Abstract

Conventional multivariable analyses may not always be the ideal method for estimating exposure effects in observational studies. Where there are large differences in the distribution of covariates between expose groups, adjusting with conventional multivariable techniques may not adequately balance the groups, and the remaining bias may limit valid causal inference. The objective of this research is to compare the result of convensional multiariate analysis versus propensity score matching analysis in case study of infant immunization using secondary data of ASUH KAP2 2003. Model will be compared without interaction variable. The results show that the OR from logistic regression (0,99) differs to propensity score matching (0,96). Propensity score matching is successfulin matching 574 subjects (68,27%). It is recommended to evaluate risk factor effect using PSM model, but to use logistic regression analysis for determinat factor analysis with many independent variables because the variables have the same position.