25 June 2014
CO14120 | Indonesian Presidential Election: Will Social Media Forecasts Prove Right?
RSIS Commentary is a platform to provide timely and, where appropriate, policy-relevant commentary and analysis of topical and contemporary issues. The authors’ views are their own and do not represent the official position of the S. Rajaratnam School of International Studies (RSIS), NTU. These commentaries may be reproduced with prior permission from RSIS and due credit to the author(s) and RSIS. Please email to Editor RSIS Commentary at [email protected].
Synopsis
Social media support for Indonesian presidential candidates are being used as indicators of possible eventual victory. However, the use of these tools presents certain challenges for analysts when relying on them to inform decision-making.
Commentary
IN RECENT years, the successful use of social media data to predict consumer behaviour by private companies has extended to its forecasting of future trends in areas such as elections and disasters. In the lead-up to the Indonesian presidential election on 9 July 2014, political pollsters have sought to leverage on social media data to provide stakeholders with up-to-date information and even predictions of possible results of the election.
Whether online support will result in eventual electoral victory, however, remains to be seen. Policymakers therefore need to be judicious when basing their decisions solely on social media data.
Traditional opinion polls versus social media
Indonesia has the fifth highest number of social media users worldwide, behind the United States, Brazil, Japan and the United Kingdom. Communications company, PT Bakrie Telecom, for example, estimated that Indonesia has 19.5 million Twitter users. Webershandwick, a public relations organisation, estimated the archipelago has 65 million active Facebook users. The popularity of social media platforms in the country suggests that analysing its content is useful for understanding the dynamics of online information diffusion as well as human behaviour.
While this knowledge has helped commercial organisations better understand their target market, pollsters are beginning to realise the usefulness of such data for understanding the possible outcome of events of national importance such as elections.
The 2012 gubernatorial election in Jakarta proved the deficiency of traditional opinion polls in predicting voter choice. Most polls had predicted the pair of Fauzi Bowo (Foke) and Nachrowi Ramli (Nara) to win, only to be proven wrong by the victory of Joko Widodo (Jokowi) and Basuki Tjahaja Purnama (Ahok). The event raised questions about the reliability of opinion polls in measuring public opinion.
Given the lack of accuracy of opinion polls, the months leading up to the presidential election have seen various social media consultancy firms attempting to gauge public sentiment towards the presidential candidates: Jokowi and Jusuf Kalla (representing the Democratic Party of Struggle, or PDI-P) versus Prabowo Subianto and Hatta Rajasa (representing the Great Indonesia Movement Party or Gerindra).
The overall objective of these opinion mining services is to answer the million-dollar question: who will be elected the next president and vice-president of Indonesia?
Social media support and eventual victory
Social media analysts believe that a high degree of positive mention on social media platforms would result in a candidate’s electoral victory. Yose Rizal, director of Politicawave, an Indonesian social media consultancy firm, observed that the results of gubernatorial elections in various parts of Indonesia showed that the candidate who netizens talked about the most and garnered high levels of positive sentiment often won elections.
In the lead-up to the current presidential election, the Jokowi-Kalla pair have been receiving more positive responses among social media users as compared to Prabowo-Hatta. Politicawave’s analysis of social media platforms showed a higher sentiment index for the Jokowi-Kalla pairing, indicating preference for the PDI-P candidates.
During the first televised presidential debate on 9 June 2014, Politicawave reported that support among Twitter users for Jokowi-Kalla was four times higher than that of Prabowo-Hatta. The hashtag #PresidenNomor2 referring to the candidacy number for Jokowi-Kalla totalled 47,610 compared to #PilihankuSatu, referring to Prabowo and Hatta at 16,000.
Jokowi continued to win netizens’ vote after the second televised presidential debate on 15 June 2014, excelling in five segments of Politicawave’s analysis in comparison to Prabowo’s victory in only two segments. If Politicawave’s argument holds true, then it would appear that the Jokowi-Kalla pair are poised to win the upcoming election.
Moreover, even prior to the announcement of his candidacy, Jokowi had been dominating the social media chatter. In September 2013, Politicawave highlighted that social media mentions of Jokowi’s name comprised 60 percent of nearly four million mentions in its survey. The presidential candidate also dominated social media discussions on a wide range of topics, including politics, natural disasters and infrastructural developments.
Since his declared candidacy on 14 March 2014, online mentions about Jokowi increased significantly from an average social mention of 20,000 to 40,000 per day prior to the declaration to almost 119,000 after the announcement. By 15 March 2014, the number increased to 164,000. The hashtag supporting his candidacy, #JKW4P became a trending topic on Twitter during that time. Given the huge popularity of Jokowi and his vice-presidential candidate, Politicawave believed that the Jokowi-Kalla pair will win the election in just one round.
Challenges to social media analysis
While the work of social media consultancy firms in informing the public about online trends is admirable, sentiment analysis remain problematic. Due to challenges inherent in sentiment analysis, there is a need to exercise caution towards the limitations of social media analytical tools even if they are state-of-the-art. This is especially so when utilising social media data for decision-making. Some of the challenges include:
Firstly, social media is inherently a “noisy” medium filled with informal language, misspelling, and grammatical errors. Sentiment analysis tools, depending on their level of sophistication, still face difficulty in correctly interpreting the meaning of social media conversations.
Secondly, social media analytical tools only capture sentiment at a particular point in time, which may not reflect future sentiment. This is especially relevant when analysing voting behaviour, which is often fluid due in part to swing voters. While analytics tools can capture data in real-time, they may not produce the best analysis in real time.
Finally, social media analysis should be contextualised through a thorough understanding of the society it is situated in. Access to the Internet and mobile devices, although growing, tends to be available to the middle and upper classes. The usage of social media is also more vociferous among the young. Therefore, the sentiments of groups that do not fall into these categories cannot be accurately captured via social media analytical tools.
In the final weeks leading up to the presidential election, the campaigns have been vigorous both online and offline. Analysing social media data, while useful for gauging possible trends for the future, is nevertheless problematic. In the case of Indonesia, it remains to be seen if online sentiment translates to electoral outcome on election day.
About the author
Jennifer Yang Hui is an Associate Research Fellow at the Centre of Excellence for National Security (CENS), a constituent unit of the S. Rajaratnam School of International Studies (RSIS), Nanyang Technological University.
Last updated on 05/09/2014
RSIS Commentary is a platform to provide timely and, where appropriate, policy-relevant commentary and analysis of topical and contemporary issues. The authors’ views are their own and do not represent the official position of the S. Rajaratnam School of International Studies (RSIS), NTU. These commentaries may be reproduced with prior permission from RSIS and due credit to the author(s) and RSIS. Please email to Editor RSIS Commentary at [email protected].
Synopsis
Social media support for Indonesian presidential candidates are being used as indicators of possible eventual victory. However, the use of these tools presents certain challenges for analysts when relying on them to inform decision-making.
Commentary
IN RECENT years, the successful use of social media data to predict consumer behaviour by private companies has extended to its forecasting of future trends in areas such as elections and disasters. In the lead-up to the Indonesian presidential election on 9 July 2014, political pollsters have sought to leverage on social media data to provide stakeholders with up-to-date information and even predictions of possible results of the election.
Whether online support will result in eventual electoral victory, however, remains to be seen. Policymakers therefore need to be judicious when basing their decisions solely on social media data.
Traditional opinion polls versus social media
Indonesia has the fifth highest number of social media users worldwide, behind the United States, Brazil, Japan and the United Kingdom. Communications company, PT Bakrie Telecom, for example, estimated that Indonesia has 19.5 million Twitter users. Webershandwick, a public relations organisation, estimated the archipelago has 65 million active Facebook users. The popularity of social media platforms in the country suggests that analysing its content is useful for understanding the dynamics of online information diffusion as well as human behaviour.
While this knowledge has helped commercial organisations better understand their target market, pollsters are beginning to realise the usefulness of such data for understanding the possible outcome of events of national importance such as elections.
The 2012 gubernatorial election in Jakarta proved the deficiency of traditional opinion polls in predicting voter choice. Most polls had predicted the pair of Fauzi Bowo (Foke) and Nachrowi Ramli (Nara) to win, only to be proven wrong by the victory of Joko Widodo (Jokowi) and Basuki Tjahaja Purnama (Ahok). The event raised questions about the reliability of opinion polls in measuring public opinion.
Given the lack of accuracy of opinion polls, the months leading up to the presidential election have seen various social media consultancy firms attempting to gauge public sentiment towards the presidential candidates: Jokowi and Jusuf Kalla (representing the Democratic Party of Struggle, or PDI-P) versus Prabowo Subianto and Hatta Rajasa (representing the Great Indonesia Movement Party or Gerindra).
The overall objective of these opinion mining services is to answer the million-dollar question: who will be elected the next president and vice-president of Indonesia?
Social media support and eventual victory
Social media analysts believe that a high degree of positive mention on social media platforms would result in a candidate’s electoral victory. Yose Rizal, director of Politicawave, an Indonesian social media consultancy firm, observed that the results of gubernatorial elections in various parts of Indonesia showed that the candidate who netizens talked about the most and garnered high levels of positive sentiment often won elections.
In the lead-up to the current presidential election, the Jokowi-Kalla pair have been receiving more positive responses among social media users as compared to Prabowo-Hatta. Politicawave’s analysis of social media platforms showed a higher sentiment index for the Jokowi-Kalla pairing, indicating preference for the PDI-P candidates.
During the first televised presidential debate on 9 June 2014, Politicawave reported that support among Twitter users for Jokowi-Kalla was four times higher than that of Prabowo-Hatta. The hashtag #PresidenNomor2 referring to the candidacy number for Jokowi-Kalla totalled 47,610 compared to #PilihankuSatu, referring to Prabowo and Hatta at 16,000.
Jokowi continued to win netizens’ vote after the second televised presidential debate on 15 June 2014, excelling in five segments of Politicawave’s analysis in comparison to Prabowo’s victory in only two segments. If Politicawave’s argument holds true, then it would appear that the Jokowi-Kalla pair are poised to win the upcoming election.
Moreover, even prior to the announcement of his candidacy, Jokowi had been dominating the social media chatter. In September 2013, Politicawave highlighted that social media mentions of Jokowi’s name comprised 60 percent of nearly four million mentions in its survey. The presidential candidate also dominated social media discussions on a wide range of topics, including politics, natural disasters and infrastructural developments.
Since his declared candidacy on 14 March 2014, online mentions about Jokowi increased significantly from an average social mention of 20,000 to 40,000 per day prior to the declaration to almost 119,000 after the announcement. By 15 March 2014, the number increased to 164,000. The hashtag supporting his candidacy, #JKW4P became a trending topic on Twitter during that time. Given the huge popularity of Jokowi and his vice-presidential candidate, Politicawave believed that the Jokowi-Kalla pair will win the election in just one round.
Challenges to social media analysis
While the work of social media consultancy firms in informing the public about online trends is admirable, sentiment analysis remain problematic. Due to challenges inherent in sentiment analysis, there is a need to exercise caution towards the limitations of social media analytical tools even if they are state-of-the-art. This is especially so when utilising social media data for decision-making. Some of the challenges include:
Firstly, social media is inherently a “noisy” medium filled with informal language, misspelling, and grammatical errors. Sentiment analysis tools, depending on their level of sophistication, still face difficulty in correctly interpreting the meaning of social media conversations.
Secondly, social media analytical tools only capture sentiment at a particular point in time, which may not reflect future sentiment. This is especially relevant when analysing voting behaviour, which is often fluid due in part to swing voters. While analytics tools can capture data in real-time, they may not produce the best analysis in real time.
Finally, social media analysis should be contextualised through a thorough understanding of the society it is situated in. Access to the Internet and mobile devices, although growing, tends to be available to the middle and upper classes. The usage of social media is also more vociferous among the young. Therefore, the sentiments of groups that do not fall into these categories cannot be accurately captured via social media analytical tools.
In the final weeks leading up to the presidential election, the campaigns have been vigorous both online and offline. Analysing social media data, while useful for gauging possible trends for the future, is nevertheless problematic. In the case of Indonesia, it remains to be seen if online sentiment translates to electoral outcome on election day.
About the author
Jennifer Yang Hui is an Associate Research Fellow at the Centre of Excellence for National Security (CENS), a constituent unit of the S. Rajaratnam School of International Studies (RSIS), Nanyang Technological University.
Last updated on 05/09/2014
