EXECUTIVE SUMMARY
- Political candidates’ personalities still matter in Indonesia’s personality-centric partisan politics. However, election campaigning can change or make up voters’ minds.
- A two-wave survey of prospective Indonesian voters shows that swing voters shifted their voting preferences as the election campaigning proceeded, due to their changing perceptions of candidates’ personalities and policy platforms.
- Extrapolating from the results, about half of swing voters in Indonesia’s 2024 presidential election chose the winning pair Prabowo Subianto-Gibran Rakabuming Raka, who in parallel had the biggest gains for their perceived personalities and policy platform.
- Polarisation (being at one extreme of the personality-political-policy ideology spectrum) was evident among swing voters who chose Prabowo-Gibran and runners-up Anies Baswedan-Muhaimin Iskandar, as they appeared more likely to vote for their preferred candidate pair if they negatively viewed the other candidate pair.
- Although theoretically we can expect that when more Indonesian voters become better educated, candidates’ policy platforms would become more important vis-à-vis personalities in swaying such voters, Indonesia’s 2024 presidential election showed that candidates tend to ‘dumb down’ their policy platforms to reach a critical mass of voters, thereby nullifying any evidence from the survey that more educated swing voters voted more based on policy platforms compared to their less educated counterparts.
* Julia Lau is Senior Fellow and Co-coordinator of the Indonesia Studies Programme, and Editor, Fulcrum at ISEAS – Yusof Ishak Institute. Her research interests include Indonesia’s foreign policy and domestic politics. Maria Monica Wihardja is Visiting Fellow and Co-coordinator of the Media, Technology and Society Programme at ISEAS – Yusof Ishak Institute, and also Adjunct Assistant Professor at the National University of Singapore.
ISEAS Perspective 2024/77, 2 October 2024
INTRODUCTION: A TWO-WAVE SURVEY ON PROSPECTIVE INDONESIAN VOTERS
Political scientists have as a rule closely studied campaigns and voting behaviour across different political systems. In their 2008 article[1] analysing European politics and political marketing (the practice of ‘selling’ candidates in an election), communications scholars De Landtsheer, De Vries and Vertessen show how “metaphors, sound bites, appearance effectiveness, and personality traits can win elections”. In modern elections, political candidates are the focal point in electoral campaigns. Voters form “intuitive impressions” of candidates and can be swayed by emotion and other irrational factors, regardless of the strength of a candidate’s policy platform.
Separately, research has shown that voters are affected by economic policy platforms[2] as well, and are discerning enough in presidential elections to adjust their vote by balancing between the institutions of the presidency and the legislature.[3] We also know that “information can affect the aggregate result” of elections “either by changing voters’ preferences or…by increasing the probability that voters vote according to these preferences”.[4]
By the time Indonesia voted on 14 February, many Indonesians had been subjected to months of slick campaigning from all candidates, including on social media. What impact did this campaigning have on voter perceptions of the candidates’ personalities and policy platforms, and on how they eventually voted?
In an earlier piece, we concluded from the first wave of a survey commissioned by ISEAS –Yusof Ishak Institute and conducted by Lembaga Survei Indonesia (LSI) on 17-27 November 2023 that “personality still outweighs policy” for Indonesia’s voters.[5] This essay extends that discussion, mainly by focusing on swing voters and closely comparing results from the survey’s first and second waves (conducted on 2-9 February 2024) to reveal what might have tipped respondents’ choices before they went to the ballot box.
We hypothesise that personality still mattered for swing voters when it came to voting day, although as the elections approached and the campaigning went into full swing, policy platforms gradually took centre stage. Generally, Indonesia’s weak party system[6] and personality-centric clientelism[7] more clearly shapes its current personality-centric partisanship than its party-centric partisanship. As Indonesian politics scholar Burhanuddin Muhtadi has argued, such personality-centric clientelism is due to the open-list proportional system, where voters directly elect their legislative representatives by choosing individuals (instead of a political party); this has resulted in election monies flowing to politicians rather than to their parties. It is beyond the scope of this essay to analyse campaign financing and how or whether each of the three candidate pairs had outspent their competitors; be that as it may, a campaign’s material resources are certainly important for winning any national election in Indonesia, given the geographical spread an aspiring presidential candidate must cover during his or her campaign.
Our findings show that prospective voters changed their voting preferences not only because of their changing perceptions about the candidates’ personalities but also because of the candidates’ policy platforms. We assume that voters learn more about each candidate’s personality and policy platform as the campaigning proceeds, such as through debate performances, dissemination of campaign materials, or attendance of campaign events. Although each pair continued to receive a higher approval rating for their personalities than their policy platform toward the end of campaigning, the gaps narrowed as the three pairs made their policy campaign promises known. However, personality and policy factors seemed to matter to voters only when they served to distinguish a candidate from other candidates.
EMPIRICAL CONTEXT AND METHODOLOGY
The ISEAS-LSI two-wave panel survey was conducted through face-to-face interviews with 2,020 respondents across Indonesia’s 34 provinces. All respondents were identified as eligible Indonesian voters, aged 17 years old or older or married if below 17, when the surveys were conducted. Multi-stage proportional and random sampling methods were used to select respondents, with villages (desa) as the primary sampling unit. Both stages preserved the composition of the true (national) population in terms of gender, rural-urban domicile, age group, religious affiliation, ethnicity, and provincial location.
Of these 2,020 in the survey, 1,919 were interviewed in both Waves 1 and 2. Of the 1,919, a total of 602 or about 31% had switched their choices between Wave 1 and Wave 2 (Figure 1 and Annex Table 1). These 602 “swing voters” are the focus of this essay. This behaviour suggests that they did not have a clear choice about who to vote for in Wave 1 but by Wave 2 had decided on one of the candidate pairs. Alternatively, these respondents changed from the preferences they indicated in Wave 1 by the time they were surveyed in Wave 2.
Figure 1. Indonesian Voters Changing Their Minds
Source: Data and chart from ISEAS-LSI Panel Survey (November 2023 and February 2024)
The winners, president-elect Prabowo Subianto and vice president-elect Gibran Rakabuming Raka, gained the lion’s share of 50% of previously undecided voters’ support (6.7% of all respondents). This contrasts with the second-placed Anies Baswedan-Muhaimin Iskandar pair’s gain of 19.3% and third-placed Ganjar Pranowo-Mahfud MD’s 19.7%. Significantly, almost a third of Ganjar-Mahfud’s early supporters switched to the Prabowo-Gibran camp by Wave 2 (see Annex Table 1).
We use the respondents’ choice of a presidential-vice presidential candidate pair in Wave 2 (after election campaigns) as the dependent variable, and changes in the respondents’ perceptions (that is, approval ratings) of the personality and policy platforms of each of the candidate pairs between Wave 1 (before election campaigning) and Wave 2 (after election campaigning) as the independent variable. (See Annex for details.) The dependent variable is a dichotomous dummy variable (1= vote, 0 = otherwise) for each of the candidate pairs, with the following breakdown (Table 1):
Table 1. Votes in Wave 2 among “Swing Voters”
| Candidate Pairs | Vote (% of total swing voters) |
| Anies-Muhaimin | 23.1 |
| Prabowo-Gibran | 54.1 |
| Ganjar-Mahfud | 15.3 |
| Undecided | 5.3 |
| Don’t know/Not applicable | 1.2 |
Source: Wave 2, ISEAS-LSI Panel Survey, authors’ calculations
Our main variable of interest is how changes in respondents’ perceptions of the personalities and policy platforms of each candidate pair affected the respondents’ likelihood of voting for a candidate, given that swing voter respondents changed their preferences from Wave 1 to Wave 2. In both waves, respondents were asked to evaluate each candidate pair’s personality and policy platforms, with their responses indexed from 1 (“strongly dislike” or “strongly disagree” depending on the question) to 5 (“strongly like” or “strongly agree”) (Annex, Part 2). The changes in perceptions range from -4 to 4.
Broadly, the data show that Prabowo-Gibran had the biggest gains in more favourable perception changes for both their personalities and their policy platform, compared to the other two pairs. For example, 40% of swing voters increased their ratings of Prabowo-Gibran’s policy platform while only 27% and 31% of swing voters increased their ratings of Ganjar-Mahfud’s and Anies-Muhaimin’s policy platforms respectively between Wave 1 and Wave 2. Only 22% of swing voters decreased their ratings of Prabowo-Gibran’s personalities while 32% and 28% of swing voters decreased their ratings of Ganjar-Mahfud’s and Anies-Muhaimin’s personalities. Tables 2 and 3 below summarise the findings:
Table 2. Swing Voters’ Perceptions of Candidate Pairs’ Personalities
| Changes in Perception of Personality | Anies-Muhaimin | Prabowo-Gibran | Ganjar-Mahfud |
| Less favourable (-4 to -1) | 28% | 22% | 32% |
| No change (0) | 43% | 47% | 44% |
| More favourable (1 to 4) | 30% | 32% | 24% |
Source: Wave 1 & 2, ISEAS-LSI Panel Survey, authors’ calculations
Table 3. Swing Voters’ Perceptions of Candidate Pairs’ Policy
| Changes in Perception of Policy | Anies-Muhaimin | Prabowo-Gibran | Ganjar-Mahfud |
| Less favourable (-4 to -1) | 30% | 21% | 28% |
| No change (0) | 39% | 38% | 45% |
| More favourable (1 to 4) | 31% | 40% | 27% |
Source: Wave 1 & 2, ISEAS-LSI Panel Survey, authors’ calculations
We then use an odds ratio, logistic regression for each candidate pair to see how the likelihood to vote in Wave 2 was influenced by changes in respondents’ perceptions of the personalities and policy platforms, controlling for respondents’ characteristics (age, gender, income, education, and religion). (The regression specification is given in Annex, Part 3.)
The survey findings show how political polarisation between the eventual winners and the runners-up became more evident after campaigning began. Swing voters were muchmore likely to vote for Anies-Muhaimin if they viewed the team’s policy platform and personalities more positively, while registering higher negative responses to Prabowo-Gibran’s policy platform and personalities. In a mirror image, swing voters were more likely to vote for Prabowo-Gibran if they registered higher positive changes towards their policy platform and personalities while showing higher negative changes in their reaction to Anies-Muhaimin’s policy platform and personalities.
Meanwhile, swing voters who voted for Ganjar-Mahfud were influenced only by increasingly liking this pair’s policy platform. Their votes were not determined by any additional dislike of the other candidates’ personalities and policies.
The following sections delve into more detail by analysing whether and how respondents’ demographic characteristics affected their prospective voting preferences.
HOW EDUCATION AND INCOME LEVELS MIGHT AFFECT VOTING
Among the key findings of Wave 1’s survey were that Anies-Muhaimin’s voter mix or supporters were “more educated, earn higher incomes, and are more connected to the Internet” than the voters for the other two pairs.[8] The survey’s second wave partially confirmed the point on income and education.
First, by looking at the composition of just the swing voters, Anies-Muhaimin’s swing voters (those who did not choose or had not chosen Anies-Muhaimin in Wave 1 but chose them in Wave 2) were more educated compared to those who ended up choosing the other two pairs (Figure 2a). Using the conditional regressions as specified in Annex Part 3, swing voters who completed a university degree or higher were more likely to vote for Anies-Muhaimin but less likely to vote for Prabowo-Gibran, compared to uneducated swing voters (defined as those with less than primary school education).
Although theoretically we can expect that if more Indonesian voters become better educated, candidates’ policy platforms would become more important vis-à-vis personalities in swaying future voters , Indonesia’s 2024 presidential election shows that candidates can ‘dumb down’ their policy platforms to reach a critical mass of voters, nullifying any evidence from the survey that more educated swing voters voted more based on policy platforms compared to their less educated counterparts. For example, offering a generous free school meal programme (as Prabowo did) is a more concrete political promise that more vulnerable prospective voters, who are often also less educated, can relate to, compared to policies on fiscal sustainability.
Second, although there was less significant difference in terms of income class, Prabowo-Gibran’s swing voters comprised more poor voters (whose incomes fall below the poverty line) compared to those the other two pairs (Figure 2b).
Figure 2a. Share of Swing Voters by Education (%) Figure 2b. Share of Swing Voters by Income (%)
Source: Wave 1 & 2, ISEAS-LSI Panel Survey, authors’ calculations
HOW RELIGIOUS AFFILIATIONS AND AGE MIGHT AFFECT VOTING
Looking at the profile of swing voters by religion, we find that Anies-Muhaimin’s swing voters comprised only Muslim voters while Ganjar-Mahfud’s swing voters comprised 79% Muslim voters and Prabowo-Gibran’s, 83% (Figure 3). This is not surprising given that Anies-Muhaimin’s political base was the Prosperous Justice Party (PKS), while Ganjar-Mahfud’s base was the Indonesian Democratic Party of Struggle (PDI-P), a more nationalist and pluralist party within the political-ideological spectrum of Indonesian politics.
Figure 3. Share of Swing Voters by Religion (%)
Source: Wave 1 & 2, ISEAS-LSI Panel Survey, authors’ calculations
Swing voters who were Hindu were more likely to vote for Prabowo-Gibran compared to Muslim ones, while Christian swing voters were more likely to vote for Ganjar-Mahfud compared to Muslim ones. Arguably, this was because Ganjar courted[9] and enjoyed strong support from various Catholic clergy and other groups, including the alumni association of prominent Catholic schools.[10]
Prior to voting day, pundits and commentators set much store on the prospective influence[11] that younger Indonesians would have in deciding the election. Looking at the profile of the swing voters by age, Ganjar-Mahfud’s swing voters were older compared to those of the other two candidate pairs (Figure 4). One possible explanation is that younger voters used more social media than older voters. Since Anies-Muhaimin and Prabowo-Gibran used social media campaigning more intensely than Ganjar-Mahfud, the younger swing voters might have been more easily swayed to choose Anies-Muhaimin or Prabowo-Gibran. Interestingly, when controlling for respondents’ characteristics using the regressions specified in Annex 3, the survey findings reveal that Gen-Z (born after 1996 to early 2010s) swing voters were more likely to vote for Anies-Muhaimin compared to their older counterparts. Millennial (born between 1981-1996) and Baby Boomer (born 1946-1964) swing voters were more likely to vote for Prabowo-Gibran compared to Gen Z ones.
Figure 4. Share of Swing Voters by Age (%)
Source: Wave 1 & 2, ISEAS-LSI Panel Survey, authors’ calculations
The likelihood of voting for Ganjar-Mahfud among swing voters was not significantly influenced by income, age or education, which further suggests that their campaign strategy might have been acceptable to all types of voters but just not appealing enough for them to win. The above results remain robust when controlled for respondents’ access to the Internet.
In addition, drawing from the above findings, for Ganjar-Mahfud, the leakage of their support to Prabowo-Gibran could be partly explained by the former pair’s failure to carve out a distinct platform and to project a political persona.[12] Although different dynamics were at play, these findings could partly be explained by Ganjar-Mahfud’s measured campaign strategy. They pitched more moderate political and policy views compared to their rivals; while this was less polarising, it left voters wanting – or even confused. Their failure to come up with distinct political personas and clear policy platforms turned out to be detrimental. For example, Ganjar’s oft-praised “close to the people”[13] personality was probably too close to President Joko Widodo’s (Jokowi) and by attribution, Gibran’s. Furthermore, in regards to policy platforms, for example, Ganjar-Mahfud’s “continuity with improvement” campaign[14] could have been seen as being too similar to Prabowo-Gibran’s.
The leakage of Ganjar-Mahfud’s supporters to Prabowo-Gibran’s camp could also be explained by how Ganjar was previously under the shadow of Jokowi’s influence and popularity. The PDI-P was Ganjar’s base and Jokowi’s then-affiliated party. Meanwhile, Anies and Prabowo may have had loyal, even staunch or fanatic, supporters as their electoral base. Consequently, after Gibran’s nomination as Prabowo’s running mate, Jokowi’s loyal supporters shifted their votes away from Ganjar-Mahfud to support Prabowo-Gibran. Moreover, Prabowo-Gibran represented stronger assurance of continuity of Jokowi than Ganjar-Mahfud. In one explanation for this shift, Muhtadi and Muslim have argued that the response of Ganjar-Mahfud’s team in starting to attack Jokowi for his political betrayal and alleged intervention in the electoral process proved counter-productive.[15] Anies-Muhaimin’s base, in contrast, remained stable as loyalists doubled down on their preferences as voting day approached.
CONCLUSION AND FURTHER RESEARCH
This essay provides a broad overview of what might influence swing voters in a three-way presidential race, particularly from the perspective of personality versus policy platforms within the personality-centric politics of Indonesia.
The results of the two survey waves shed light on a small piece of the electoral puzzle that all politicians aspiring for national office aim to understand: how to convince swing voters to vote their way come election day. Though a three-horse fight, Indonesia’s 2024 election had seemingly ended up as a two-way highly polarised race in terms of ideology, personality and policy between Team Prabowo-Gibran and second-placed Anies-Muhaimin. Although swing voters who shifted to Ganjar-Mahfud were less polarised in their views of the candidates, the election showed clearly that for Indonesia’s politicians, failing to distinguish oneself from other political candidates in terms of political persona and policy platform can be fatally detrimental.
As Indonesian voters become more informed and educated, we can expect (and hope) theoretically that politicians’ policy platforms will come to play a greater role in swaying voters despite the structural constraints in Indonesia’s political institutions, which will continue to entrench a personality-centric electoral system. However, it is not evident from the survey that more educated swing voters voted more based on policy platforms compared to their less educated counterparts. This is perhaps because presidential candidates can ‘dumb down’ their policy platforms. What is evident from the survey is that general Indonesian voters made more informed decisions based on the policy platforms and not just personalities of the candidates, as election campaign proceeded.
There are various inquiries that scholars can further explore. For instance, future studies could examine more closely what role voters’ access to different election campaign media (television, social media, et cetera) play, and how disinformation or misinformation and election-related propaganda narratives – with or without deepfake content – may further sway the minds of swing voters.
ANNEX and ENDNOTES
For annex and endnotes, please refer to the pdf document.
| ISEAS Perspective is published electronically by: ISEAS – Yusof Ishak Institute 30 Heng Mui Keng Terrace Singapore 119614 Main Tel: (65) 6778 0955 Main Fax: (65) 6778 1735 Get Involved with ISEAS. Please click here: /support/get-involved-with-iseas/ | ISEAS – Yusof Ishak Institute accepts no responsibility for facts presented and views expressed. Responsibility rests exclusively with the individual author or authors. No part of this publication may be reproduced in any form without permission. © Copyright is held by the author or authors of each article. | Editorial Chairman: Choi Shing Kwok Editorial Advisor: Tan Chin Tiong Editorial Committee: Terence Chong, Cassey Lee, Norshahril Saat, and Hoang Thi Ha Managing Editor: Ooi Kee Beng Editors: William Choong, Lee Poh Onn, Lee Sue-Ann, and Ng Kah Meng Comments are welcome and may be sent to the author(s). |