Master's Thesis

Beyond the Screen

Media Effects on Voter Turnout & Vote Direction in Indonesia's 2019 Legislative Election

Ali Al Harkan · Universitas Indonesia

In 2019, Indonesia held one of the most complex single-day elections in the world. Against a backdrop of 17,000 islands and a booming digital landscape, a critical question emerged: Does the media we consume actually dictate how we vote?

0M
Eligible Voters
0
Provinces Data
0
Regression Models
0
Media Companies

The New Democratic Landscape

Since the 1998 Reformasi, Indonesia's media landscape has undergone a radical transformation. The repeal of restrictive laws birthed an ecosystem of 1,362 press companies across print, broadcast, and cyber/digital media.

Today, traditional TV and radio coexist with a massive surge in internet users. As media becomes the "fourth pillar" of Indonesian democracy, it serves as the primary bridge between political parties and the public.

But in an era of information saturation, we must ask: Does increased media access lead to higher democratic participation, or simply more noise?

Source: BPS Indonesia, Kominfo, 2000–2019

Challenging the Traditional Models

Academic studies of Indonesian voters typically rely on three pillars:

  1. Sociological: The "Columbia School" approach — how religion, ethnicity, and class bind voters to their groups.
  2. Psychological: Loyalties to specific party figures or historical "Downsian" utility.
  3. Rational Choice: Economic benefits and government performance.

However, external factors — specifically Media Effects — are often relegated to the sidelines. This research integrates media consumption into these traditional models, using aggregate data from all 34 Provinces to see the "big picture" of the 2019 legislative race.

Source: Academic literature review, CSIS surveys

The Statistical Engine

To uncover these hidden relationships, we utilized SEM-PLS (Structural Equation Modeling Partial Least Square). This technique allowed us to test and estimate complex relationships between multiple variables simultaneously.

We executed 32 distinct test models, analyzing how provincial trends in media consumption (TV, Radio, Newspaper, Internet, and Google Search) correlated with the performance of major parties like PDIP, Gerindra, and PKB.

By using aggregate data from BPS and KPU, we move from "what people say" to "what the numbers show" across the entire archipelago.

Source: BPS Statistics Indonesia, KPU Election Commission 2019

The Surprising Power of Radio

In the digital age, we expect the Internet to be the primary driver of participation. However, our data suggests otherwise.

The effect of the Internet on Voter Turnout was found to be weak. Instead, Radio emerged as a significant positive predictor. In provinces where more citizens listened to the radio, participation in the 2019 election was consistently higher.

This suggests that radio remains a potent, perhaps overlooked, tool for civic mobilization in Indonesia's more remote or traditional regions.

Radio listeners showed a Strong Positive Effect (β=0.612) on turnout compared to low-radio provinces, controlling for other variables.
Source: KPU 2019 Turnout Data, BPS Media Survey

Targeted Influences

Media effects are not a "blanket" influence; they are highly fragmented across different political parties:

  • G
    Google Search: Higher search volumes were significant predictors for Gerindra and PKB support.
  • T
    Television: Remains the primary vehicle for parties like Nasdem.
  • R
    Radio: Showed a specific link to support for Demokrat.

This confirms that different media platforms serve as "triggers" for different segments of the electorate. The "echo chamber" in Indonesia isn't one big room — it's a series of smaller, media-specific corridors.

Source: KPU 2019 Party Vote Shares, Google Trends, BPS

The Sum is Greater than the Parts

Media alone is rarely the decider. However, when we look at the Combined Model — merging Media with Sociology (Religion/Ethnicity) and Economics (Poverty) — the predictive power skyrockets.

For PDIP, the combined model explained a staggering 83.9% of the variance in provincial vote direction. This proves that media acts as a "multiplier" for existing social and economic conditions.

It doesn't replace the voter's identity; it reinforces it.

R² = 83.9% for PDIP in the Combined Model — one of the highest predictive fits in Indonesian electoral research.
Source: SEM-PLS Model Output, SmartPLS 3.0

A Call for Granular Data

Our findings show that media effects in 2019 were fragmented and often weak at a provincial level. This suggests that the "battle for the voter" happens at a much more local scale.

To improve future research and democratic transparency, we recommend:

  1. Shifting to District-Level Analysis: Provincial data masks local nuances; we need Kabupaten/Kota granularity.
  2. Expanding Data Access: Public access to Indonesian statistical data must be improved for more sophisticated aggregate research.
  3. Active Monitoring: Stakeholders should actively monitor media's "trigger" effects to ensure a balanced democratic discourse.
Source: Research Recommendations, UI Graduate Program
Indonesia's Media Growth, 2000–2019
Internet penetration surpassed radio listenership around 2015, creating a dual-media landscape heading into the 2019 elections.
The Research Variable Map
Three categories of independent variables — Media, Sociological, and Economic-Political — orbiting two outcome constructs.
SEM-PLS Path Model (Simplified)
Animated paths show causal direction between media variables and electoral outcomes. Line thickness represents path coefficient magnitude.
Radio vs. Voter Turnout — 34 Provinces
Each point is a province. The upward trend for Radio consumption contrasts with the flat trend for Internet, revealing radio's unexpected civic power.
Media–Party Significance Matrix
Highlighted cells indicate statistically significant relationships (p < 0.05). Each media platform triggers support for different parties.
Explaining Power (R²) — Media vs. Combined Model
The combined model that integrates media with sociological and economic variables dramatically outperforms media-only models.
From Provincial to District-Level Analysis
Future research should zoom into Kabupaten/Kota level data to uncover the truly local dynamics shaping Indonesian democracy.