Validando estágios de desenvolvimento do raciocínio indutivo

• Ano de defesa: 2012
• Autor(a) principal: Hudson Fernandes Golino
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Estágios; Raciocínio indutivo; Desenvolvimento; Validade estrutural; Raciocínio (Psicologia); Psicologia do desenvolvimento; Psicologia
• Link de acesso: http://hdl.handle.net/1843/BUBD-9EGM5J

The developmental stages validity has been focus of controversy in the literature about developmental psychology. However, in the past twenty years a serie of quantitative methodologies have been developed or applied to empirically identify discontinuities, both on people performance and items and tasks difficulties. The present dissertation investigates the validity of inductive reasoning developmental stages, throught the construction and validation of the Inductive Reasoning Developmental Test (IRDT). It will be presented in two parts, representing two papers. The first paper investigates if the IRDTs items measures six developmental stages, forming six different and spaced clusters, in two samples, being one composed by167 people (50.3% men) with ages varying from 6 to 58 years (M = 18.90, SD = 9.70), and the other composed by 188 people (57.7% woman) with ages varying from 6 to 65 years (M = 21.45, SD = 14.31). The result shows an adequate data fit to the Rasch model (infit mean = 0.94, SD = 0.22), six clear item clusters with gaps between them, with adjacent clusters presenting statistically significant differences. The second paper investigates the structural validity of the IRDT 3rd version, constructed to overcome some limitations founded in the first two versions. Three quantitative methodologies are used: 1) Confirmatory Factor Analysis; 2) Dichomotomous Rasch Model; and 3) Latent Class Analysis. The sample was composed by 1,459 people (52.5% woman) aging from 5 to 86 anos (M = 15.75, SD = 12.21). The result shows a factorial structure with seven first-order latent variables (one for each stage) and a second-order geral factor [2 (61) = 8832.594, p = .000, CFI = .96, RMSEA = .059]. The 56 items presented adequate fit to the Rasch model (infit mean = 0.96; SD = 0.17), with a high item reliability (1.00) and a moderately high person reliability (0.82). The evidences point to a seven latent class model (AIC: 263.380; BIC: 303.887; Loglik: -111.690). Both studies show that is possible to empirically identify developmental stages of reasoning applying specific quantitative methodologies. The evidences point to the validity of the IRDT items to assess developmental stages of inductive reasoning.