Still trusting the pollsters despite their numerous false predictions?

Tri Wibowo Santoso
Researcher in the People’s Sovereignty Movement (Pergerakan Kedaulatan Rakyat)

IO – The Nazi Minister of Information and Propaganda Paul Joseph Goebbels once said that “If you tell a lie often enough and systematically enough, it will seem to have turned to reality. The perfect lie is the truth twisted just a little bit.”

If we connect Goebbels’s statement with the results of the surveys that state that Presidential Candidate Number 01, Joko Widodo (Jokowi) is winning at a wide margin over his opponent, Presidential Candidate Number 02 Prabowo Subianto, is it relevant?

Quite possibly.

The results of the latest Indonesian Survey Circle (Lingkaran Survey Indonesia – “LSI”) under the leadership of Denny J. A. Shows that the Jokowi-Ma’ruf Presidential Candidate Pair has an electability rate of 58.7%., while the Prabowo Subianto-Sandiaga Uno Presidential Candidate Pair has an electability of 30.9%. The remaining 9.9% voters surveyed claim not to have made up their minds.

Survey agencies are actually an important instrument that measure an individual candidate or party’s electability. However, the issue is that nowadays many of the predictions have missed. The question arises: “Are the results of pre-election surveys engineered by survey agencies to direct public opinion?”

The DKI Jakarta Regional Head Election is still fresh in our minds. At the time, nearly all survey agencies predicted that Gubernatorial Candidate Pair Basuki Tjahaja Purnama-Djarot Saiful Hidayat would win. Charta Politica even stated that on the second round of campaigns, Basuki Tjahaja Purnama-Djarot Syaiful Hidayat’s electability was higher than that of Anies Baswedan-Sandiaga Salahuddin Uno at 47.3% to 44.8%.

Denny J. A.’s LSI predicted something even more extreme: in mid-January 2017, it published a survey result that states that Agus Harimurti Yudhoyono-Sylviana Murni had the highest electability rate among the three pairs at 36.7%. Gubernatorial Candidate Pair Number 2 Basuki Tjahaja Purnama (Ahok)-Djarot Saiful Hidayat was elected by 32.6% of respondent, while the third Pair Anies Baswedan-Sandiaga Uno had an electability of 21.4%. Yet in fact, by the end of the second round of DKI Jakarta Gubernatorial Campaigns, the position was won by Gubernatorial Candidate Pair Number 3 Anies Baswedan-Sandiaga Uno with a landslide vote of 3,240,987 or 57.96%. Gubernatorial Candidate Pair (Ahok)-Djarot only obtained 2,350,366 votes or 42.04%.

However, we still need to acknowledge that LSI predicted that Anies Baswedan-Sandiaga Uno would come first at the second rounds of campaigning, even though he thought that they would be last in the first round.

An even worse miss occurred during the 2018 West Java Regional Head Elections. Nearly all survey agencies made a huge error when predicting the electability of the Gubernatorial Candidate Pair Sudrajat-Ahmad Syaikhu. Saiful Mujani Research and Consulting (SMRC) predicted their electability at only 7.9%; LSI Denny J. A. at 8.2%; and Indo Barometer at 6.1%. The Elections Commission (Komisi Pemilihan Umum – “KPU”)’s final tally recorded Sudrajat-Syaikhu’s votes at 28.7%.

Then there’s also the Gubernatorial Elections in Central Java. In mid-June 2018, LSI Denny J. A. published its electability predictions: Ganjar-Yasin at 54%, Sudirman-Ida at 13%. However, the final tally showed that the margin between the two Gubernatorial Candidate Pair was actually very slight. Ganjar Pranowo dan Taj Yasin, who were supported by PDIP, PPP, Democrat, and National Democrat Party obtained 10,362,694 votes or 58.78%; while Sudirman Said and Ida Fauziyah, who were supported by Gerindra, PKB, PKS, and PAN obtained 7,267,993 votes or 41.22%. The survey agencies’ predictions were off again, as the difference was just too extreme.

Differences between survey results and real count is not the real issue here. The issue is the independence of these survey agencies themselves. This is why the public frequently correctly concludes from the results that survey agencies tend to be biased towards or against certain candidates.

Identifying “Made-to-Order” Survey Agencies
Identifying whether the survey results of an agency are correct and unrigged is actually not hard. First, the data-taking method for the survey must be accurate and the survey must be made according to demographic data. This data is obvious from the list of respondents in each province – all correct, no skewing made by the survey agencies. Second, look at the list of questions made by the survey agencies. If the questions submitted to respondents tend to direct them to elect specific candidates, it would naturally affect survey results.  Third, the methodology applied for the survey and sampling must be correct. Efforts at manipulation would naturally reduce the accuracy of the survey. Finally, the costs of the survey must be transparent.

How to Deal With “Made-to-Order” Survey Agencies
Nowadays, there is no law that states that survey agencies are liable for prosecution if they are proven to use methods to direct respondents’ opinion. Consequently, survey agencies continue to lie to the people by creating surveys that direct opinion, seeming to be credible and based on solid methods.

Even though there are no specific regulations for this, we can still work around it, such as by amending Law Number 16/1997 concerning Statistics. The amendment would also include articles that regulate what will happen if survey agencies are proven to be false and lie to the people. Further, Article 131 Paragraph (3) Letter a of Law No. 10/2016 concerning Regional Elections, which states that “The people’s participation (including surveys) must not be directed to show a bias that would benefit or damage any single Gubernatorial and Vice-Gubernational Candidate Pair, Regent and Vice-Regent Candidate Pair, and Mayor and Vice-Mayor Candidate Pair” must also include sanctions against the survey agencies that violate it.

Most importantly, the membership of any association or agency that monitors survey agencies must be people who are not involved in any specific survey agency to avoid conflict of interest. Such people can be taken from the ranks of academicians and activists. Therefore, there is no burden on the Ethical Council of the Indonesian Public Opinion Survey Agencies’ Association (Perhimpunan Survei Opini Publik Indonesia – “Persepi”) to perform audits against survey agencies that are suspected of data manipulation. Finally, the media should stop publishing the results of surveys made by agencies that are known to be flawed. Therefore, the agency should receive social punishment from the public and disappear automatically.