Authored by Tavian MacKinnon
In January 2018, President Joko Widodo appointed Major General Djoko Setiadi as Head of the newly formed National Cyber and Encryption Agency to help national intelligence agencies and law enforcement efforts combat online misinformation and fake news in advance of the April 2019 general elections. In the run-up to the general elections, social media companies worked with the government to block and remove fake content to combat the spread of misinformation online. Although the government and private sector have taken substantial steps to mitigate misinformation and fake news, the increased usage of social media platforms as a source of news has escalated the prevalence of hate speech and disinformation in Indonesia.
This poses a direct threat to the fourth most populous country in the world by eroding public trust in institutions and the electoral processes, as well as increasing extremism and creating an environment of intolerance in one of the most diverse countries in Southeast Asia. The problem of misinformation and fake news became even more apparent in the post-election violence that followed the reelection of President Joko Widodo on April 17, 2019 which was flared by numerous inaccurate and false news stories surrounding the election results that were shared on social media.
In order to quantify Indonesia’s adult population access to traditional media and social media, as well as their exposure to misinformation and fake news, the University of Notre Dame in collaboration with GeoPoll, designed a comprehensive baseline survey instrument. From March 27 – April 24, 2019, GeoPoll’s team of interviewers in Jakarta, Indonesia, implemented a computer-assisted telephone interview (CATI) for 1,000 successful interviews. The survey was nationally representative proportionate to 2019 census data estimates at the first administrative level (province), age group and gender. The main interview language of the survey was Indonesian, but the interviewer team was able to switch other languages such as Balinese and Javanese if requested by the respondent.
From March 27 – April 2, 2019, GeoPoll conducted a training to its data collection team for the survey in Jakarta. GeoPoll’s local team of interviewers consisted of a fourteen-member unit, twelve of which were interviewers and two managers. Before data collection commenced, the team members were trained on standard CATI best-practices such as research ethics, operator refusal aversion, questionnaire design and aim, practice interviews as well as conducting a pre-test of the survey.
During the planning stages of the survey, a nationally representative framework with interlocking quotas at the first administrative level, age group and gender was created proportionately to the population of Indonesia using 2019 census estimates. A probability sampling method was utilized for the project, as sample members were selected using a simple random sampling approach of verified mobile phone numbers in Indonesia. During data collection, the sampling framework was used as a guideline for quota management. The data collection team only closed the survey once all interlocking quotas were met in the sampling framework.
During data collection, the enumeration team made more than 11,000 phone calls that resulted in a 21 percent response rate. From the overall call attempts, 1,045 or 9 percent agreed to participate in the survey while 1,299 or 12 percent refused to participate in the survey. Approximately 43 of the respondents that were reached ended the survey prematurely resulting in a 0.38 percent break-off rate. Additionally, two sample members contacted were not eligible for the survey as they were below the age of 15. These dispositions resulted in 1,000 completed surveys for a 9 percent completion rate. These survey dispositions are aligned with what GeoPoll normally experiences with CATI surveys in Indonesia.
The margin of error in survey research defines the amount of the random variation underlying a survey’s results. This can be thought of as a measure of the variation one would see in reported percentages if the same survey were taken multiple times by the population of interest. The larger the margin of error, the less confidence one has with the survey results. The margin of error for the entire survey project at 95% confidence level is +/- 3.10%.
All 34 provinces in Indonesia are represented proportionately to Indonesia’s population for 1,000 completed surveys. Although the age and gender quotas are proportional within each province, overall 34.9 percent of the respondents that completed the survey were women and 56.1 percent were men. For age groups, 21.5 percent of the respondents were between the ages of 15-24, 29.4 percent between the ages of 25-35, 20.7 percent between the ages of 36-45, while 15.4 percent were between the age of 46-55. For the older subset of the population, 8.5 percent of those that completed the survey were between the ages of 56-65 and 4.5 percent were 66 or older.
Other demographics that were collected, but that were not part of the sampling framework includes education levels, religious affiliation as well as if the respondent lives in an urban or rural area. For educations levels, one percent of the respondents replied they have never attended school and 1 percent Quranic education. Approximately two percent indicated that they did not complete primary education, while 15 percent completed primary school. A majority of the respondents at 59 percent completed secondary education, while 22 percent reported they obtained some form of higher education. The vast majority of respondents self-reported Islam as their religion at 86 percent, while 12 percent indicated they were Christian, and two percent had another religious affiliation. Lastly, 57 percent of the respondents noted they lived in a rural area and 42 percent identified the area they lived in as urban.
Media Consumption – Indonesia
To evaluate the type of media in terms of median and broad topics, the survey measured access to various media platforms as well as determining which type of content that an individual regularly views and follows.
Only 25.3 percent of those that completed the survey indicated that they often or always follow national political affairs and events, while the remaining respondents indicated sometimes, seldom or never. In terms of international news, 85 percent of the respondents indicated they sometimes, seldom or never follow international events and affairs. Male respondents were more likely to follow international events and affairs compared to female respondents, as 16.4 percent of male respondents indicated they almost always, or often, follow international events and affairs, compared to 8.7 percent of female respondents. Local news was more impactful as approximately 41 percent of those surveyed indicated they almost always or often follow local events and affairs.
A focus of the survey was on users of social media in order measure how misinformation and fake news stories are spread. In the survey results, 75.9 percent of the respondents use social media, of which 93.6 percent reported they access social media on a daily or weekly basis. From the subset of respondents that use social media, other information medians such as television were viewed daily by 72 percent, while 54 percent replied they did not listen to radio at all. Print media such as a physical newspaper or magazine was generally split by respondents, with notable differences among men and women. Overall, only 12 percent of respondents read a physical newspaper or magazine daily, while 37.8 percent of men reporting they read a newspaper or magazine at least weekly. However, 59 percent of the women that completed the survey either never read a newspaper or magazine (51%) or do so less often than once in a month (8%).
For the subset that use social media, when asked what the primary source of news was, television led as the median to a plurality of respondents with approximately 35 percent indicating this was the main platform for news. Facebook was the second most reported platform for news as 14 percent overall with a slight gender difference as 16 percent of women and 12 percent of men respectively selecting this median as the main source of news. The remaining 51 percent of the subset indicated an array of medians such as other social media platforms on the internet traditional news platforms such as magazines, newspapers and radio. Overall, social media platforms were selected by 38 percent of the subset that use social media as the primary source of news in Indonesia.
Indonesian Social Media Use
Respondents that had earlier in the survey indicated that they use social media outlets were asked a series of questions to determine what popular platforms they have an account on and where they get news from. The popular platforms asked in the survey are Facebook, YouTube, Twitter and WhatsApp. Other social media platforms were captured in the survey through an open-ended style question to record what other platforms social media users in Indonesia utilize.
Respondents that use social media were more likely to have a WhatsApp account (86%) compared to all other social media platforms. Facebook came in second for the subset that use social media, as 79 percent of the respondents reported having a Facebook account. Only 18 percent of those that use social media and that were surveyed in the study indicated they have a Twitter account. For YouTube, 45 percent of the subset that use social media replied they have an account on the popular video platform. Approximately 16 percent of the subset that uses social media said they have an account on other platforms. Instagram led with this subset at 67 percent.
For the subset of respondents that use social media, 54 percent replied that they use Facebook regularly to read and watch local news and events on. Approximately 38 percent of the subset indicated they use YouTube to read and watch local news and events on, with male respondents having a propensity to use YouTube for this purpose compared to female respondents. Furthermore, 32 percent of the subset that uses social media noted that WhatsApp was used regularly to read and watch local news and events on. Only 8% of the subset that uses social media said using Twitter for reading and watching local news and events on, while 17 percent reported another social media platform and 4 percent indicate no social media platform.
For global news and events, YouTube led among the subset that uses social media at 43 percent, with 45 percent male respondents and 41 percent of female respondents selecting this social media platform. Facebook was the second most selected platform for accessing global news and events at 36 percent compare. Interestingly, about one-tenth of respondents overall replied there are no social media platforms they use to read and watch global news and events on, with a 16 percent female and 12 percent of the male breakdown.
Overall, accessing social media accounts and how local and global news is read and watched provide insights into how information is spread. For example, a large majority, at 86 percent of those surveyed that use social media, have a WhatsApp account, but only one-third uses the platform to read and watch local news and events on, suggesting that content is not widely disseminated on this platform compared to other social media platforms. Perhaps not surprisingly, social media platforms that are known to monetize advertisements and targeted content, such as Facebook and YouTube, are the leading platforms for those surveyed to read and watch both local and global news and events on.
Misinformation & False News
Respondents were asked how often they read news stories before sharing them with others on social media. A plurality of the subset that use social media noted that they sometimes (30%) read the full news story before sharing. Approximately 42 percent of the subset indicated they rarely or never read the news story before sharing, while only 28 percent indicated they always or very often read the news story before sharing on social media. Similarly, when asked if the respondent “likes” or shares a news item on social media and what part of the post would the respondent typically read before sharing, 25 percent said they read just the headline. Only 40 percent of the subset admitted to reading the entire article before liking or sharing the news item on social media.
One of the more compelling data points of the survey examined if respondents intentionally share news stories that they know to be false with others. When asked directly, 4 percent of the survey participants that use social media reported they shared false stories to others. Reasons behind sharing false stories include enlightening others on false news as 56 percent of the subset, 16 percent for fun and 24 percent that indicated fake news stories do not harm anyone, they are interesting, and that people enjoy it.
Accessing Real, Misleading & Misinformation
To measure how a representative sample in Indonesia accesses the accuracy of real, misleading and misinformation news in the media, a randomized module was developed for the survey. In coordination with Cekfakta, University of Notre Dame team members provided GeoPoll with a series of news story headlines that were classified as real, misleading and misinformation. Five real news headlines –from news stories that are widely regarded as accurate and truthful information from reputable media sources– that would be well-known in Indonesia were selected. Additionally, five misleading new headlines were provided, which were classified as misleading as a part of the story was inaccurate but had a portion that was accurate. Lastly, four news stories that were classified as misinformation, or fake news, were provided in descriptive headlines. These news stories contain false information and unsubstantiated events and occurrences. All news story headlines provided in each category were widely shared through social media in Indonesia and were selected to be used in the module.
The randomization module developed within the survey instrument provided each respondent that completed the survey were to evaluate two real, one misleading and one misinformation news headline. Respondents were first asked if they had heard of the news story through the descriptive headline. For the portion of respondents that heard of the headlines, a question was asked for each headline to rank the accuracy of the individual news story. This module was developed so that each respondent that took the survey had a random chance of being provided the news stories selected overall for the study at a 2:1:1 ratio.
Exactly 35.7 percent of the respondents that completed the survey had heard the two real news stories from the headlines they were randomly provided in the module. From this subset, 30 percent believed the stories were very accurate of somewhat accurate when asked to access each of the two stories’ accuracy. Interestingly, 41 percent of the subset believed the real news stories were neutral, while 24 percent said the stories were somewhat inaccurate or very inaccurate. The remaining 7 percent of the subset either did not know or refused to answer the question.
For the part of the module that presented a misleading news headline randomly, only 24 percent of those that completed the survey had heard the one story they were asked about. Out of this subset, 24 percent believed the misleading story was very accurate or somewhat accurate, while 46 percent reported the story was neutral. A quarter of the subset (25%) believe the misleading news story was somewhat inaccurate or very inaccurate. The remaining 4 percent of the subset either did not know or refused to answer the question.
The misinformation news story headlines were known by approximately 41 percent of the respondents that completed the survey. Thirty-seven percent of the subset that heard of the news stories in the misinformation portion of the module believe the news story they knew of was very accurate or somewhat accurate, which was the highest out of any of the news story classifications. Approximately 31 percent noted the news story was neutral, while 28 percent said it was somewhat inaccurate or very inaccurate. Similarly, to the misleading portion of the module, 4 percent of the subset did not know or refused to answer the question to rank the story’s accuracy.
The baseline survey results display that a sample of 1,000 Indonesians proportionately representative to the overall population by gender, age and provinces is exposed to and spreads misinformation, unbeknownst to the individual. Nearly 80 percent of the survey respondents regularly use social media and about 70 percent of this subset admitted to sharing a news stories without always reading the full article. Furthermore, in the module that measured respondents’ knowledge of real, misleading and misinformation news stories, results varied greatly among survey participants. Proportionately, more respondents heard of the misinformation news stories compared to the real and misleading stories. More importantly, this subset believed that the misinformation news stories in the module were more accurate comparatively to the real and misleading stories. Even the respondents that had heard of the real news stories were split on whether or now the stories were accurate, neutral or inaccurate. Only four percent of the respondents that completed the survey indicated they knowingly share news stories they know to be false. These results suggest that media literacy and the ability for Indonesians to accurately evaluate media is lacking among the general population regardless of location, age, gender, education and religious affiliation.