The bivariate probit regression model is a probit regression model consisting of two response variables with errors between the two variables correlate each other. The correlation between the two response variables can occur as a result of the presence of endogeneity, a condition in which a response variable becomes an exogenous variable in another response variable. Besides, the important issue that cannot be underestimated is undetectable nonlinear relationships between response variables and predictors, especially discrete or continuous predictor variables. The bivariate probit regression that does not ignore endogeneity cannot detect the nonlinear relationships between response variables and predictors, so one of the regression models that can overcome the problem is bivariate probit regression model with a semiparametric approach. The first step in semiparametric bivariate probit modeling is testing the hypothesis of exogeneity to determine whether there is a case of endogeneity or not. The exogenous test used in this study is the Lagrange Multiplier (LM) and Likelihood Ratio (LR) test. The data used in this study consisted of two binary categorical response variables, they are parity status of the mother and basic immunization giving to infants in North Kalimantan Province in 2017. The results of the exogenous test using the LM test and LR test stated that there was a significant correlation between response variables. The AIC value of the semiparametric bivariate probit model is 1301.602, while the bivariate probit model produces AIC of 1316.789, so it can be concluded that the semiparametric bivariate probit model provides better modeling results than the bivariate probit model.
|IOP Conference Series: Materials Science and Engineering
|Published - 1 Jul 2019
|9th Annual Basic Science International Conference 2019, BaSIC 2019 - Malang, Indonesia
Duration: 20 Mar 2019 → 21 Mar 2019