"I thought I had won, but it felt like I had been scammed."
A staffer at one Democratic Party of Korea campaign, which exit polls by the three major broadcasters had shown leading the People Power Party candidate by a wide margin, said this on the 4th after the final vote count ended in defeat. This year as well, exit polls and actual results diverged in key battleground races, reviving the argument that exit polls are useless.
In exit polls jointly conducted the previous day by the three terrestrial broadcasters KBS, MBC, and SBS along with the Korea Broadcasters Association, Democratic Party candidate Jung Won-oh registered 51.4% and People Power Party candidate Oh Se-hoon 46% in the Seoul mayoral race, putting Jung ahead by 5.4 percentage points. However, in the final results based on a 99.54% count as of 3 p.m. on the 4th, Oh secured the win with 49.15%, ahead of Jung (48.13%).
In the South Gyeongsang governor race, the exit poll and final result also differed greatly. The exit poll predicted Democratic Party candidate Kim Kyoung-soo would lead with 54.3% over People Power Party candidate Park Wan-su (45.7%), but in the actual count Park won with 51.28%.
In the Daegu mayoral race, the exit poll showed a tight contest with a 0.8 percentage point margin, but in the end People Power Party candidate Choo Kyung-ho won 53.92%, outpacing Democratic Party candidate Kim Boo-kyum (45.05%) by a relatively wide margin.
The by-elections also failed to match the exit poll predictions. In Pyeongtaek-B, Gyeonggi Province, Cho Kuk of the Rebuilding Korea Party (31.1%), Yoo Eui-dong of the People Power Party (30.6%), and Kim Yong-nam of the Democratic Party (30.3%) showed an extremely close race, with Cho predicted to lead, but the actual winner was Yoo. In Busan Buk-A, the survey showed Democratic Party candidate Ha Jung-woo (42.6%) leading independent candidate Han Dong-hoon (41.6%), but in reality Han succeeded in entering the National Assembly.
Under the Public Official Election Act, exit polls are conducted only on the day of main voting, and their failure to sufficiently reflect the sentiment of voters who participated in early voting is cited as the main cause of the error. To correct for early voting, a telephone survey of about 28,500 people was also conducted in parallel, but it was assessed as having revealed limitations.
Political commentator Kim Sang-il said, "In this local election, the early voting rate reached 23.51%, the highest ever for a local election, but the exit polls did not sufficiently adjust for the early voting sentiment." He added, "The approach is to increase the scale of the survey to raise accuracy, but ultimately there are also issues of cost and structure."
