In this study, the hypothesis testing of geographically weighted bivariate logistic regression (GWBLR) procedure is proposed. The GWBLR model is a bivariate logistic regression (BLR) model which all of the regression parameters depend on the geographical location in the study area. The geographical location is expressed as a point coordinate in two-dimensional geographic space (longitude and latitude). The response variable of BLR model is constructed from the (2x2) contingency table and follows the multinomial distribution.

The purpose of this study is hypothesis testing of the GWBLR parameters which includes a test of similarity between BLR and GWBLR model (goodness of fit test), simultaneous test, and partial test. The Vuong test method is applied to the goodness of fit test. The simultaneous test and partial test are carried out using the likelihood ratio test method and the Wald test method.

The test statistic of the goodness of fit test and the partial test was asymptotically normally distributed, whereas the test statistic of the simultaneous test was asymptotically chi-square distributed.