Multilevel Multinomial Logistic Regression Analysis of Contraceptive Prevalence in Dakahlia Governorate | ||||
| The Egyptian Statistical Journal | ||||
| Article 2, Volume 55, Issue 2, December 2011, Page 93-106 | ||||
| DOI: 10.21608/esju.2011.314313 | ||||
 View on SCiNiTO | ||||
| Abstract | ||||
| This study uses Multilevel Multivariate Logistic Regression Model as a suitable method for multilevel analysis of the data is collected from different levels of hierarchy. Thus, is necessary to elicit the inherent variations which occur for a particular phenomenon between those levels. The model was applied on 2008 Egypt Demographic and Health Survey (DHS) contraceptive multinomial multilevel stratified cluster data. The design of this study allows working with multilevel logistic regression models, including model conceptualization describing model, understanding of the structure of required multilevel data, estimating of the model via the statistical package MLwIN, and interpretating of the results The results reflect the existence of significant variations in the mean effects of the predictors in both cluster and division level, the effect for place of residence has been found considerably different in rural and urban areas across the divisions. The study has finally suggested integrating a strong awareness program that targets currently married women in Dakahlia in those divisions and clusters where the particular determinant of contraceptive use has been found to be less effective. | ||||
| Keywords | ||||
| Multilevel Multivariate Multinomial Logistic Regression Model; Random Effects; Fixed Effects; Hierarchy Data; Penalized Quasi; Likelihood (PQL) | ||||
|
Statistics Article View: 29 |
||||
