Political Advertising on Facebook During the 2022 Australian Federal Election: A Social Identity Perspective

This study analyzes Facebook and Instagram political advertising during the 2022 Australian federal election using Meta's Ad Library, revealing that major parties focused on reinforcing partisan identities to prevent defection while smaller parties emphasized issue-specific messages to capture disaffected voters, a strategic divergence interpreted through Social Identity Theory within the context of compulsory voting.

The Gist

Imagine the 2022 Australian Federal Election as a massive, high-stakes game of "Who can shout the loudest and most effectively in a crowded stadium?" This paper acts like a pair of binoculars, letting researchers watch exactly how the different teams (political parties) used Facebook and Instagram to shout their messages, who they shouted at, and what they shouted about.

Here is a breakdown of their findings, explained simply:

1. The Setup: A Different Kind of Game

In many countries, political parties spend their time trying to convince people to show up to vote (like a coach trying to get players out of the locker room). But in Australia, voting is mandatory. Everyone has to show up.

Because everyone is already in the stadium, the parties don't need to worry about getting people to the game. Instead, they act like salespeople trying to convince the crowd to buy their specific brand of soda instead of the other guy's. The goal isn't "Get Out The Vote"; it's "Get Out The Vote for Us."

2. The Heavy Hitters vs. The Underdogs

The researchers looked at the "receipts" (the money spent and the number of people who saw the ads).

• The Giants: The two big teams, the Labor Party (ALP) and the Liberal Coalition, were the heavyweights. Together, they spent over 70% of all the money and reached over 70% of the people. It was like two massive cruise ships dominating the ocean, while everyone else was in small dinghies.
• The Spenders: Labor spent the most (about $5 million AUD), followed closely by the Liberals.
• The Efficiency: Interestingly, the smaller parties, like the Greens, spent much less money but still got a decent number of votes. It's like a small, agile speedboat that managed to get noticed without needing a massive engine.

3. Timing: The "Last Minute Rush"

The researchers noticed a pattern in when the ads appeared.

• The Slow Start: After the election was officially called in April, the ad activity was like a gentle hum.
• The Crescendo: As the election day (May 21) got closer, the noise got louder and louder.
• The Peak: Just before the "media blackout" (a rule where TV and radio can't run ads right before voting), the Facebook ads went into overdrive. It was like a sprinter realizing the finish line was seconds away and giving it everything they had.

4. Who Were They Talking To? (The Targeting)

You might think political parties use complex algorithms to whisper specific secrets to specific people (micro-targeting). The paper found that, surprisingly, they mostly used a shotgun approach rather than a sniper rifle.

• Geography: Most ads were just aimed at entire states (like Queensland or New South Wales), similar to how a radio station broadcasts to a whole city rather than a single house.
• Age: Most parties aimed at everyone over 18. They didn't try to hide ads from older people or show them only to young people.
• Gender: There was a slight tendency to show more ads to women, but it wasn't a strict rule.
• The Exception: The United Australia Party (UAP) was a bit different; they seemed to aim more at men and focused heavily on Queensland, their home turf.

5. The Message: "Us vs. Them" vs. "The Issues"

This is where the researchers used a psychological lens called Social Identity Theory. Think of this as understanding how people feel about their "team."

• The Big Parties (Labor & Liberals): Their ads were like team jerseys. They focused heavily on their own party name and attacking the other team. They were trying to say, "We are the good team, they are the bad team." This reinforces the idea that if you are already a fan, you should stay loyal. They were trying to stop their own fans from defecting to the other side.
• The Small Parties (Greens, Independents): Their ads were like issue-specific flyers. They talked about specific problems like "Climate Change" or "Mining." They were trying to build a new team identity based on what they believed in, rather than just who they were. They were trying to catch the attention of people who were unhappy with the big teams.

6. The Big Picture Takeaway

The paper concludes that in a country where everyone must vote, the big parties play a defensive game (protecting their loyal fans), while the small parties play an offensive game (trying to win over the undecided or unhappy voters by focusing on specific issues).

They found that while the technology could be used to whisper secrets to individuals, the parties mostly used it to shout loudly at broad groups, relying on the power of "team identity" and "us vs. them" narratives to win the election.

In short: The election was a battle of the giants shouting at everyone, while the underdogs tried to win by focusing on specific problems, all while the clock ticked down to the final whistle.

Technical Summary

Technical Summary: Political Advertising on Facebook During the 2022 Australian Federal Election: A Social Identity Perspective

Problem Statement
While research on social media's impact on politics has evolved from early Twitter-based opinion polling to examining precise audience targeting and campaign efficiency, a significant empirical gap remains regarding digital advertising in compulsory voting systems. Most existing literature focuses on voluntary voting contexts where the primary strategic goal is mobilization ("Get Out The Vote"). In contrast, Australia's compulsory voting system ensures high turnout, shifting the strategic focus from mobilization to persuasion. This unique institutional setting requires a distinct analysis of how political actors utilize digital platforms to influence undecided voters and reinforce party loyalty, rather than simply driving turnout. Furthermore, there is a need to interpret these digital strategies through established political psychology frameworks, specifically Social Identity Theory (SIT), to understand the mechanisms of voter motivation in a digital environment.

Methodology
The authors conducted a comprehensive analysis of political advertising on Meta platforms (Facebook and Instagram) during the 2022 Australian federal election (March 1, 2022 – June 1, 2022). The study employed a multi-stage data pipeline:

1. Data Collection: Using the Meta Ad Library API, the researchers collected 56,224 advertisements from 1,549 sponsors. Data was harvested at six-hour intervals, capturing attributes such as creation time, demographic distributions, ad text, sponsor, and spending/impression metrics (calculated as the mean of Meta's provided bounds).
2. Data Preprocessing and Filtering:
   • Keyword Filtering (PAK Dataset): To refine the dataset and exclude non-political content mis-flagged as political, a snowball sampling technique was used to generate a list of 43 relevant keywords (e.g., "government," "election"). This reduced the dataset to 47,503 ads (PAK).
   • Party Affiliation Annotation (PAP Dataset): A political science expert annotated sponsors and page names to identify political party affiliations. A second expert independently annotated 10% of the data, achieving a Cohen's $\kappa$ of 0.906. This resulted in a focused dataset of 30,767 ads (PAP) sponsored exclusively by recognized political parties contesting the election.
3. Analytical Approach: The study utilized descriptive statistics, correlation analysis (Pearson's $R$ and Spearman's $\rho$), and time-series analysis (Mann-Kendall test, KPSS test for stationarity, and first-order differencing) to examine temporal, demographic, and geographical patterns.
4. Theoretical Framework: Findings were interpreted through three lenses of voter motivation: (a) rational self-interest, (b) automatic pilot behavior (bounded rationality), and (c) Social Identity Theory (SIT).

Key Contributions
The paper makes four primary contributions to the field of digital political advertising:

1. Dataset Curation: The creation and release of a manually curated dataset of political advertisements from the 2022 Australian election, including a verified list of party affiliations to support reproducibility.
2. Comparative Analysis: A general analysis of the campaign comparable to international studies, specifically tailored to the unique context of Australia's compulsory voting system.
3. Strategic Insight: A detailed examination of major parties' online strategies, covering quantitative aspects (spending, timing, targeting) and qualitative content themes.
4. Theoretical Application: The application of Social Identity Theory to explain observed disparities in advertising strategies between major and minor parties, offering a psychological interpretation of digital persuasion tactics in a high-turnout environment.

Results

• Dominance and Scale: The Australian Labor Party (ALP) and the Liberal Coalition dominated the landscape, collectively accounting for over 70% of total ad impressions and spending. The ALP spent A$5M (40.2% of total) across 13,047 ads, while the Liberal Coalition spent A$3.9M (31.3%) across 9,364 ads.
• Temporal Patterns: Advertising activity surged immediately after the election call on April 10th, peaking on May 18th, just before the mandated broadcast blackout. Time-series analysis confirmed a significant upward trend in spend and impressions over time.
• Demographic Targeting:
  • Age: Most parties targeted the 18–34 age groups, reflecting Facebook's user base. The Greens specifically focused on the 18–24 demographic, while the United Australia Party (UAP) skewed toward older demographics and showed minimal engagement with the youngest group.
  • Gender: The ALP reached more women, particularly those over 65. The Greens heavily targeted young women (18–24), while the UAP exhibited a male-skewed audience in the 18–54 range.
• Geographical Distribution: Major parties followed national population densities, focusing on eastern states (QLD, NSW, VIC). Smaller parties employed more targeted regional strategies: the Greens concentrated on Victoria, Independents on specific electorates in NSW, and the UAP heavily on Queensland (39.4% of impressions).
• Content and Themes:
  • Major Parties: Focused on self-promotion (party names) and negative campaigning (attacking opponents), aligning with strategies to reinforce existing partisan identities.
  • Minor Parties: Emphasized issue-specific messages. The Greens prioritized climate-related keywords (50% of spend on anti-mining messaging), and Independents leveraged climate crises to differentiate themselves.
• Targeting Granularity: Contrary to assumptions of hyper-micro-targeting, the study found that 84.7% of ads were geographically targeted to a single region, but 85.6% of ads targeted all adult age groups, and only 4.8% were exclusively gender-targeted. This suggests a preference for broad-reach strategies over sophisticated micro-targeting.

Significance and Claims
The paper argues that the observed advertising strategies are best understood through the lens of Social Identity Theory within the context of compulsory voting.

• Strategic Differentiation: Major parties (ALP, Liberal Coalition) employed strategies to reinforce existing "rusted-on" partisan identities and mobilize the in-group against an out-group (negative partisanship). This is consistent with a system where turnout is guaranteed, and the goal is to prevent voter defection.
• Issue-Based Identity: Smaller parties (Greens, Independents) utilized issue-based appeals to cultivate new identity connections among disaffected voters who are obligated to vote but dissatisfied with major parties.
• Broad vs. Micro-Targeting: The prevalence of broad-reach strategies (targeting all ages, specific regions) is interpreted as a logical adaptation to the compulsory voting system. Without the need for micro-targeted mobilization, resources are reallocated toward broad persuasive messaging aimed at the universal electorate.

The authors conclude that while their data reveals strong correlations between party strategy and identity-based appeals, the study does not claim to prove causal effects on voter behavior. They emphasize that the apparent simplicity in targeting may be partly due to API limitations, and future work should incorporate multimodal analysis (images, video) and experimental designs to further validate these theoretical interpretations.