Case Study 25.2: Digital Micro-Targeting and Democratic Accountability — The Whitfield Digital Program

Background

While the Garza-Whitfield air war on television has been extensively documented through commercial ad monitoring and FEC disclosure, the digital advertising programs of both campaigns exist in a substantially less transparent information environment. This case study reconstructs what can be known about the Whitfield campaign's digital advertising program through available transparency sources—and then examines what cannot be known, and why the gap matters.

The case draws on: Meta Ad Library data, Google's political advertising transparency report, FEC expenditure disclosures, Whitfield campaign communications (public-facing), and an analysis conducted by a journalism school's digital advertising research team that accessed targeted ads through a panel of volunteer voters.

What the Transparency Sources Reveal

Meta Ad Library Analysis

A researcher queried the Meta Ad Library for "Tom Whitfield Senate" over the final ten weeks of the race and identified 847 unique advertising creatives across the campaign committee and two aligned PACs. The library disclosed:

Spending ranges (campaign committee only, as required): - Total estimated Facebook/Instagram spend, 10-week period: $280,000–$360,000

Audience targeting disclosures: - Gender distribution: 48% male, 52% female (within platform's stated ranges) - Age distribution: 76% of impressions to users 35–65+ - Geographic targeting: 100% within the state

What the library did NOT disclose: - Specific demographic or interest targeting parameters used to show specific ad versions - Which ads were shown to which audience segments - Custom audience uploads (voter file data matches to platform users) - Lookalike audience parameters - A/B test variants and their respective target audiences

FEC Expenditure Data

FEC filings disclosed the following Whitfield campaign digital advertising expenditures with vendor and purpose information:

Total campaign digital spend disclosed: $703,000. This does not include aligned PAC digital spending.

Note that the FEC description "digital advertising placement" and "digital media placement" reveal that advertising occurred but provide no information about content, targeting, or audience.

The Journalism School Panel Study

A journalism school's advertising transparency research project recruited 340 registered voters in the state as a paid panel. Panelists installed a browser extension that captured political advertising they encountered on social media and the open web during the final eight weeks of the race. Of the 340 panelists, 217 provided at least some data.

Panel demographics: The panel skewed more educated and digitally engaged than the registered voter population (78% college-educated vs. 47% of registered voters). This demographic skew is important for interpreting the data's representativeness.

Ads captured by the panel (Whitfield-related, n=834 ad instances across 127 unique creatives):

By message type: - Crime-focused attack on Garza: 31% - Economic anxiety (general): 22% - Immigration enforcement: 18% - Whitfield biography/positive: 16% - Cultural values (anti-"woke," parental rights): 9% - Other: 4%

By audience segment (inferred from which panelists saw which ads): - Crime attack ads: concentrated among men 45+, mixed partisan registration - Immigration ads: concentrated among men 55+, Republican-leaning registration - Economic anxiety ads: broad demographic distribution - Cultural values ads: concentrated among women 35-55, Republican or no-party registration, inferred suburban - Biography ads: broad distribution

The Targeting Analysis: What the Data Reveals

When the journalism school researchers analyzed which panelists saw which types of advertising, several patterns emerged:

Finding 1: Issue-Specific Demographic Targeting

Immigration-enforcement advertising was concentrated among panelists with demographic characteristics that appear to correlate with voter file data signals: older, male, Republican-leaning. Cultural values advertising was concentrated among a different demographic: women in the 35-55 age range who registered as either Republican or no-party. This is consistent with campaign micro-targeting organized around voter file segments with different issue priorities.

If the campaign is delivering immigration messaging primarily to men over 55 and cultural values messaging primarily to suburban women 35-55, it is having two distinct conversations about its candidate's identity with different voter segments—conversations that those segments cannot see each other having.

Finding 2: The Absent Audience

Panel analysis revealed that several significant demographic groups were rarely or never seeing Whitfield campaign advertising of any type:

• Panelists in precincts with >60% non-white voter registration: 94% reduction in Whitfield ad impressions compared to white-majority precincts
• Panelists who were Democratic party registrants: 89% reduction compared to independent and Republican registrants
• Panelists with modeled low turnout propensity: 67% reduction compared to high-propensity voters

The campaign appears to be largely (though not entirely) excluding non-white majority precincts, Democrats, and low-propensity voters from its advertising universe. The first of these findings—the concentration of ad exclusions in non-white majority precincts—is the most ethically significant.

Finding 3: Message Inconsistency Across Audiences

Two panelists received contradictory implicit signals about Whitfield's economic positions. A panelist in a manufacturing-heavy county saw advertising emphasizing Whitfield's support for local industry and manufacturing jobs. A panelist in a suburban area with many white-collar workers saw advertising emphasizing Whitfield's opposition to "government overreach" in business regulation. These are not technically contradictory messages—a politician can be pro-manufacturing and anti-regulation simultaneously—but the implicit framing of the economy in each case is different in ways that matter to different voter constituencies.

Finding 4: The Escalation Pattern

Tracking ad types by week, researchers found that immigration-focused advertising increased substantially in the race's final three weeks, concentrated heavily in specific suburban county segments identified through the voter file. This escalation coincided with a national news cycle about immigration enforcement that the Whitfield campaign appears to have exploited—increasing advertising that exploited issue salience generated by events outside the campaign's control.

What Cannot Be Known

The journalism school study represents the most detailed public analysis of the Whitfield digital program available, but even it leaves significant gaps:

1. The targeting parameters are inferred, not documented. Researchers can observe that immigration ads were concentrated among older Republican-leaning men, but they cannot see the actual targeting parameters the campaign used. The campaign may be targeting by voter file segments, platform demographic signals, behavioral interests, or some combination. The distinction matters for legal and ethical evaluation.

2. The panel underrepresents key populations. A panel that is 78% college-educated systematically underrepresents non-college voters, lower-income voters, and communities of color relative to the registered voter population. The most analytically significant question—what were non-white majority precinct voters actually seeing?—is precisely the question the panel is worst positioned to answer.

3. Cross-platform visibility is incomplete. The panel captured open web programmatic advertising and social media, but not CTV, streaming audio, YouTube mobile app, or SMS/MMS advertising. The campaign's programmatic buys may have included significant advertising on platforms not captured by the browser extension tool.

4. Outside group advertising is separate. The aligned PACs are running their own digital programs with potentially different targeting parameters and messaging. The FEC and Meta Ad Library data on outside groups is less detailed than on the campaign committee.

The Democratic Accountability Question

This case study's data enables a direct examination of the democratic accountability concerns raised in Chapter 25. Consider the following specific accountability issues:

Accountability Issue 1: Message consistency. Tom Whitfield is running for a Senate seat that requires representing all of the state's voters, including those in non-white majority precincts. If those precincts are systematically excluded from the advertising audience, those voters have less information about what Whitfield is promising, less opportunity to respond to those promises, and less ability to hold him accountable if elected. The accountability relationship between candidate and constituent is attenuated by the targeting exclusion.

Accountability Issue 2: Message documentation. In traditional broadcast advertising, everything a campaign says publicly is documented, monitorable by journalists, and available for fact-checking. When a campaign delivers immigration advertising to specific demographic segments through digital channels, the content of those messages reaches only the targeted audience. Fact-checkers cannot monitor what they cannot see; journalists cannot report on messages delivered through micro-targeted digital channels unless they are members of the targeted segment.

Accountability Issue 3: Racial exclusion. The 94% reduction in ad impressions to panelists in non-white majority precincts is not direct evidence of racially discriminatory targeting—the targeting may be based on partisan registration or turnout propensity, which happen to correlate with racial composition. But the effect is functionally the same: political information is being systematically withheld from non-white communities, reducing their ability to make informed electoral decisions.

Discussion Questions

1. The journalism school's panel study is the most comprehensive transparency analysis available for this campaign, but it has documented demographic limitations. Specifically: the panel is 78% college-educated while the registered voter population is 47% college-educated. For each of the three major findings (issue-specific demographic targeting, absent audience, message inconsistency), assess how seriously the panel's demographic skew affects the reliability of that finding. Are some findings more robust to this limitation than others?

2. The chapter distinguishes between targeting by partisan registration/turnout propensity (which produces racial disparities as a side effect) and explicitly racially discriminatory targeting. Is this distinction legally and ethically meaningful in your view? If a campaign explicitly excludes "likely non-supporters" from its targeting, and likely non-supporters happen to be concentrated in non-white majority precincts, what obligations does the campaign have to ensure its targeting is not functioning as racial exclusion?

3. The campaign is running different economic messages to manufacturing-area voters and suburban white-collar voters. Using the framing concepts from Chapter 24, analyze whether this represents legitimate audience adaptation (emphasizing different aspects of a consistent position for different audiences) or accountability-compromising message inconsistency. Where is the line?

4. Design a political advertising transparency regime that would address the accountability gaps identified in this case study. Specifically: what additional disclosures should platforms be required to make? What information should campaigns be required to disclose beyond current FEC requirements? Be specific about what information would need to be disclosed and what process would enable the disclosure.

5. The Finding 4 escalation of immigration advertising in the race's final three weeks coincided with a national news cycle that made immigration salient. The campaign appears to have increased advertising to exploit this externally-generated salience. Using the priming concepts from Chapter 24, analyze the relationship between the external news cycle, the campaign's strategic response, and the ultimate priming effect on voters. What is the campaign doing that differs from simply "being responsive to voter concerns"?

6. Nadia Osei's Garza campaign has conducted its own digital program with its own targeting choices. A journalist asks her: "Is the Garza campaign also excluding certain communities from its digital advertising?" How should she respond? Draft a 150-word response that is both honest and strategically sensible for the campaign's position.