Pustaka Ilmiah Universitas Padjadjaran

Abstrak

A Spatial Autoregressive Model Of Diarrhea Prevalence In Java Island Districts

Diarrhea diseases remain ones of the most challenging public health problems in developing countries. Information on contribution of diarrheal risk factors become an important input for development and evaluation the policies and activities for combating diarrhea. In this paper we estimate the contribution of diarrhea risk factors by means of a spatial autoregression (SAR) model to account for spatial dependence among diarrhea occurrences in Java Island districts. We find that SAR improves the standard regression (SR) diarrhea model. It provides a better model fit and lower standard error estimates than SR does. It also corrects parameter estimates bias due to ignoring diarrhea contagiousness. From the estimates we infer that diarrhea prevalence rate in neighboring districts is the most dominant risk factor i.e. 1% increase of average diarrhea prevalence in neighboring districts will increase 0.36% diarrhea prevalence in a district. The other important risk factors are house wife literacy, safe water supply (non-pipe and pipe water), and private toilet. 1% increase in each risk factor rate, respectively, will decrease 0.098%, 0.023%, 0.020% and 0.010% diarrhea prevalence. Nevertheless, we suspect that the spatial effect to be overestimated, while the other factors to be under estimated. Further improvement is required for reducing omitted variable bias caused by diarrhea risk factors which are not in the model.