Made available in DSpace on 2016-03-15T19:40:16Z (GMT). No. of bitstreams: 1 Daniel de Sousa Filho.pdf: 797124 bytes, checksum: b6136c0e2a983cf0eaad975c2f8445b7 (MD5) Previous issue date: 2014-02-06 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Background: Autism Spectrum Disorders (ASD) are developmental disorders characterized by qualitative impairment in reciprocal social interaction, language and communication and restricted, repetitive, and stereotyped patterns of behavior, interests, and activities and they are defined based on standardized clinical criteria as such those from international manuals as ICD and DSM, or based on clinical tools, mainly the semi-structured interviews, such as the Autism Diagnostic Interview- Revised (ADI-R), a recognized gold standard diagnostic tool for ASD. This interview was recently translated into brazilian Portuguese, and it was preliminarily validated for this language. However, this process has been occurred with a small sample in a specific state of Brazil. Besides, other validity evidences, as convergent validity were not explored in this study. Aim: The aim of this study was to investigate the Convergent Validity between the Brazilian version of the ADI-R and a screening tool Autism Behavior Checklist (ABC). Method: For this study, 20 parents or caregivers of children and teens diagnosed with ASD were recruited from the Presbyterian Mackenzie University ASD Clinic, and from the Federal of São Paulo State University Social Cognition Clinic. The children were submitted to a clinical multidisciplinary evaluation, and neuropsychological evaluation that also consisted on application of the ABC. Another researcher, a Child Psychiatrist with expertise on ASD was previously trained and allowed for conducting and codificating the ADI-R. Results: The age of the probands has ranged between 6 and 19 years old, 90% of them were male. The correlations between both the tools, using the Pearson coefficient were mostly positive, and they have ranged between average to high with statistical significance., considering the 3 mains domains of the ADI-R and the five main domains of the ABC. Conclusions: Evidence of convergent validity was found when comparing ADI-R results with ABC/ICA. / Introdução: Os Transtornos do Espectro do Autismo (TEA) são transtornos do desenvolvimento caracterizados por prejuízos qualitativos na interação social recíproca, comunicação e linguagem e pela presença de padrões de comportamentos estereotipados, restritos e repetitivos e sua caracterização se faz a partir de critérios clínicos definidos e padronizados por classificações internacionais como as da CID ou DSM ou por instrumentos, principalmente as entrevistas, como a Autism Diagnostic Interview-Revised (ADI-R), instrumento considerado pela literatura padrão-ouro para diagnóstico de TEA e que recentemente foi traduzido e preliminarmente validado para o português brasileiro. Entretanto, tal processo ocorreu com uma amostra reduzida e numa região específica do Brasil, assim como outras evidências de validade do instrumento tais como validade convergente não foram exploradas. Objetivo: Investigar evidências de validade de convergente entre a versão brasileira da entrevista diagnóstica ADI-R e os instrumento de triagem Inventário de Comportamentos Autísticos (ICA). Método: Foram selecionados 20 pais e/ou cuidadores de 20 sujeitos a partir da Clínica de TEA da Universidade Presbiteriana Mackenzie e da Universidade Federal de São Paulo. Os sujeitos foram submetidos à avaliação clínica multidisciplinar médica e neuropsicológica a qual incluía a aplicação do ICA. Um outro avaliador, psiquiatra da Infância e Adolescência com experiência em TEA e habilitado a aplicar a ADI-R foi responsável pela aplicação e codificação da entrevista. Resultados: A idade dos sujeitos variou de 6 a 19 anos, com média de 10,1. Desses, 90% eram do sexo masculino. Os coeficientes de correlação de Pearson entre os 3 domínios da ADI-R (comunicação, interação social e comportamentos restritos) entre si e destes com o e os do ICA (escrever todos), foram moderados e altos, positivos e a maioria com significância estatística. Conclusões: Evidências de Validade Convergente foram encontradas, comparando-se a ADI-R com o ABC/ICA.

The measurement of reliable and valid indicators of early child development is necessary for assessing phenomena and is useful in the monitoring of ongoing efforts to eradicate inequalities in the social determinants of health. There is an increasing awareness of the contextual, cultural, and developmental influences on constructs used in early child development (ECD) research. Using a measurement perspective, this dissertation examined the issue of measurement invariance and psychometric validity in early child development research. A construct violates the principle of invariance when two persons from different populations who are theoretically identical on the construct being measured have different scores on it. This dissertation consists of three journal-style manuscripts (published or under review) that were used as examples to address the importance of the issue of measurement invariance and psychometric validity in ECD research using data from two unique areas: autism and executive functioning. The three data sets were collected on pre-school children with parents and or teachers as informants and were chosen to represent different levels of data collection – clinical, community, and population. These data sets allowed for the examination of measurement invariance by type of informant, sex, and age of child. The results from the three studies illustrate the importance of assessing measurement invariance in ECD and whether or not the instruments examined can be used to assess sub-group differences with confidence. A lack of measurement invariance found for two of the studies, suggests that observed group differences in latent constructs could be attributed, in part, to measurement bias. More importantly, bias in the measurement of the constructs of severity of social impairment symptoms in autism, and executive functioning across groups could have an impact on services such as patient treatment. These biases could also influence public policy development, particularly when there may be an underlying need for a cross-group approach where belief systems may affect the meaning and structure of constructs. In summary, measurement invariance should be a prerequisite for making any meaningful comparisons across groups. A requirement of establishing measurement invariance should be included in the guidelines for comparative research studies as a necessary first step before an instrument is adopted for use.

The measurement of reliable and valid indicators of early child development is necessary for assessing phenomena and is useful in the monitoring of ongoing efforts to eradicate inequalities in the social determinants of health. There is an increasing awareness of the contextual, cultural, and developmental influences on constructs used in early child development (ECD) research. Using a measurement perspective, this dissertation examined the issue of measurement invariance and psychometric validity in early child development research. A construct violates the principle of invariance when two persons from different populations who are theoretically identical on the construct being measured have different scores on it. This dissertation consists of three journal-style manuscripts (published or under review) that were used as examples to address the importance of the issue of measurement invariance and psychometric validity in ECD research using data from two unique areas: autism and executive functioning. The three data sets were collected on pre-school children with parents and or teachers as informants and were chosen to represent different levels of data collection – clinical, community, and population. These data sets allowed for the examination of measurement invariance by type of informant, sex, and age of child. The results from the three studies illustrate the importance of assessing measurement invariance in ECD and whether or not the instruments examined can be used to assess sub-group differences with confidence. A lack of measurement invariance found for two of the studies, suggests that observed group differences in latent constructs could be attributed, in part, to measurement bias. More importantly, bias in the measurement of the constructs of severity of social impairment symptoms in autism, and executive functioning across groups could have an impact on services such as patient treatment. These biases could also influence public policy development, particularly when there may be an underlying need for a cross-group approach where belief systems may affect the meaning and structure of constructs. In summary, measurement invariance should be a prerequisite for making any meaningful comparisons across groups. A requirement of establishing measurement invariance should be included in the guidelines for comparative research studies as a necessary first step before an instrument is adopted for use.