Abstract
This paper aims to investigate the effects of region, ethnic group and electoral system on the Democratic Progressive Party's (DPP) vote across elections by using aggregate panel data. Comparing the pool data model to the panel data fixed effect model based on Hausman and Taylor (1981), we find that the pool data model overestimates (underestimates) the effects of the governing counties (ethnic groups) on the DPP's votes. However, the panel data fixed effect model does not. Since the traditional fixed effect model cannot reveal estimated information on individual effect, we present the Hausman and Taylor model, which can estimate individual effect proper1y and specify the impact of region. We can conclude that the result of the Hausman and Taylor model that avoids estimation bias by specifying the individual region is not just superior to other model results, but prove to be a proper model for further studies on geographical correlation on voting behavior.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
