Case Study 13-2: The Asian American Electorate — The Most Underanalyzed Community in the State

The Situation

Six weeks before election day in the Garza-Whitfield race, Nadia Osei received a pointed memo from one of Garza's major fundraisers — a tech entrepreneur named Jonathan Kwan, who was Filipino-American and deeply involved in the local South and Southeast Asian professional community.

The memo was brief but direct:

"Your campaign has done zero targeted outreach to the Asian American community in this state. This community is 12% of the population and skews high-propensity. You are leaving votes on the table. I've been talking to community leaders and there's real interest in Garza, but also real frustration that nobody from the campaign has called. I need to understand what your plan is."

Nadia pulled up the state demographic data and realized Kwan had a point — but it was more complicated than he knew.

The Data Problem

The state's Asian American population of approximately 12% masked enormous internal heterogeneity. ODA's voter file analysis, which Nadia had on her desk, broke it down:

Estimated Asian American voter registration by national-origin community (major groups): - Filipino-American: ~31% of registered Asian American voters in the state - Vietnamese-American: ~24% - Indian-American: ~18% - Chinese-American: ~14% - Korean-American: ~6% - Other/multiple: ~7%

The political behavior data — assembled by ODA from precinct-level analysis of heavily Asian American precincts and exit poll fragments — was fragmentary but suggestive:

Estimated Democratic vote share by community in the most recent statewide race: - Indian-American: ~72% Democratic - Filipino-American: ~61% Democratic - Chinese-American: ~58% Democratic - Korean-American: ~54% Democratic - Vietnamese-American: ~44% Democratic

The Vietnamese-American figure was particularly striking: while other Asian American communities leaned clearly Democratic, Vietnamese-Americans in this state were roughly split, with a history of Republican alignment driven by older voters' connections to the anti-communist South Vietnamese cause and a newer generation that was more politically mixed.

Nadia's campaign data scientist, Marcus Adeleke, ran a quick calculation: if the campaign ran targeted Asian American outreach treating the community as uniform (a single D+20 group), they would over-invest in reaching Vietnamese-American voters where the expected persuadable-to-voted ratio was low, while potentially under-investing in Indian-American and Filipino-American communities where Democratic enthusiasm was high but turnout was uneven.

The Measurement Challenge

There was a deeper problem: the data itself was thin.

The state's election administration did not systematically track voter race/ethnicity — voters self-registered without providing race information, so the racial composition of the voter file had to be inferred. ODA used a probabilistic name-based methodology (assigning race/ethnicity probabilities based on surname and geography) combined with census block demographics. This methodology was imperfect: it worked reasonably well for Spanish-surnamed voters (high coverage, clear surname patterns) and moderately well for common Asian surname patterns, but it struggled with Vietnamese names that were less commonly represented in its training data, with multiracial individuals who had non-ethnic surnames, and with recent immigrants whose names didn't match the training data's expected patterns.

Marcus flagged this in a campaign analytics meeting: "Our Vietnamese-American voter file data is probably the least reliable of any demographic group in the state. The name-matching algorithm performs about 15-20% worse on Vietnamese names than Filipino or Indian names. We might be underestimating their Democratic lean, or overestimating it — we honestly don't know."

This was a case where measurement uncertainty was directly consequential for strategic decisions.

What the Campaign Did — and Didn't Know

The Garza campaign had done something many campaigns skip: a small focus group series in the large Asian American professional community in the major metro area. The sessions — conducted in English, with separate sessions for Filipino-American and Indian-American participants — yielded several important insights that didn't show up in the voter file data.

Healthcare was the top issue. In both communities, participants cited healthcare costs and access as their primary concern — not immigration policy, not trade, not cultural identity issues. Indian-American participants specifically mentioned the H-1B visa backlog, but ranked it below healthcare and education as an electoral priority.

Garza was relatively well-known and positively regarded — more so than Whitfield, who many participants had never heard of before the campaign started.

The ask was missing. Multiple participants noted that they had received no campaign communication — no mailers, no phone calls, no digital ads in relevant languages (Tagalog, Hindi, Mandarin). "I'm going to vote for Garza," one Filipino-American participant said, "but I'm going to do it on my own. Nobody's asking me to."

This was the specific form of the self-fulfilling prophecy that Adaeze at ODA had written about: communities that weren't being targeted were not unmotivated, they were simply uncontacted. Their turnout being lower than their registration rates suggested wasn't about enthusiasm — it was about the absence of mobilization infrastructure.

Nadia's Response to Kwan

Nadia wrote back to Jonathan Kwan three days after receiving his memo. Her response laid out the campaign's constraints honestly:

"Jonathan — you're right, and I hear you. Here's where we are:

The honest answer is that our data on the Asian American community in this state is thinner than on almost any other group, which has made it hard to build a targeted outreach strategy with confidence. The community is also genuinely diverse — our Vietnamese-American data suggests a more competitive picture than our Filipino and Indian data — and we've been uncertain about how to calibrate messaging across a community this heterogeneous.

That said: your point about turnout stands. We have good reason to believe that targeted, culturally competent outreach would increase Democratic turnout in Filipino-American and Indian-American communities specifically without requiring a lot of persuasion — these voters lean our way already.

Here's what I'm proposing: we stand up a specific Asian American outreach operation for the final six weeks. We're going to need your help identifying community validators, event venues, and language-competent staff. We can commit $180,000 to this effort. We're going to prioritize Filipino-American and Indian-American communities in the metro area, where our data is strongest and the expected yield is highest."

Kwan's response came within two hours: "That's a start. But I want to talk about Vietnamese-American community organizers I know who think the picture there is more Democratic than your data shows. Can we set up a call?"

The Analytic Epilogue

After the election, ODA conducted a post-election analysis of Asian American voting patterns in the state. Their conclusions:

The campaign's decision to prioritize Filipino-American and Indian-American outreach over Vietnamese-American outreach was probably correct, but for partly the wrong reasons. The Vietnamese-American community did skew more Democratic than the pre-election data suggested — ODA's post-election analysis using post-election surveys and census block correlations estimated the Vietnamese-American Democratic lean at approximately D+12, not the nearly even split the name-based probabilistic matching had suggested.

The name-matching methodology had systematically undercounted Democratic-leaning younger Vietnamese-American voters who had more anglicized or mixed surnames and lived in mixed-demographic census blocks where the algorithm assigned lower Vietnamese probability.

More broadly, ODA's post-election report noted: "Asian American voters remain the group most likely to be misidentified in voter files, most likely to be absent from campaign targeting lists, and most likely to lack culturally specific campaign communication. This represents a persistent equity gap in political representation that no campaign-level fix can fully address — it requires investment in voter file infrastructure at the community and civic organization level."

Discussion Questions

1. The Data Quality Problem

The campaign's data on Vietnamese-American voters was systematically less reliable than data on other groups due to the limitations of name-based probabilistic matching. This is an example of how measurement methodology can produce inequitable outcomes — some communities are measured better than others, and those measured poorly receive less campaign attention. What alternative data collection approaches might improve the accuracy of racial/ethnic voter identification? What are the privacy and equity concerns with each alternative?

2. Heterogeneity Within the Category

The "Asian American" category in this case concealed five distinct national-origin communities with meaningfully different political behavior patterns. Generalizing across them produced a misleading average. When is it appropriate to use an aggregate demographic category in political analysis, and when does aggregation actively mislead? What criteria should guide the choice?

3. The Self-Fulfilling Prophecy in Practice

ODA's post-election analysis found that the Vietnamese-American community leaned more Democratic than pre-election data suggested — partly because the young, more Democratic voters were systematically harder to identify in the voter file. If the campaign had targeted Vietnamese-American communities more aggressively, they might have found more Democratic voters than their data predicted. How should a campaign think about data uncertainty of this type? Is it rational to invest in communities where the data is worst, knowing that the data might be wrong in your favor?

4. The Resource Allocation Decision

Nadia committed $180,000 to Asian American outreach in the final six weeks. Given the data uncertainties described in this case, evaluate her resource allocation decision. Was $180,000 too little given the community's size and political potential? Too much given the data uncertainty? How would you have approached this decision differently, and what additional information would have been most valuable?

5. Structural vs. Campaign Solutions

ODA's post-election report concluded that the Asian American data gap is "a persistent equity gap that no campaign-level fix can fully address." What structural solutions — beyond any single campaign's decisions — might improve the representation of Asian American and other underrepresented communities in political data infrastructure? Who is responsible for building those solutions, and what incentives or funding models would support that work?