In this study, we investigate the data of the exit poll for the 19th general election of 2012 to analyze the causes and characteristics of exit poll forecasting error. The empirical studies show that systematic negative bias of candidates affiliated t...
In this study, we investigate the data of the exit poll for the 19th general election of 2012 to analyze the causes and characteristics of exit poll forecasting error. The empirical studies show that systematic negative bias of candidates affiliated to Saenuri Party is serious in the specific regions. To explore the main causes and effects of the forecasting error, the total survey error is divided by sampling error of the polling sites and non-sampling error which comes from selecting voter and collecting responses from sampled voters in each polling sites. And we consider relationships among several factors, which are anticipated as main causes of error, and non-sampling error as well as sampling error. As a result of correlation analysis, we can figure out that the nonsampling error of the exit poll mainly come from non-response. Also we consider the house effect in exit poll.