01 20 33.3) 46.01 1.1) 55.5) 53.78 0.7) 40. The p values from Fisher's Least Significant Difference (LSD) post hoc test following an 18 × 2 two-way ANOVA are presented to compare genders within each A.01 27 48.39 1.5)* p b 0.4 (12.6) 40.02) 1.54 (0.01 1.1 (16.28) 1.39 (0.23 (0.20) 1.04)* p b 0.2) p = 0.93) 0.4) 32.0)* p b 0.9)* p b 0.06)* p b 0.06) 1.01 1.54 (0.06) 1. / Early Human Development 89 (2013) 561–576 age interval. The reliability index.30 (0.89 (0.7) p = 0.23 54 52.46 (0.77 53.65 (0.10 (0.01 0.75 52.52 (0.08)* p b 0.5 (9.27 (0.44 (0.6) p = 0.30 (0.02)* p b 0.4 (14.03) 1.03) 1.5 (18.27 (0.07) 1.22 (0.2) 48.72 (0.6) 48.2) 50.09) 1.9 (12.2 (13.13 0.06) 1.8) 35.06)* p b 0.01 1.02) 1.0 (8.08) 1.27) p = 0.78 54.6 (13.09) 1.01 1. with significant differences between groups marked with an asterisk.6) 34.8 (12.12 0.16 (0. Filgueiras et al.8 (14.1 (17.09 (0.2 (8.06) 1.89 1.13 (0.8) 54.55 (0.5 (16.7) 49.5) 40.03 1.1) 55.03) 1.66 54.02)* p b 0.49 0.9) 56.1 (16.23 1.06 (0.82 (0.34 (0.2 (15.6) 35.76 38.12 (0.75 52.14 (0.9)* p b 0.17) p = 0.1 (10.6)* p b 0.01 30 50.1 (12.27 (0.06) 0.15 1.3 (16.3 (15.02 (0.18 (0.02)* p b 0.19 (0.05) p = 0.23 (0.03)* p b 0.82 (0.03) 1.05)* p b 0.20 (0.82 (0.01 0.63 (0.06 1.3 (11.7) p = 0.1 (17.1 (19.05) 1.22) 1.33 (0.07 (0.02)* p b 0.7) 55.01 1.3) p = 0.01 1.4 (14.02) 1.7) 53.02 0.03) 1.8 (13.06) 1.53 (0. Respective p values from Student's t-test comparisons between the two groups are presented.2 (8.5 (10.8 (12.01 42 46.29) 1.8)* p b 0.47 (0.8) 41.17) p = 0.2 (14.1 (13.5) 50.2 (13.47 (0.1) 38.16) p = 0.14 0.02) p = 0.04) 1.0 (11.73 (0.3) p = 0.53 (0.05 (0.03) 1.05) 1.54 (0.29) p = 0.1 (8.59 (0.6 (13.01 1.51 (0.19 (0.8 (18.55 1.13) p = 0.03) 1.9) 53.06) 1.3) 36.3) 54.3) p = 0.9 (18.12) 1.03) 1.81 8 37.2) 27. The theta measure and standard error (SE) are presented.9) 37.0) p = 0.02 24 44. Age intervals Communication α Gross motor (months) Mean (SD) Theta (SE) Mean (SD) Theta (SE) Global Boys Girls t-test Global Boys Girls Fisher's Global Boys Girls t-test Global Boys Girls Fisher's LSD LSD 6 37.88 10 26.94 (0.8) 38.06 16 29.18 (0.7) 47.56 (0.9) 50.2) 47.03) 1.06) 0.16) 1.09) 1.01 0.31) 1.7 (17.36 (0.85 (0.87 (0.01 1.

9) 36.7(18.06) p b 0.01 36 0.03) p b 0.6(13.3) 54.30 0.04) 0.12) 1.4) 45.93(0.34(0.71 44.7) 44.08(0.1) 36.90(0.06) 1.06) p b 0.3(12.03)* p b 0.6(14.6(13.42(0.9(13.29(0.01 1.28) 1.14(0.04) 1.08(0.3(16.79 37.5) 42.4) 41.9(12.06) p b 0.6(15.4) 42.05) p = 0.38 30 0.7) 40.04(0.20(0.3) 38.6(15.1(15.35(0.33) 1.10(0.5) 40.4)* p b 0.6(17.14) 1.22(0.01 1.03) 1.24(0.3) p = 0.7(17.0)* p b 0.07(0.67(0.4(13.19(0.05) 1.4(13.01 0.4(18.3(18.86(0.45(0.52 1.7(17.18(0.07(0.54 8 0.79 48.5) p = 0.05) 1.56(0.4) 40.45 (continued on next page) 569 .07) 1.85 38.89(0.7) 35.8(14.8) 38.12) 1.7) 43.04) p b 0.03) 1.03) 1.2(13.37(0.20 0.12) 1.19 12 0.01(0.71 41.7(15.05) p = 0.03) 1.1(13.9(13.0(16.01 1.04 1.72 40.2) 47.6) 48. Filgueiras et al.03 1.45(0.87(0.08) 1.5) 37.1(15.9(17.6) 39.4(15.04) 1.10(0.05) 0.20(0.05) 1.1)* p b 0.6(9.9(13.01 1.94(0.8(15.9) 37.78 39.45(0.2) 42.6(14.10) 1.72 38.7)* p b 0.2) 37.7(14.14 33 0.0(10.14 22 0.14(0.8) 44.01(0.86(0.74 1.15(0.4) 28.3) p = 0.26(0.3) p = 0.5(2.01 0.9(12.03) 1.7(13.25 1.79 37.6) 46.94 1.33(0.4) 47.03)* p b 0.01 20 0.70 44.78 35.77 34.75 36.24) 1.5) p = 0.14(0.5) 41.5) 40.3(17.22(0.36(0.30) p = 0.8)* p b 0.46(0.04)* p b 0.06) 1.07) 1.64 1.1(17.52(0.71(0.4) 47.9) p = 0.0(13.3(5.04) 1.50(0.41(0.02 1.13) p = 0.9) 40.18(0.25(0.4(13.3(13.5) 43.01 0.05(0.16) p = 0.04) 1.04) 1.39(0.36(0.6) p = 0.67 36.0) 47.9) 41.5)* p b 0.8(18.38) p = 0.3(13.35(0.8) p = 0.4(19.14) 1.01 48 0.79 40.Table 2 (continued) A.71 34.02) 1.11) 1.3) 40.05) 1.4) 46.01 0.27(0.22(0.7) 33.4(15.02) 1.14) p = 0.34) 1.12) 1.01 1.4) 38.01 0.28 14 0.10) 1.04) 0.14 0.04) p = 0.4) 34.9)* p b 0.5(12.38 0.5) 37.23(0.07)* p b 0.78 46.5(16.33 18 0.02) 1.31(0.03) 1.71 36.8(17.73 0.1) 42.79 48.14 1.5(15.36(0.23 1.02 1.01 0.9)* p b 0.12) 1.1(15.3)* p b 0.4) 34.5(17.07)* p b 0.3(15.01 1.8) p = 0.03) 1.01 24 0.18(0.5(14.03) 1.24 1.5) 41.86 0.3(16.6) 33.48 1.05) 1.28 0.61(0.85 44.05) 1.8)* p b 0.08 1.6(15.1(17.06) 1.19 1.13 0.95(0.16 1.5(18.39 1.22) p = 0.28(0.2(14.04) p = 0.11) p = 0.07) 1.25(0.85 35.46(0.1) p = 0.2) 33.5) 32.12 1.31(0.17(0.22(0.71 46.5(18.5) 37.6(17.18(0.32(0.0(18. / Early Human Development 89 (2013) 561–576 Age intervals α Fine motor α Problem solving (months) Mean (SD) Theta (SE) Mean (SD) Theta (SE) Global Boys Girls t-test Global Boys Girls Fisher's Global Boys Girls t-test Global Boys Girls Fisher's LSD LSD 6 0.7(13.03) 0.1) 45.10) 1.3) p = 0.21) p = 0.70 45.8(17.8) 40.47 1.01 54 0.06) 1.7(10.8(12.1(14.3(13.08) p = 0.28(0.5)* p b 0.03) p b 0.57(0.3(14.7(13.9(14.39(0.77 35.06) 1.1(13.8(10.2) 33.0(14.5(8.8) p = 0.01 1.1) p = 0.5) 54.5) 47.5(14.03) 1.5) p = 0.17) p = 0.05)* p b 0.30(0.7(12.35(0.3) 38.4) 36.3) 36.1) 42.48(0.02(0.34(0.6) 45.08) p = 0.8) 32.04)* p b 0.60(0.85(0.38 10 0.19) 1.02) 1.01 1.01 0.19(0.3(15.12) p = 0.8)* p b 0.9(16.4) 41.11(0.29(0.16) 1.7) 48.9) p = 0.05)* p = 0.04) 1.72 46.60(0.3)* p b 0.9) p = 0.41(0.3) p = 0.45(0.3) 38.03) 1.16) p = 0.7) 49.8) 35.34(0.86 16 0.6(17.0(17.19(0.64 41.3(17.5) 41.02 0.91(0.7)* p b 0.57(0.7) 33.8)* p b 0.04) 1.2(17.06) 1.7(14.5) 37.6) 42.05) p b 0.67 34.09) 1.38 1.09) 1.62 47.2) 34.7(15.21) 1.4) p = 0.03) 1.2(16.8(17.4(14.01 60 0.78(0.3) 40.03) 1.05) 1.8) 36.78(0.15(0.9(18.08) p = 0.44) 1.2(16.64 41.8(15.44(0.9(13.09(0.2(18.33(0.04) 1.67 1.03 27 0.04) p b 0.5(17.14(0.01 1.6)* p b 0.55 1.05 1.04) 1.02 42 0.03(0.47 0.22(0.84(0.9) 37.48(0.7)* p b 0.81 1.64 39.79 41.6(14.03(0.2(16.2(17.43(0.74(0.01 0.4) p = 0.01 0.89 43.71 42.04) p = 0.4(15.01 1.8(14.43(0.8(18.4(10.7(16.00(0.03) 1.8(12.04) 1.6) 36.28 0.5(16.69 42.74 40.7(15.02(0.90(0.5) 47.41(0.03)* p b 0.04) 1.4(13.2(17.67 39.21(0.6(12.31(0.14) 1.03) 1.

570 A.78(0.04) 1.7) 51.23) 1.36(0.08) 0.51(0.17(0.9) 45.9(11.66 42 0.03) 1.9) 30.72 36.03 0.5) p = 0.19(0.7) 32.0(13.6(11.1) 37.01 1.05) 1.01 1.6)* p b 0.1(11.79 33.83(0.02) 1.11(0.52 54 0.76 14 0.28) 0.0(13.0(14.5) 35.9)* p b 0.1) 31.8(17.2(11.1(16.67 47.7(15.01 0.01 0.3) 45.6(18.9(15.57 24 0.14(0.7) 32.5(10.08(0.1(13.17) 1.1) 48.6(9.14(0.9) 33.77 33.0) p = 0.06) 1.05)* p b 0.09(0.06) 1.04) 1.01 1.67 48.94(0.01 0.25(0.71 18 0.06) 1.5) 29.52 .04) 1.16) 1.12) 0.6)* p b 0.64 22 0.46 0.4) 47.9) 42.6(14.03)* p b 0.48(0.2) 39.22) 1.50(0.65(0.07) 1.59 27 0.81 0.03) 1.5) 40.13(0.0(11.63 38.9(12.01 1.03) 1.3) 50.01 0.53 60 0.74(0.6) 47.67 10 0.7) 46.86 0.64 0.4(16.49(0.16 0.2(12.77 30.5) 32.04)* p b 0.8)* p b 0.06) 0.68(0. / Early Human Development 89 (2013) 561–576 Table 2 (continued) Age intervals α Personal–social α (months) Mean (SD) Theta (SE) Global Boys Girls t-test Global Boys Girls Fisher's LSD 6 0.7(12.64 20 0.73(0.06) 1.80 34.2)* p b 0.01 1.08) p = 0.8(13.22)* p b 0.19 0.61 36 0.69(0.63 34.21) 1.8(12.43(0.45(0.72(0.78 45.5(15.5) 35.68(0.00(0.0(15.3(15.04) 1.4(12.7(11.04) 1.1(11.10(0.12(0.7(16.2(13.73(0.03) 1.8(11.18) p = 0.03) 1.53(0.1)* p b 0.57 48 0.03) 1.01 1.02)* p b 0.19(0.91(0.09(0.2)* p b 0.5) 38.18) p = 0.01 1.68 49.08)* p b 0.8(12.05)* p b 0.0)* p b 0.50(0.6(11.03(0.01(0.05) 1.2) 46.01 1.0(8.24) p = 0.8) p = 0.09(0.4) 32.6(14.4(12.1) 33.00(0.14) 1. Filgueiras et al.03 0.67 49.07) 1.7)* p b 0.85(0.01(0.0(15.01 1.6) 35.18(0.0(10.8) 39.7) 38.7(13.11(0.2) 42.03) 1.1(10.07)* p b 0.18) 1.65 35.1(12.63(0.69 46.01 0.3(11.01 0.01 0.03(0.54(0.53 33 0.01 0.34 1.20(0.7(18.5) 45.01 0.9(13.2(10.46(0.08)* p b 0.47(0.5) 48.01 0.72 35.04(0.2) 35.63 30 0.70 51.6) p = 0.9)* p b 0.79 41.10) 0.02 0.43(0.62 46.24 1.04)* p b 0.1(14.03 1.03 1.76 16 0.6(16.06)* p b 0.03)* p b 0.22(0.17(0.58 12 0.70 8 0.7) 48.5) 50.20(0.5)* p b 0.5(13.05 0.39 1.01 1.9)* p b 0.06) 1.04) 1.6(11.05)* p b 0.07) 1.1) 37.3(10.4)* p b 0.0(11.4) p = 0.09) p = 0.38 0.2) 46.7(16.8(16.

61 1.09 (0.42 1.72 1.77 2.98 (0.17) 0. such as might have problems related to culturally inappropriate content or self-regulation.23 0.64 1. For example. The items with misfit Infit and outfit statistics evaluate different aspects of an item based depicted in the RS analyses probably showed problems in one specific on differences in how they are calculated. adaptive functioning.21 1.73 1.83 1.96 (0.97 (0.23 (0.37 0.05) 0.18) 0.99 (0.97 (0. / Early Human Development 89 (2013) 561–576 571 Table 3 Minimum (Min) and maximum (Max) mean square infit and outfit statistics and respective mean and standard deviation (SD) for each of the five ASQ-BR domains across the 18 age intervals.30) 0.88 1.98 (0.07 (0.00 (0.34) 0.89 1.01 (0.15) Out fit 0.5).89 1.00 (0.01 (0.59 1.19) 20 Infit 0.23) 0.38 (0.56 1. suggesting that they evaluates different aspects of the personal–social domain.04 (0.23 1.17) 0.83 1.08) 0.83 1.09) 0.13) 0.47 (3.22) 0.00 (0.18) Out fit 0.25) 0.22 0.81 1.13 1.68 1.79 1.07 (0.84 1.35)a 0.02 (0.01 (0.75 1.99 0.07 (0.70 1.00 (0.86 1.26 0.10 0.18 0.56 1.7%) presented instrument for possible developmental delays in social–emotional func.50 0.22 1.99 (0. but when calibrated with other scales.35 1.07) Out fit 0.72 1.29) 0. 72.78 1.20 (0.03 (0.03 (0.10) 0.06) 0.40 1.13 0.78 1.20 0.30 1.11 1.00 (0.74 1.48 (1.42)b 0.74 1.49 1.26) Out fit 0.13) 0.62 1.89 1.01 (0.20 (1.32 1.5 (32 of 44.14 1.91 1.34 1.02 (0.43 1.99 (0.09 1.08) 0.65 2.88 1.76 1.56 1.19) 0. Accordingly.06 (0.38)b 0.32 (0.77 1.08) Out fit 0.23 1.98 1.98 (0.67 1.00 (0. b 2 item with MsSq fit statistics outside limit.20) 14 Infit 0.31) 0.16) 0.65 1.02 (0.25) 0.00 (0.67 1.15 0.00 (0.67 4.10 1.45 2.50 1.27) 0.07 (0.32) 0.07 (0.92 (0.14) 0.30 0.62 1.05 (0.00 (0.03 (0.82 1.11 0.36) 0. There- tonomy.10) 0.18) 0.14 0. the misfit was dilut- importance on extreme responses and is weighted according to items ed.00 (0.01 (0. and 27 month questionnaires.96 (0.79 1.27) 0.25) 18 Infit 0.92 1.13 1. ric characteristics that cannot be analyzed using Classic Test Theory To understand the relationship among items in the entire set for each techniques.46 0.5 provided additional psychometric support for the good properties and are well distributed through the latent trait.00 (0.82 1.12) 0.98 (0.74 1. indicating that the majority of the ASQ-BR items targeted statement is based on the fact that no items presented misfit problems the intended construct.13) 0.36 0.18 1.61 1.53 (1.98 (0.50 1.11) 0.70 (1.14 1.82 1.81 1.07 (0.01 (0.13) Out fit 0.3%).62 1.46 1.01 (0.25 1.27) 0.00 (0.49 1.86 1.09 0.29) 0.20 0.36 1.16 0.34 5.70 1.26)a 0. Accordingly. Squires et al.04 (0.35 1.19 0.58)b 0.33) 0.09 (0.50 1.82 1.17 1.81 1.27 1.79 1.03 (0.40 1.15) 36 Infit 0.24 1.16 1.00 (0.57 1.21) 0.69 2.07) 42 Infit 0.09) Out fit 0.99 (0.12) 0. d 4 items with MsSq statistics outside limit.59 1.68 1.13 0.00 (0. This ASQ-BR.14 1.20) 0.37 1.24) 0.15) 0.14) 0.98 (0.24 1.34 1.65 1.19 1.12 0.14 1.79 1.08) Out fit 0.47 1.42 (0.49 1.99 (0.97)a 22 Infit 0.34 1.53 1. The outfit sta. This is a screening regard to infit statistics.72 1.81 1.01 (0.11 (0. once they were calibrated in the common scale.30 1.96 (0.82)c 0. fitting problems.99 (0.51 1.96 (0. compliance.38 1.00 (0.11) 0.49) 0.10) 0. only four of the 540 items (0.02 (0. In these other scales.50 0.06 (0.45 1.18) 0.14) 16 Infit 0.23 0.76 1.48 1.87 (0.73 1.00 (0.22 1.08 0.10 0.64 1.59 1. Our hypothesis is that the calibration in Equating makes room for dif- that have better alignment to the person's level of ability.24 0.99 (0.18 1.55 1.47)a 0.93 1.00 (0.02 (0.11) 0.79 1. development domain from the ASQ-BR.00 (0.26 0.32 1.07) 0. specialists to consider alternative adaptations or language evaluation.68 1.13 0.72 1.14 0.99 (0.15) 54 Infit 0.01 (0.97 (0.60)c 0.49 1.21 1.50 1.17 0.82 1.99 (0.70 1.26 1.19) 0.81 1.62 0. difficulty and the person's ability.65 1.00 (0.22) Out fit 0.16 (0.18) 60 Infit 0.13 0.73 1.25 1.04 (0.27 0.00 (0.10 0.70 1.36) 0.19) 0.26) 0.02 (0.10 0.33) 0.15 0.28 1.83 1.01 (0.90 1. To deal with the complexity of personal–social abilities.11) 0.18) 0.00 (0.74 1.70 1.20 0.06 (0.32 1.51].02 (0. ferent fits of the entire item sets.16) 0.68 1.73 1.00 (0.08) Out fit 0.84 1.03 (0.00 (0.38 1.94 (0.10) 0. ASQ-BR items fitted the model particularly well with [54] developed the ASQ Social–Emotional (ASQ-SE).18) 0.72 1.83 1.97 (0.26 0.14) 0.18) Out fit 0.96 (0.85 1.26 1.97 (0.55 1.08 0.13) Out fit 0.94 (0.80 1.88 1.04 (0.14 0.79 1.00 (0.77 1.16 1.16 (0.00 (0.11) 0.86 1.11) 0.24)a 0.59 1.08 1.51 (0.68 1.48 (0.84 1.37) 0.88 (0.85 1.88 1.11) 0.21) 0.88 3.31 1.28) 0.31 0.18) 0.18 1.01 (0.17) 0. such as how well an item fits into a unidimensional logis.81 2.86 1.25 1.13) 0.87 1.17) 0.58 1.17) 0.84 1.92 (0.76 1.48 0.27 1.77 1.20) 0.10 (0.99 (0.99 (0. c 3 items with MnSq fit statistics outside limit.53 1. ambiguous wording and thus lead to item misunderstanding.81 2.19 (0.85 1.12) 0.18 1.76 1.12) 0.70 1.61 2.05 (0.44)b 24 Infit 0.03 (0.23 2.26) 0.77 1.20) 0. these items might need to be further reviewed by our panel of Rasch family analyses allow the evaluation of ASQ-BR psychomet.83 0.09 1.02 (0.50 1.15) 0.97 (0. e 5 items with MnSq fit statistics outside limit.76 1.85 1.79 3.91 1.37 1. the same item ap- sensitivity in instances when a large difference exists between an item's pears in the 20.63 1.74 1.19) 0.13 0.19 0.45 1.81 1.80 1.28) 0.99 (0.82 1.35 1.59)b 30 Infit 0.20) 0.83 1. A misfitted item in one specific age tistic is not weighted and is sensitive to extreme off-targeted responses. The infit statistic places less age interval.22) 0.77 1.98 (0.11 1. Most of the misfit items presented outfit statistic tioning for children 6 to 60 months of age.46 1.14) 33 Infit 0.49 1.01 (0.87 1.85 1.22 0.54 1.30 1.28) 8 Infit 0.57)b 0.17 (0.01 (0.14) 0.70 1.22 1.05 (0. This analysis allowed us to infer that items from the ASQ-BR have range of 0.68 1.03 (0.12) 0.99 (0.83 1.81 1. the ASQ-SE values greater than 1.77 1.13) 0.14)d 0.14) Out fit 0.29)a 0.66 1.67 1.86 1.44 1.14) 0.76 1. infit statistics are more useful for assessing the utility of 3 from the 22-month-old personal–social questionnaire presents an outfit an item because they are not affected by outliers [55].24 (0.17 1.01 (0.98 (0.69 1.24) 0.63)a 0.11) 0.93 1.10 (0.30 0.39 1.89 1.78 1.01 (0.15 (0.77 1.26) 0.00 (0.02 (0. au.69 1.91 1.99 (0.14) 0.02 (0.22 1.78 1.26 1.08) 0.19 1.03 (0.12 (0. However.55)b 0.22) 0.03 (0.97 (0.20) 0.09) 0.14) 0.27 2.99 (0. affect.06 (0.83 1.31) 0.29) 0.07) 0.81)b 0.17)a 0.22) 0.87 1.19) 0. communication.20 1.36 0.08 (0.16) 0.86 (0.13) 0.71 1.15 1.01 (0.02 (0.03 (0.03 (0.97 (0.55 1.22) 0.78 1.00 (0.11) Out fit 0. interval could be fit to items from other age intervals.92 1. Age intervals Statistics Communication Gross motor Fine motor Problem solving Personal–social (months) Min Max Mean (SD) Min Max Mean (SD) Min Max Mean (SD) Min Max Mean (SD) Min Max Mean (SD) 6 Infit 0.07) 0. Conversely.89 (0.90 1.98 (0.99 (0.01 (0.86 1.09) Out fit 0.08 0.13 1.11) Out fit 0.20 1.28) 0.79 1.00 (0.36)b 0.99 (0.77 1.52 1.32 1.77 1.17 0.19 1.89 1.75 1.15) 0.97 (0.18) 0.09) 0.15 (0.06) 0.89 (0.21) 0.04 (0.22) 0.83 1. The fact that more than 90% ent age intervals in a common scale using the One-step Equating method of the 540 items had infit and outfit MNSQ within the acceptable [50.98 (0.69 1.09) Out fit 0. 24.48 1.73 1.13) 0.03 (0.06 0.09)a 12 Infit 0.44 4.63 1.76 1.18 1.01 (0.85 1.20) 0. and interaction with people.43 9.77 1.26 1.48 1.25 1.84 1.04 (0.21 0.92 1.78 1.99 (0.15) 48 Infit 0.28) 0.15 0.57 1.32)a 0. A.25 1.28) 0.5 to 1.00 (0.26 1.11) 0.49 1. we linked the scales of the differ- tic model for the probability of responses.10) 0. item For that reason.39 (0.35 0. It asks about the “toileting skills” of the child but is the only statistics are less critical for item evaluation because they are focused on item that addresses this matter in the scale.18 1.31) 0.77 1.28 1.00 (0.97 (0.83 1.14 1.32)b 0.11 (0. outfit issue (>1.20) 10 Infit 0.56)b a 1 item with MsSq fit statistics outside limit. The calibration .97 (0.83 1.95 (0.49 0.88 1.84 1.99 (0.31 1.85 1.26) 0.89 1.22) 0.14 1.11 1.24 1.12) 0.39) 0.84)a 27 Infit 0.75 2.87 (0.67 1.65 1.47)e 0.09) 0.65 1.18 0.89 1.96 (0.51 1.74 1. fore.50 1.00 (0.98 (0.48 1.72 1.12 1.34) 0.50 1.36)a 0. the outfit problem is only shown in the 27-month scale.06 (0.10) 0.15 (0.14) 0.15 0.02 (0.85 1. Filgueiras et al.07) Out fit 0.12 0.19) 0.20 1.91 1.

38 .99 32 .07 1.30 .23 26 .02 1.15 25 −.01 26 −.93 .95 11 .04 .19 .92 .94 14 .01 1.96 .03 1.11 1 −.80 3 .01 .48 .03 1.04 .02 1.90 20 −.98 .94 .01 1.95 10 .79 23 −.02 1.33 1.39 .11 41 −.88 .01 1.01 1.03 .19 .31 .96 28 .51 .07 .09 20 −.06 1.07 .03 .00 19 −.27 1.04 1.26 .02 .56 .01 15 1.06 1.14 1.00 .04 1.96 23 −.01 .19 .07 .96 17 .96 .84 15 −.11 .32 .04 .02 3 −.97 . Infit Outfit Item Item difficulty Stand.02 1.12 .01 1.17 .05 1.03 1.03 22 .89 24 −.02 1.01 .02 1.96 .98 2 −.99 1 −.96 32 −.08 1.06 1.14 .97 .16 1.11 1.84 – – – – – 26 −.92 .03 .02 .03 1.05 .22 1.95 .08 .90 33 1.02 1.03 16 .98 .15 1.14 .16 .04 1.10 16 −.05 .92 21 1.05 – – – – – 1 −.05 3 −.89 .30 .97 .39 .98 39 −.97 34 −.51 .32 .07 .99 .04 1.91 1.04 .01 25 .00 38 .98 .21 .08 21 −.21 .07 1.30 1.08 42 −.18 1.00 .94 1.99 .03 .19 1.53 .14 .05 .90 2 −.56 .59 .02 .07 .40 .05 1.22 .89 18 −.12 .02 .01 1.15 .02 1.86 38 −.91 7 .08 1.01 .00 1.97 29 −.97 A.01 5 .04 1.63 .04 1.86 .04 1.78 28 −.44 20 1.03 1.21 .01 .85 .03 1.03 1.19 1.12 .97 .03 1.02 1.24 .63 .05 .08 32 .04 6 .53 .97 8 −.13 . Common scales of the ASQ-BR Communication Gross motor Fine motor Problem solving Personal–social Item Item difficulty Stand.01 13 .98 .11 1.04 12 .00 4 −.08 1.96 .04 .08 43 −.94 .04 1.99 .01 1.15 .64 .e.05 4 .38 .15 .87 .02 1.17 .18 .01 1.98 .17 .02 14 −.05 1.01 .81 .86 .82 .03 .05 1.04 41 .04 1.03 1.00 10 .08 1.18 .01 .06 1 −.10 39 .04 1.01 1.96 – – – – – 5 −.10 38 −.26 .95 41 −.98 .10 1.23 .01 .16 .01 1.69 .02 15 .10 .09 37 .92 .97 30 −.08 27 .02 .05 .00 .09 1.86 .04 1.07 14 .02 1.95 .02 1..96 16 −.17 .02 1.04 33 −.01 16 −.03 1.93 8 .98 1.99 30 −.54 .03 1.79 7 −.70 .87 .04 1.97 .02 1.01 .00 .97 .11 32 .46 .10 .80 .12 .09 1.12 1.69 .01 .24 .02 1.99 .00 37 −.20 4 −1.01 .11 1.11 .03 .01 .04 .86 27 −.24 .09 34 −.03 1.02 1.92 36 −.02 1.96 28 −.97 29 −. / Early Human Development 89 (2013) 561–576 19 .04 1.08 1.06 .97 9 −.02 1.06 1.44 .02 1.08 .00 39 −.06 33 .04 1.48 .98 36 −1.02 .96 .99 .94 37 .11 11 .89 .23 1.20 .03 1. 572 Table 4 Item numbers in the common scale (Item).40 .02 1.83 .53 .29 .05 .01 1.16 13 .01 .17 .02 1.91 25 .93 .34 .03 9 .07 .02 12 .05 1.03 1.09 1.07 1.02 1.89 5 −.02 .99 .09 .02 21 −.96 39 −.93 .03 1.28 .96 .02 1.07 1.14 8 −.98 22 .97 12 −.96 38 .96 .16 23 2.94 13 .13 .03 27 −.09 1.03 1.26 .78 .02 .01 1.95 37 .01 1.73 11 −.80 38 −.87 1 .02 19 .02 .01 .04 24 .57 .00 31 .05 1.86 43 −.88 26 .88 .41 .01 . Infit Outfit in logits error in logits error in logits error in logits error in logits error 16 1. logits.96 27 .88 .09 1.60 .10 1.02 1.28 .01 1.36 .25 .00 6 −.94 .08 1.01 1.98 .73 .14 1.82 .72 – – – – – . when calibrated to the equated scale.04 1.40 .39 .94 – – – – – – – – – – – – – – – – – – – – 4 −1.01 .03 1.09 15 .02 .03 1.11 1.14 .09 1.81 30 .02 1.95 .01 .42 .06 .32 .05 .04 .80 42 −.91 17 −.47 .99 6 .03 30 .09 .72 .10 1.13 .02 1.01 .88 .00 .11 1.03 1.02 .03 1.18 .06 36 .00 41 −1.01 1.42 .73 .33 .24 .03 1.07 .18 .81 .92 .02 1.09 1.93 26 .00 .05 1.19 1.05 1.81 5 .04 1.97 .01 1.13 1.01 .02 1.01 .35 .89 .06 1.93 .93 39 .96 .09 .00 .06 1.32 .48 .07 13 .89 10 .42 .05 1. Infit Outfit Item Item difficulty Stand.17 1.86 24 −.18 .00 1.17 .92 21 −.17 1.01 31 −.12 .96 24 −.90 .05 29 .03 23 −.12 .03 .27 .03 35 .00 .02 .39 .03 1.00 1.82 .15 1.35 .07 – – – – – – – – – – – – – – – – – – – – 28 −1.04 1.96 .07 1.03 .90 .04 1.01 .03 .01 .22 .06 35 −.02 34 .03 1.86 .03 .91 14 .01 1.07 35 .27 .04 1.04 7 .98 27 .01 1.07 .26 .64 .87 .04 1.13 .86 .95 .01 1.06 .01 .98 2 .90 19 .93 29 −.01 .05 1.06 1.99 31 −.02 .05 .13 .03 10 .10 .88 .04 1.05 1.04 .04 1.95 .04 .05 1.06 .03 2 .03 .03 1.85 18 −.10 .96 36 .04 .04 1.01 1.05 1.01 .89 .02 1.04 33 .03 19 .49 .01 .95 17 .35 .01 9 −.03 37 −.02 .46 .01 .19 .35 1.11 .77 36 −.97 .01 .00 .18 .95 .24 .09 1.03 1.08 33 .02 1.03 .52 .02 1.03 .99 18 .02 30 −. and infit and outfit statistics for each development domain when the items were calibrated together using the One-step Equating method for linking the scales of the ASQ-BR.94 .03 3 −.05 1.37 .09 11 .02 1.02 1.39 .19 .05 1.05 1.10 .10 1.06 .58 .05 .05 .16 .05 1.09 10 .07 1.01 .04 1.16 .01 1.07 .03 1.96 42 1.02 1.83 29 .45 .97 3 −.01 .10 .26 .02 .40 .83 .35 .04 .92 .01 1.27 .04 6 .22 .08 9 −.92 5 −.16 .50 .13 .97 .01 1.09 1.10 1.37 .04 1.24 .01 .02 1.05 1.03 .85 25 .97 .90 43 .25 .27 .94 .08 1.98 9 −.96 25 −.53 .02 1.07 1.77 .98 35 .54 .03 1.19 8 .03 40 . standard errors.99 . Filgueiras et al.01 1.97 12 .02 .42 .05 8 −.03 35 .03 1.01 1.72 .00 – – – – – 21 −.97 28 .01 .43 .43 .00 1.02 .33 . both common and non-common).90 .07 17 −.01 .05 1.65 .07 .02 .11 1.83 .02 1.09 .25 .03 11 .99 .01 1.03 12 −.01 1.02 .99 22 .03 1.13 1.43 . considering all items (i.03 1.11 .14 .85 .14 1.35 .53 .00 18 −.01 .02 .22 .01 34 −.07 1.05 1.82 32 −1.03 .14 .01 31 −.02 1.02 1.01 .08 15 −.99 .06 1.11 . Infit Outfit Item Item difficulty Stand.02 1.03 45 −.01 1.22 1.04 22 −.10 .13 .11 1.00 .92 .08 1.04 4 .36 .04 34 .60 .01 23 −.08 1.07 40 .45 .17 1.48 .28 .92 .09 1.07 .96 22 .91 .98 2 −.05 .97 31 .03 1.46 .82 18 −.35 .83 20 −.15 .00 40 −1.02 .90 24 .86 40 −. Infit Outfit Item Item difficulty Stand.01 .05 .13 .03 1.97 .67 20 −.08 1.02 .97 14 −.04 .08 1.04 1.96 .04 1.44 .01 1.07 1.19 .05 1.95 17 .01 .94 13 .11 1.91 .03 1.99 .02 1.89 6 −.95 .79 .28 .07 44 .92 .80 7 .01 1.96 7 −.11 42 −.07 .

Note: Example of items: logit +0. logit −1. logit +0. Note: Example of items: logit +1. Construct Map for the personal–social common scale of the ASQ-BR. item 19.8. item 1.2. item 23. “Does your child say 15 or more words in addition to ‘Mama’ and ‘Dada’?”.’ ‘stronger. 2. item 27. his blanket on his bed. item 40. does your baby pick up a crumb or Cheerio with the tips of her thumb and a finger?”. or shirt by himself?”. ‘Tell me about the ball. item 13. item 26. does he know his toys belong on the toy shelf. This procedure enabled us to build a Construct ty of the distribution of items and persons in a scale [48]. “If you call your baby when you are out of sight. item 37. Construct Map for the gross motor common scale of the ASQ-BR. does she reach for it with both hands?”.5.2. item 28. logit 0. item 15. Construct Map for the problem solving common scale of the ASQ-BR. logit −0. “Does your baby get into a crawling position by getting up on her hands and knees?”. item 6.5. logit −0. “Does your child wash her hands and face using soap and water and dry off with a towel without help?” .5.9. logit 0. “After watching you draw a line from the top of the paper to the bottom with a crayon (or pencil or pen).4. The item dif- Map of the five ASQ-BR developmental domains. “Does your baby climb onto the furniture or other large objects. such as ‘heavier. item 32. ficulties of all of the ASQ-BR's domains were normally distributed and Fig. “Does your child turn the pages of a book by himself?”. logit −0. logit +0. or pen. item 3. “When a toy is in front of your baby. item 6. item 7.8. “When a loud noise occurs. item 26.4. logit +0. “Without resting her arm or hand on the table. logit 0. ask him to make a line like yours. item 42. “Does your child eat with a fork?”. item 43. logit 0. logit +0. one in each hand. logit +0. does she look in the direction of the voice?”.5. logit −1. logit −1. ‘It's round. “If you copy the sounds your baby makes. “Does your baby move around by walking. such as large climbing blocks?”. Note: Example of items: logit +1. does your baby repeat the same sounds back to you?”. “After your child watches you draw a line from the top of the paper to the bottom with a pencil.1. and hold onto them for about 1 minute?”. logit −0. “Does your child climb the rungs of a ladder of a playground slide and slide down without help?” C. It's big’?”. putting it down again with little spilling?”. item 38. “Does your child tell you at least two things about common objects? For example.0. logit +0. logit −1. A. Does your child copy you by drawing a single line in a vertical direction?” D. “Does your child put on a coat. I throw it. “Does your baby say four or more words in addition to ‘Mama’ and ‘Dada’?”. “Does your child know the names of numbers? 3–1–2”.4.6.4. jacket. item 21. Do not let your child trace your line. “Does your child put things away where they belong? For example. Construct Map for the fine motor common scale of the ASQ-BR. and dishes in the kitchen?” E. crayon. does your baby turn to see where the sound came from?” B. “Does your child color mostly within the lines in a coloring book or within the lines of a 5-centimeter circle that you draw?”. item 17. Construct Map for the communication common scale of the ASQ-BR. item 12. / Early Human Development 89 (2013) 561–576 573 allowed the item to fit in the common scale when the item was compared The Construct Map is a tool that is used to overlook the difficul- with a larger set of items. A. item 20. “Does your child use comparison words. logit −0. item 16.5. item 8. “Does your child drink from a cup or glass. “Does your baby pick up two small toys.5. does your baby take several steps forward?”. does your child copy you by drawing a single line on the paper in any direction?”. rather than by crawling on his hands and knees?”.8. item 4. if you say to your child. logit 0. “When you hold one hand just to balance her.’ does she say something like.9. “Does your baby feed himself a cookie or fruit?”. logit −0.’ or ‘shorter’?”. logit −1. “Does your baby stack three small blocks or toys on top of each other by herself?”. Note: Example of items: logit +2. Filgueiras et al. Note: Example of items: logit +0.3. item 17.3.

e. The procedure revealed an average theta for boys and ing adequately. It only means that both children are develop. This child does not necessarily have the same development as a calibrating and studying items across both age intervals and persons child with the same score. with the exception of the The distributions of persons' abilities were skewed. it should be used only for screening allowed to use forks for safety reasons. Conclusions difference between genders remained. Filgueiras et al. scores than girls.e. It is expected that children whose items are marked found the same gender differences using the Norwegian [29] and by the caregivers already present complete or nearly complete develop. the ASQ-3 [13]. In Rio de Janeiro's childcare centers. motor skills [57–60]. including Brazilian preschool children [61]. early development. Dutch [17] versions of the ASQ. The The Construct Map provided a glimpse into the persons' distributions fact that the ASQ-BR was able to detect gender differences in develop- for a screening test.574 A.. A high concentration of “Yes” mental milestones at early ages confers psychometric value to these responses was expected. The results of the ANOVA corroborated . compromised once teachers and professionals involved in caring for The present work also addressed the issue of gender differences in the children did not feel sufficiently secure to test the children. The only exception was item 23 from the personal–social not simply one [56]. Thus. range. you use it more carefully. higher than in boys in all of the domains. The and corresponds to an already discussed problem (i. well spread. The purpose of this study was to analyze whether domain. The original proposition of the ASQ-3 [13] instruments have also indicated that preschool girls from different was to determine whether the child has adequate development for cultural regions have better developmental scores outside gross her age and not to separate persons by their abilities. We compared differences across age intervals. Overall. This item was clearly the most difficult item in the ASQ-BR the ASQ-BR is an adequate instrument in the educational context. “never” was the most marked answer. 2 (continued). The item evaluation was likely purposes and not to separate children. ASQ-BR scores in girls tended to be Thus. Different developmental assessment ment of the measured trait. a total score of 60 (i. / Early Human Development 89 (2013) 561–576 Fig. such as ASQ-BR.. The ASQ-BR. gross motor skills domain. were possible. score) only indicates that the child is within a normal developmental Once the domain scales were linked-equated in a common scale. As expected. the maximum questionnaires. thus. The interactions revealed by about any evaluated person should be made using several instruments. The importance of this fact is that once you understand girls. children are not trate at the top of the scale. the ANOVA were interesting. “feed himself Construct Map allowed us to understand that persons tend to concen- with a fork”). is essentially a screen. in which boys tended to present higher such as the original instrument. These results are consistent with other studies that ing instrument. the the ASQ-BR as a screening tool.

[9] Maternal Child Health Bureau. Developmental screening.312:1900–2. Shevell MI.333: motor skills [58. . For example. The [12] Bricker DD. Cultural and linguistic ad- aptation allowed caregivers to complete the questionnaires without major difficulty. ferent genders explore their world through experiences with parents. we may conclude that the ASQ-BR is a brief and low-cost method that can be employed cost-effectively to gather information from a large number of children enrolled in public daycare centers during a short period of time. ably go through paths with few intersections since preschool [64].65]. Assessing the development of young children in child care: a survey of formal assessment practices in one state. Zeanah CH. [13] Squires J. Mirela Carvalho.34:455–65. Oskoui M. Twonbly E. Neurol 2005. children. Brookes Publishing. These differences are [6] Heckman JJ.63]. Sheila Najberg. ASQ-BR results might help improve the quality of public child daycare centers and help the decision-making process with regard to political and economic govern- mental investments. Fregni F.59. Finally.pdf. might limit the generalizability of the present findings. did not appear until 14 months of age and tended to disappear around [2] Knudsen E. Acknowledgments The present work was only possible because of the invaluable sup- port of many people. Differences in the gross motor domain. have different pathways during development. Marcia Gil. Curr Pediatr 2003. 2009. Baltimore: Paul H. The statisti- care needs: chartbook 2005–2006. With regard to the problem [8] Gleason MM. 2011]. 2nd ed. Investing in early human develop- ment: timing and economic efficiency. Daniel Santos. test–retest analysis would be important to evaluate the consistency of ASQ-BR scores over time.62].20:4–21. they were drawn from child daycare centers with low-income families who live in Rio de Janeiro. Infant Mental Health J 2010.. vention and education. Science 2011. ANOVA results in the present study are important to understand how child-monitoring system. Tremblay RE.16:1412–25. With regard to gender. Heckman JJ. [7] Allen SF. and other children [60. motor skills are more related to movement skills than physical fitness.31:335–57. ASQ-BR results might also help the child screening process for possible developmental delays as early as possible so that children can be referred to trained examiners for further evaluation and different forms of interventions can be planned in advance. which probably restrained the cultural diversity of our sample. Dickstein S. We are deeply grateful to Ricardo Paes e Barros. social skills emerge in Brazilian children around 20 months of age. [3] Pascual-Leone A. Fig. Econ Hum Biol 2009. Rosana Mendonça.7:1–6. The national survey of children with special health continuously grows in both genders in two parallel paths). therefore. / Early Human Development 89 (2013) 561–576 575 Ii is important to note that the present study had several limita- tions that can be explored in future research. This may pro. fine motor skills and problem solving skills also [5] Doyle O. 2 (continued). This likely occurred because gender differences in gross Neurosci 2004. family. Potter L. Acknowledging these limitations. no interaction was found in the ANOVA. The plastic human brain cortex. Baltimore: Paul H. Sensitive periods in the development of the brain and behavior. the sensitivity and specificity of each of the ASQ-BR questionnaires could be evaluated with the presence of a clinical sample of children with developmental disabilities. [accessed August 4. mental health assessment: development and preliminary validity of the Early vide evidence of parallel development of the skill [63] (i. Developmental neuroscience: implications for early childhood inter- age-dependent. Brookes with regard to gender differences. 3rd ed. Harmon CP. Eduardo Pádua. Skill formation and the economics of investing in disadvantaged well-known in the literature and likely are related to the way that dif. and preschoolers parent-completed child monitoring system.e.28:377–401. for example. and Jane Squires. 1999. References the t-test results and indicated that the gender differences were [1] Hannon P.13:58–63. Bricker DD. to the ASQ-BR results. A. Merabet L. influenced by gender. This is an important issue because the geographically narrow region. Science 2006. The convergent validity of the ASQ-BR could also have been investigated if the study used another early developmental screening instrument. or the Bayley Infant Neurodevelopment Screener which have been validated for the Brazilian population [61. Filgueiras et al. Recognizing young children in need of solving domain. such as the Denver Developmental Screening Test. http://www. Publishing. not simply stimulating minds. although the present study employed an exceptionally large number of subjects. Majnemer A. According to our empirical findings with regard 982–3.uconnucedd. Moreover. Conflict of interest None. Protecting brains. the children's experiences during development impact gross [4] Shonkoff JP. toddlers. Ages & Stages Questionnaires: a we deal with the development of infants. New York: Psychological Corpora- tion.org/actearlyct/PDFs/ cal difference remained the same across age intervals but was not National. Squires J. the ability Childhood Screening Assessment. J Cogn 4 years of age. [11] Rydz D. school. differences in personal– [10] Bayley N. 1969. Classical Test Theory and RS analysis from the Rasch model family indicated that the ASQ-BR has suitable psychometric properties for assessing early development. Ages and Stages Questionnaires: a parent-completed. Bayley Scales of infant development. J Child The development of social skills in boys and girls is different and prob. Amedi A. Early Childhood Educ J 2007. Annu Rev Neurosci 2005.

Infant winsteps.93:1525–9. 2001. Swerdlik ME. A Rasch model for partial credit scoring.96:748–52. development and evaluation of the intervention model for the Florida Infant Men. tal Health Pilot Program. [57] Hyde JS. analysis: a tutorial on parallel analysis. Mahwah. [34] Instituto Brasileiro de Geografia e Estatística.43:561–73. Puente AE. children based on effect size calculated by meta-analysis. Developing preschool surveillance tools for adaptive functioning: les. Comparison of systematic develop. 617–23. Momento da [51] Lee OK. Greisen G.28:130–50. [23] Macy M. 2005. San Antonio. Gender influences on preschool children's social 2006.98:1494–9.34(1):126–35. Psychometrika and age. Gender differences of [30] Plomgaard AM. [48] Linacre JM.52:195–206. Indian Pediatrics 2012. / Early Human Development 89 (2013) 561–576 [14] Handal AJ. Brookes Publishing. Psychological testing and assessment: an intro- on neurobehavioral development in infants and young children in a flower. Support [41] O'Connor BP. 5th ed. Fisher R. p. Infant 149–74. The evidence behind developmental screening instruments. 2010. Dixon G. Hansen BM. ing. [58] Lung FW. movement skills and physical fitness among 3. Squires J. Scarpello V. Brookes Publishing. editor. Third Edition (ASQ-3): user's guide. Pesquisa Nacional por Amostra de [63] Halpern DF. Behav Res Methods Instrum screener. Rasch Meas Trans 2000. Proceedings of the fourth Berkeley symposium on mathematical Pediatrics 2009. Serra AMPH. Pan X. Squires J. SPSS and SAS programs for determining the number of components for the global feasibility of the Ages and Stages Questionnaire as developmental using parallel analysis and Velicer's MAP test. Acta Paediatr sional psychology. human sciences. Boris NW. Adaptation of the 36-month Ages and meta-analysis. et al. Lederman JC.com. Characteristics. Comput 2000. Katz L. [21] Osofsky JD. Mitchell EA.21:138–53. Encyclopedia of bio. Adams S. de Meer G.52(10):1091–102. Joyce DP.85:443–7. [49] Bond TG. Berkeley: University of California Press. Aoyagi O. Mental Health J 2007. and developmental delay in 287 consecutive infants attending a plagiocephaly [40] Hayton JC. Smith DK. Butcher PR. editors. Effect of community of residence [37] Cohen RJ. Robaey P. de Carvalho-Leite JC. 1992. Comparing “Partial Credit Models” (PCM) and “Rating Scale Models” [22] Beeber LS. accessed Febru. J Dev Behav Pediatr 2011. Zubrick SR. [20] O'Leary C. In: language delay in 2-year-old children: the importance of dose and timing on risk. Harlow SD. Wright BD. Anderson RE. tal delay in children presenting for primary care. Shu BC. Prof Psychol Res Pract 2000. [27] Bonin M.104:53–69.mrw. [39] Velicer WF. London: Routledge. Howell DC. Feng JY. 1992. Mathematics and Reading Test Equating.51(2):154–9. (RSM). Allen DG.41:321–7.95:1488–94. pnad2006/pnad_sintese_2006.br/home/estatistica/populacao/trabalhoerendimento/ Psychol 2011. Ages and stages questionnaire as a screen. 2012]. Psychological test usage: implications in profes- population sample: a comparison with US normative data.28:259–80. ing tool for developmental delay in Indian children. Kopelman BI. Marshall T. Campos ARS. Baltimore: Paul H. Infant Behav Dev ary 28. [53] Branson D. Black WC. CA: Mayfield Publishing. 1961.interscience. Rasch Measurement Transactions 1993. 2006. TX: Paul H.120:381–9.31(2):141–54. 2003. Nickel R. Chichester: Wiley.32:396–402.14:368–79. Cross-cultural adaptation of a pre-school screening [61] Drachler ML. Berthelsen D. Applying the Rasch model: fundamental measurement in the [24] Thomas SA. Encyclopedia of statistics in behavioural science. Janson H. Early Hum Dev 2009. A user's guide to Winsteps: Rasch-model computer programs. San Antonio. Reno: University of Nevada. Filgueiras et al.16:297–334. [25] Graça PRM.163:197–209. Lozoff B.pdf. Colton T. Lacroix V. Chen PF. TX: Paul H. [17] Kerstjens JM.49(6): [62] Crombie G. [26] Ages & Stages Questionnaires. Taylor CL. 2009. Bastin GL. Fox CM. questionnaire. A rating scale formulation for ordered response categories. Chichester: John Wiley & 499–511. growing region of Ecuador.98:260–82. Nathan JS. screener in Brazilian preterm children under risk conditions. [65] Guedes DZ. Infant Mental Health J 2007. Stages Questionnaires. [42] Little TD. Impact of implementing developmental [38] Kline P. Gender differences across age in motor performance: a [31] Tsai HLA. Stages Questionnaire in Taiwan: results from a preliminary study.576 A. Marks K. 2000. [55] Wright BD. An easy guide to factor analysis. Acta Paediatr 2007. Squires J. 1979. Harden BJ. p. [18] Limbos MM. Measuring developmental deficit in children children's developmental trajectory from 6 to 60 months in the Taiwan Birth born at gestational age less than 26 weeks using a parent-completed developmental Cohort Pilot Study. On general laws and the meaning of measurement in psychology. statistics. 2012]. 298–9. Ratliff-Schaub K. A validation study of the Norwegian version of the Ages and Bull 1988. [50] Wright BD. Org Res Methods 2004. Neyman J.17(2):116–20. http://www. Psychometrika 1976. 2007. Acta Paediatr 2009. Yovanoff P. Bower C. social– de développement: Évaluation de l'enfant par les parents. Psychometrika 1978. Equitable Test Equating.gov. Best test design. Bos AF. Pediatrics 2007. Eur J Paediatr Neurol 2010. Lopes SCG. NJ: Lawrence Erlbaum. Multivariate data analysis. Stone MH. [15] Hix-Small H. [46] Linacre JM.7(2): ClinPediatr (Phila) 2012. Cronbach's alpha. problem-solving strategies. Psychol Bull 1985. www.115:128–33. J Early Intervent [60] Walker S. qualidades psicométricas do ASQ −2 dos 30 aos 60 meses. Determining the number of components from the matrix of partial [16] Hutchison BL. 2012]. [59] Thomas JR. Prenatal alcohol exposure and [44] Rasch G. [19] Msall ME. Young Child 2012. head shape measurements correlations. Breilh J. Factor analysis: multiple groups. Gender differences in verbal ability: a meta-analysis. The Early Promotion and Intervention Research Consortium (E-PIRC): five ap. screening at 12 and 24 months in a pediatric practice. assessment abilities. Factor retention decisions in exploratory factor clinic. BINS validation — Bayley neurodevelopmental com/emrw/9780470011812/eob/article/b2a13018/current/html. Ramji S. emotional. Lin SJ. duction to tests and measurement. School Health 2009. Mountain View. 9–23. Twombly E. A continuous-scale measure of instrument: comparison of Korean and US populations. in cross-cultural research. Bricker DD. Acta Paediatr 2006. Parent-completed developmental screening in a Norwegian [56] Camara WJ.123:547–54. 2nd ed. Int J Psychol Res 2010. 2010. Chazan-Cohen R. 2009. Relationship between gender difference in motor performance [35] Cronbach LJ. Pratt C. Paediatr Perinat Epidemiol [33] Juneja M. Am Domicílios. J Intellect Disabil Res child development for population-based epidemiological surveys: a preliminary 2008. Ages and Stages Questionnaires.3(1):111–21. Vandaele S. 321–44. McClellan MM.wiley. ten Vergert EMJ. Cotton W.25:19–61.5: [36] Dukes KA. 2004. Coefficient alpha and the internal structure of tests. 2011. J Genet Psychol 2002. Tatham RL. Mohanty M. Spanish ver. Jain R.32:100–6. [43] Milfont TL.28:213–25. [32] Heo KH. [47] Masters GN. statistics and probability. In: Armitage P. Slegers DW. Teixeira MSC. Environ Health Perspect 2007. Brookes Publishing. [28] Janson H. Sex differences in intelligence: implications for education.7:191–205. study using Item Response Theory for the Denver Test.32: editors. Questionnaire sur les étapes [54] Squires J. French KE. et al. Gender differences: implications for social skills assessment and train- 457–61. Testing measurement invariance across groups: applications sons for neuro-oncology. 2007. Kronenberg M.14(3):768. Linn MC. The [45] Andrich D. Stewart AW. mental surveillance with standardized developmental screening in primary care.to 6-year-old Japanese 1951. Squires J. [64] Ikeda T. New York: Prentice Hall. 2nd ed. Irving K. Psychometrika 1982. Psychol [29] Richter J.ibge.16:177–96. 2005 [http://www. In: Everitt BS. Chicago: Mesa Press. Chiang TL. Annual meeting of avaliação na intervenção precoce: o envolvimento da família: estudo das the American Educational Research Association: San Francisco. Res Dev Disabil 2011.47: proaches to improving infant/toddler mental health in Early Head Start. . [accessed February 28. Hammer JH. 1994. Comparison of the ASQ and PEDS in screening for developmen. [accessed February 28. Squires J. Primi R. Rev Bras Educ Esp [52] Hair JF. Sons. J Clin Child Psychol 1988. Reijneveld SA. Scott-Heller S. The effects of using the ASQ on childcare providers' developmental sion.