Case Study 23.2: Spanish-Language Media and the Garza-Whitfield Information Gap

Background

The state in which Maria Garza and Tom Whitfield are competing for a Senate seat has a significant and growing Hispanic/Latino population—approximately 19 percent of registered voters, concentrated in three counties with high population density in the southern and western regions of the state. Spanish-dominant and bilingual households constitute an estimated 7 percent of the total registered voter population.

From ODA's perspective, this population presents both a measurement challenge and an analytical imperative. Sam Harding's media monitoring dashboard, as documented in Chapter 23, was built primarily for English-language media. The 200 outlets tracked by the ODA system cover the full range of mainstream English-language media but include only two Spanish-language outlets—the state's dominant Spanish-language television affiliate and one major Spanish-language daily newspaper based in the largest city.

This case study examines what a more complete picture of the Spanish-language media ecosystem reveals about how the Garza-Whitfield race is being covered differently for different communities—and what the implications are for campaign strategy, media analysis, and democratic representation.

The Spanish-Language Media Ecosystem in the State

A research team hired by ODA conducted a six-week audit of the state's Spanish-language media ecosystem. Their findings:

Outlets Identified

Coverage Patterns: Garza-Whitfield

The audit coded all coverage of the Senate race in these outlets during the six-week period. Findings:

Volume: Spanish-language media produced approximately 23% of the per-outlet article/segment volume of English-language media on the race—reflecting both smaller newsroom capacity and different news priorities.

Topics covered: Spanish-language coverage was significantly more likely to discuss immigration policy (86% of race coverage touched on immigration) compared to English-language mainstream media (31%). Spanish-language coverage was significantly less likely to cover horse-race polling (12% vs. 47% in English-language media).

Candidate access: Maria Garza participated in one Spanish-language television interview and one Spanish-language radio call-in show during the audit period. Tom Whitfield's campaign declined requests for Spanish-language media interviews during the six-week window, though one audio clip from a Whitfield rally—in which he made remarks interpreted by some listeners as negative about undocumented immigrants—was widely shared across Spanish-language social media.

Frame analysis: Spanish-language coverage of Whitfield was more consistently negative in tone (79% negative/unfavorable) than English-language coverage of Whitfield (38% negative/unfavorable across outlets with varying partisan leanings). However, the audience of Spanish-language media is not the same as the audience of English-language media—the negative framing in Spanish-language coverage was reaching a voter population already likely to lean toward Garza.

The WhatsApp Dimension

The most analytically challenging component of the audit was the WhatsApp ecosystem. The research team identified 47 WhatsApp groups with some political content related to the state or the Senate race; they estimated combined membership of approximately 34,000 unique accounts, though membership overlapped significantly between groups.

WhatsApp groups presented specific research challenges: - Access: Researchers could only audit groups that community members shared voluntarily; groups were not publicly searchable - Verification: Accounts within groups were not publicly linked to voter file records; demographic estimation was based on participant survey rather than direct observation - Content volume: Content shared in groups ranged from forwarded news articles to original commentary to videos of uncertain provenance - Language: Content was predominantly Spanish but included significant English content, code-switching, and community-specific vocabulary not captured by general Spanish-language sentiment tools

Within the accessible WhatsApp groups, researchers observed heavy sharing of three types of political content during the six-week period: 1. Forwarded articles from Spanish-language news outlets (primarily immigration coverage) 2. Short video clips from both candidates' events—with community commentary that shaped interpretation significantly 3. Content of uncertain origin making factual claims about immigration enforcement, voting procedures, and candidate positions

The third category included both accurate and inaccurate information. Researchers identified three widely-shared pieces of content making specific inaccurate claims: one claiming (falsely) that Whitfield had promised to deport people who had obtained DACA status, one claiming (falsely) that Garza had voted to reduce immigration enforcement funding (she had not served in a legislature and had no such vote), and one containing incorrect information about voter registration deadlines.

Strategic Implications for the Garza Campaign

When ODA presented this audit to the Garza campaign, Nadia Osei identified three distinct strategic questions:

Question 1: Is the current advertising strategy reaching Spanish-dominant voters effectively?

The campaign's media buying had been entirely in English-language television markets and digital placements in English-language news environments. The audit revealed that approximately 7 percent of registered voters were potentially being missed by English-only media buys. In a competitive race expected to be decided by 2-4 percentage points, this segment could be decisive—particularly because the district's geographic concentration of Latino voters placed them in a critical county that was among the race's most contested.

The campaign's digital targeting did include voter file segments identified as likely Hispanic based on census tract and name-based modeling, but the creative content for those segments was English-language. The question of whether English-language digital ads were effectively reaching Spanish-dominant voters was unanswered by campaign data.

Question 2: How should the campaign respond to the WhatsApp misinformation environment?

Two of the three pieces of widely-circulating misinformation—one falsely attributing statements to Whitfield, one falsely attributing a vote to Garza—were both operationally inaccurate and strategically complex. The false Whitfield claim was false in fact but directionally aligned with Garza's actual message; the false Garza claim was false in fact and directly harmful to Garza.

The campaign's options for responding to WhatsApp misinformation were severely constrained by the medium's architecture: content spreads through private networks not accessible to campaigns, fact-checking campaigns work best in public media environments, and attempts to insert campaign messaging into closed WhatsApp groups risked appearing manipulative.

Question 3: Should the campaign invest in Spanish-language media relations and advertising?

The audit documented that Garza's English-language earned media advantage was not translating into Spanish-language earned media—simply because the campaign's communications team had not developed relationships with Spanish-language outlets. Spanish-language journalists had been given less access than English-language reporters, and the campaign's press operation did not have Spanish-language capacity.

Discussion Questions

1. The chapter argues that media monitoring tools systematically undercount the political information environments of non-English-speaking communities. Using this case study as evidence, estimate the magnitude of the error in ODA's initial analysis of the Garza-Whitfield media environment. What specific conclusions from the English-language-only analysis would have been wrong or misleading?

2. The Spanish-language coverage of Whitfield was 79% negative, compared to 38% negative in English-language coverage. Before concluding that Spanish-language media is more biased against Whitfield, what alternative explanations should an analyst consider? How would you investigate these alternatives?

3. The WhatsApp political information ecosystem presents fundamental methodological challenges for media monitoring. For each of the following methodological options, evaluate both what it would reveal and its ethical limitations: - Recruiting Spanish-language community members as paid research informants - Using a research disclosure approach (researchers publicly joining groups, disclosing their purpose) - Using computational text analysis on publicly-accessible Spanish-language social media as a proxy - Conducting survey-based estimation of WhatsApp content exposure

4. Nadia Osei's strategic question about whether to invest in Spanish-language media involves a resource allocation decision. The campaign has limited budget. Design a simple decision framework for how to allocate advertising resources between English-language and Spanish-language media channels. What variables should enter this decision, and what data would you need to make it?

5. The case documents three pieces of circulating misinformation in WhatsApp groups, two of which are false claims that affect the campaign. Using the framing concepts from Chapter 24, analyze each piece of misinformation: what frame does it invoke, and why does that frame make it believable to its intended audience? What would an effective counter-frame strategy look like for each?

6. This case illustrates a broader phenomenon: political campaigns, media analysts, and researchers are systematically better equipped to monitor and engage with political information environments that resemble the English-language, mainstream digital media ecosystem. Write a 200-word reflection on what this systematic gap implies for claims that political analytics is a tool for democratizing political power. Whose political agency is being enhanced by analytics capabilities, and whose is being left behind?

Methodological Note: Research Ethics in Community Information Environments

The audit described in this case study was conducted with approval from an Institutional Review Board and with explicit disclosure to the WhatsApp community administrators who granted access. The research team's protocol included:

• Consent: Only publicly accessible content and content from groups where administrators gave explicit permission was included in the analysis
• Anonymization: Individual users were not identified; only aggregate content patterns were reported
• Community benefit: Preliminary findings, including the misinformation identified, were shared with community organizations before publication to enable community-level responses
• Language accessibility: The final report was produced in both English and Spanish

The researchers noted limitations of these protocols: consent from group administrators may not represent consent from all group members; community organizations receiving preliminary findings may have had conflicts of interest; the research team's Spanish-language capacity was a limiting factor in interpretation. These limitations do not invalidate the research but should inform how its findings are interpreted and applied.