Sections

Services

Information

UK United Kingdom

Quick counts, exit polls and surveys in Indonesia’s election

Exit polls and quick count results from seven pollsters in Indonesia show Jakarta governor Joko Widodo as the winner in Indonesia’s…

Pollsters wait until sample polling stations completed the vote count before releasing quick count results. Popularity surveys, exit polls and quick counts in Indonesia's election, CC BY-SA

Exit polls and quick count results from seven pollsters in Indonesia show Jakarta governor Joko Widodo as the winner in Indonesia’s presidential election.

According to the quick count results, which have proved accurate in the past, the candidate popularly known as Jokowi won the race by a margin of 3-5% from former military general Prabowo Subianto. Eleven pollsters have released quick counts. Prabowo also claims victory based on surveys released by four pollsters favourable to him.

What is the difference between exit polls and quick counts?

Dirga Ardiansa, a statistician and research manager at the Centre for Political Studies at the University of Indonesia, said that to predict the winner of presidential elections pollsters use three consecutive surveys: the popularity surveys, the exit poll and quick count. The quick count, based on sampling of the actual vote, is the most reliable.

What is…

Quick Count: “is a survey on the election results. Quick count samples are no longer individuals but the polling stations,” Dirga said.

The pollsters wait until their sample polling stations have counted the votes. The result of quick counts comes from the aggregate percentage of voters.

Margin of error: Margin of error is the amount of sampling error in the survey’s result. “In the context of the Indonesian election, if we take the total number of polling stations, which is 479,000, then taking 2,400 polling stations as samples will give you a 2% margin of error,” Dirga said. The more samples survey centres take, the smaller the margin of error and the more precise their data would be.

According to Dirga, of the 11 pollsters' published quick counts, Saiful Mujani Research Centre has the smallest margin of error. But for him, the most credible surveys come from the Kompas research and development team and RRI (Republic of Indonesia Radio) pollsters for their methodology and independence, he said.

“If we look at the four pollsters that show Prabowo leading, from the results they publish they still have many weaknesses,” he said. “We are not sure about their independence as well.”

He said the CSIS-Cyrus quick count is methodologically credible. “But we have to admit they are part of Jokowi’s campaign team, so there is a conflict of interest there.”

Exit Poll: “is a survey taken immediately after the respondents cast their votes”, Dirga said. The target respondents are people who have exited the poll booth and are asked who they voted for.

“This type of survey is relatively more accurate than popularity surveys because the respondent answered just after they have voted,” he said.

For example, the Centre for Strategic and International Studies with Cyrus had interviewers in 2,000 polling stations scattered in 34 provinces across Indonesia. In each polling station, they interviewed four people (two men and two women, based on Indonesia’s gender distribution).

Popularity surveys: Remember when Prabowo closed the gap between him and Jokowi in electability rates? Dirga said that these electability rates were based on pre-election surveys, or opinion polls. “The method of survey is to take samples before the election. Prospective voters are asked about their political preference,” he said.

Pre-election surveys are the weakest in predicting who would win the race, Dirga said. “Because it’s done before the election, people can still change their mind.”

Sign in to Favourite

Join the conversation

2 Comments sorted by

  1. Ken Alderton

    PhD student, former CEO

    Whenever I use opinion polls from Indonesia I am always very keen to get the "Tidak tahu", Don't know, number. My general experience is that if the question is political or about a sensitive social issue like religion thsi number is in the 20% range. In a political poll this number is bigger than many of the other responses. If the question is not these areas it is down around the normal 5%. I have asked a prominent Australian survey company and they said that a "dont know" response over 5% would worry them. I am as sure as I can be that the methodology of Indonesian surveys is sound.
    Any idea why the number is so high?. I would suggest that this is the reason "Pre-election surveys are the weakest in predicting who would win the race". In the 2004 DPR election a prominent US election organisation ran a whole series of tracking surveys from a couple of months out to election eve an the 'tidak tahu" number hardly changed.

    report
  2. Ken Alderton

    PhD student, former CEO

    There were reports that the voter turnout was likely to be +80% againt but that there would be a significant proportion of voters who would spoil their ballots. Had anyone seen the informal vote numbers in any of the early returns?

    report