Pollsters attribute more accurate runoff surveys to better timing
Paper Edition | Page: 9
Recheck: Staffers of the Menteng election committee in Central Jakarta recapitulate the tallies counted in each polling station in the subdistrict on Friday. Quick counts on Thursday showed a slim lead by Joko “Jokowi” Widodo. (JP/P.J. Leo)
Pollsters say that their voter surveys for the gubernatorial runoff on Thursday were held closer to election day and were thus more accurate than polls held in advance of the first round of voting in July.
Most polls conducted before the runoff said that Surakarta Mayor Joko “Jokowi” Widodo would beat incumbent Governor Fauzi Bowo by a small margin in the runoff, which accorded with quick counts on Thursday that said Jokowi topped Fauzi by about 10 percent.
Agus Herta Sumarto from the Political Research Institute said on Friday that the closer a survey was done to election day, the more accurate it would be.
“Closer to the runoff, voters have generally made up their minds and are less likely to change their decisions,” Agus said.
According to a survey conducted by the Center for Political Studies (Puskapol) at the University of Indonesia (UI) between Aug. 27 and Sept. 2, 69.82 percent of 600 respondents said that they had already decided on who they would vote for in the runoff.
On the other hand, Puskapol UI’s previous survey, conducted more than a month before the first round of voting in July, said only 45.7 percent of 594 respondents had made up their minds.
According to Dirga Ardiansa, a researcher at Puskapol UI, the first survey was held too early to accurately reflect voter perceptions, among other things.
“Voters were faced with six candidates to choose from in the first round, as opposed to two in the runoff,” Dirga said. “It’s easier to get an answer when there are fewer options.”
Dirga also questioned the credibility of polls funded by the campaigns.
“There were indications that some of these pollsters manipulated their surveys by selecting respondents that fit with the characteristics of a person who would vote for a certain candidate,” he said.
The Indonesian Survey Institute (LSI) also conducted one of the runoff polls that was more accurate than its first-round poll.
LSI executive director Kuskridho Ambardhi said that the institute conducted an unpublished poll in the first round that showed Fauzi leading Jokowi, albeit by a small percentage.
Meanwhile, a survey conducted between Sept. 2 and Sept. 7 by the LSI said that 45.6 percent of 399 respondents would vote for Jokowi and running mate Basuki “Ahok” Tjahaja Purnama, while Fauzi and Nachrowi Ramli would receive 44.7 percent of the vote on Thursday.
Kuskridho said that the institute’s first-round survey was inaccurate because the LSI used multistage random sampling, selecting respondents from each of the city’s five municipalities and one regency.
Things were done differently for the runoff, Kuskridho said.
“For the runoff, we decided to use simple random sampling by directly choosing respondents from the voter lists,” he said. “This increased our accuracy.”
Kuskridho said that the LSI also increased its quality control from 20 percent in the first round to 100 percent for the runoff.
“We also did not substitute respondents who failed to be interviewed for the runoff survey to increase the accuracy of the survey, because doing so would defeat the purpose of random sampling,” he said.
Pollsters have been criticized for inaccurate surveys for the first round of voting in the gubernatorial election in July.
Most said Fauzi would finish with a plurality of the votes cast. Some even predicted that Fauzi would clinch reelection in the first round by netting more than 50 percent of the votes.
The reality, however, was different as Jokowi won the first round with 42.6 percent of the votes, compared to Fauzi with 34.05 percent. (han)