Modelo de regressão logística ordinal em dados categóricos na área de ergonomia experimental

• Ano de defesa: 2009
• Autor(a) principal: Montenegro, Santhiago Guedes
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal da Paraí­ba; BR; Engenharia de Produção; Programa de Pós Graduação em Engenharia de Produção; UFPB
• Palavras-chave em Português: Ergonomia; Regressão Logística Multinomial; Modelagem de Dados; Índice de Capacidade para o Trabalho; Ergonomics; Multinomial Logistic Regression; Data Modeling; Work Ability Index; CNPQ::ENGENHARIAS::ENGENHARIA DE PRODUCAO
• Link de acesso: https://repositorio.ufpb.br/jspui/handle/tede/5273

On analysis performed at experimental ergonomics, still is rare the use of Ordinal and Nominal Multinomial Logistic Regression, having been employed their simplification, Binary Logistic Regression, even on cases where Dependent Variable (DV) have more of two categories. To make the DV becomes binary leads to damages at data analysis, caused due lose of information by category agglutination and ordination disrespect. An analysis using Ordinal Multinomial Logistic Regression was performed on a data set containing a categorical DV, the Work Ability Index (WAI) of Nurses working on João Pessoa city Public Hospital Intensive Care Unit (ICU), and as Independent Variable (ID) thermal comfort variables, environmental variables, personal variables, and work organization variables. Through this analysis, was found out risk factors that lead to increase the probability of the WAI falls on an inferior category. The DV and used data set features allows to conclude that the Ordinal Multinomial Logistic Regression use made possible a more accurate result and analysis.