Desenvolvimento de um banco de itens para avaliar o transtorno depressivo maior

• Ano de defesa: 2017
• Autor(a) principal: Sales, Hemerson Fillipy Silva
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal da Paraíba; Brasil; Psicologia; Programa de Pós-Graduação em Neurociência Cognitiva e Comportamento; UFPB
• Palavras-chave em Português: Transtorno depressivo maior; Banco de itens; Unidimensionalidade; Calibração; Major depressive disorder; Bank of items; Unidimensionality; Calibration; CNPQ::CIENCIAS HUMANAS::PSICOLOGIA
• Link de acesso: https://repositorio.ufpb.br/jspui/handle/123456789/12003

The aim of this study was to evaluate the Major Depressive Disorder (MDD), focusing on the severity of the disease. For this, 150 items were elaborated, based on the DSM-V diagnoses for TDM and BDI items. The study was carried out in two phases: theoretical analysis of the items and pre-test. The theoretical analysis comprised two stages: an analysis of judges, which counted on the collaboration of three experts and a semantic analysis, in which seven subjects were divided into two groups. For the pre-test use a non-probabilistic sample of 501 participants from the cities of Parnaíba-PI and João Pessoa- PB. The majority of the female sex (61.9%), unmarried (58.5%) and aged between 18 and 76 years (M = 30.09 / SD = 12.01). For the data collection phase, the items were divided into two sets (A and B). About the results of the theoretical analysis, 14 items were eliminated, 35 were modified by suggestion of the tests and four at the suggestion of the participants. The results of the preliminary analysis were divided into two parts, the first to verify the unidimensionality and local independence of the sets of items and the second to calibrate them. To verify the Local Independence assumption, factorial analyzes and the Hull method were performed. These analyzes showed that the two sets (A and B) presented only one factor with excellent alpha values (0.93 and 0.96, respectively), thus corroborating the unidimensionality of the sets and consequently assuring local independence. The calibration of the items was performed using the Gradual Response Model (GRM). As the first estimates showed a need to exclude some items for presentation of low discrimination (a), besides showing that others would need to be recategorized. After recategorization, a new evaluation was carried out in which all the items, in short, were adequate. Finally, it is hoped that the proposed goal has been met. The item bank presented good psychometric concepts, being calibrated and ready to be used to prove Computerized Adaptive Tests (CATs).