How South Asian Americans voted, 2016 to 2024

Two-party share is the Democratic candidate's votes as a percentage of Democratic plus Republican votes. Third-party voters were 2 to 6% of South Asian voters in these surveys.In the 2016 National Asian American Survey (NAAS), 86.0% of South Asian American voters who chose between Hillary Clinton and Donald Trump chose Clinton (90% interval 82.4 to 89.6) [1]. The Collaborative Multi-racial Post-election Survey (CMPS) put the Democratic share at 86.0% in 2016 and 80.2% in 2020 [2, 3]. Other Asian American voters in the same three surveys gave the Democrat 72.9%, 74.2% and 69.7%. Nationally, Clinton won 51.1% of the two-party vote and Joe Biden 52.3% [4].

South Asian here means a respondent who gives Indian, Pakistani or Bangladeshi as their primary ethnicity or ancestry. NAAS also has codes for Nepali, Sri Lankan and Bhutanese respondents, but none appear in the post-election file. The CMPS public files mask the write-in answers. Indo-Caribbean voters cannot be identified in any of these surveys.The national exit poll and AP VoteCast ask voters their race but not their national origin, so they report Asian American voters as a single group. The previous chapter measured how often South Asian citizens vote, using the Census Bureau's turnout survey, which does not ask about vote choice. This chapter uses the surveys that do ask both. All of them are self-reports from samples of a few hundred South Asian voters, so each estimate below carries a 90% interval.

Three surveys identify South Asian voters

The NAAS and the two CMPS waves ask Asian American respondents for their national origin and include large enough South Asian samples to estimate their vote: 728 voters in the 2016 NAAS, 162 in the 2016 CMPS and 452 in the 2020 CMPS. NAAS interviewed by telephone from a listed sample drawn by race and national origin, and it sampled Pakistani and Bangladeshi Americans as separate groups (320 and 320 respondents in the public file). The CMPS is an online survey that recruits registered voters from voter-file email lists and panels and weights each racial group to the American Community Survey. The 2020 CMPS was fielded from April to August 2021, five to nine months after the election.

The Pew Research Center's 2022–23 survey of Asian Americans has a probability sample but no vote question, and its public file separates only Indian Americans among South Asians [5]. AP VoteCast has thousands of Asian American voters but no national-origin question [6, 7]. Both appear below as benchmarks.

Figure 1

Survey files used in this chapter

Fielding dates, mode and sample of each survey, how it identifies South Asian respondents, and the unweighted counts of South Asian respondents and voters

Table of the six survey files used in this chapter, with South Asian respondents and South Asian voters naming the Democrat or Republican: NAAS 2016 post-election: 1,151 respondents, 728 voters; CMPS 2016: 541 respondents, 162 voters; CMPS 2020: 881 respondents, 452 voters; Pew Survey of Asian Americans: 446 respondents, no vote question; AP VoteCast 2020: South Asian respondents not identified, 885 Asian voters; AP VoteCast 2024: South Asian respondents not identified, 1,609 Asian voters.

Survey	Fielded	Mode and sample	How South Asians are identified	South Asian respondents	South Asian voters naming Democrat or Republican
NAAS 2016 post-election	Nov 2016 to Mar 2017	Telephone; listed sample stratified by race and national origin	Primary ethnicity: Indian, Pakistani, Bangladeshi	1,151	728
CMPS 2016	Dec 2016 to Feb 2017	Online; voter-file email sample plus opt-in panels	Primary ethnicity or ancestry: Indian, Pakistani, Bangladeshi	541	162
CMPS 2020	Apr to Aug 2021	Online; voter-file email sample plus panels matched to voter files	Primary ethnicity or ancestry: Indian, Pakistani, Bangladeshi	881	452
Pew Survey of Asian Americans	Jul 2022 to Jan 2023	Mail to web or paper; address sample plus surname lists	Indian only (public file)	446	No vote question
AP VoteCast 2020	Oct 26 to Nov 3, 2020	Web and phone; voter-file sample plus probability and opt-in panels	Not identified (Asian only)	–	885 Asian
AP VoteCast 2024	Oct 28 to Nov 5, 2024	Web and phone; voter-file sample plus probability and opt-in panels	Not identified (Asian only)	–	1,609 Asian

South Asian voters were more Democratic than other Asian American voters in each survey

The South Asian margin over other Asian American voters was 13.1 points in the 2016 NAAS (90% interval 6.6 to 19.6), 11.7 points in the 2016 CMPS (5.8 to 17.6) and 10.4 points in the 2020 CMPS (5.9 to 14.9). Both groups voted more Democratic than the national electorate.

Between the two CMPS waves, which used similar recruitment and weighting, the South Asian Democratic share fell by 5.8 points (90% interval −11.9 to +0.3), and the other Asian American share fell by 4.5 points (−8.5 to −0.5). Biden's national share was 1.2 points higher than Clinton's. The 2020 CMPS and AP VoteCast give similar estimates for Asian American voters as a whole: 72.1% and 71.4% for Biden.

Estimates of the 2024 Asian American vote differ by survey. AP VoteCast gives Harris 66.8% and Trump 30.5% of Asian American voters. Pew's study of voters matched to voter files gives Harris 57% and Trump 40%, up from 30% for Trump in 2020 [8].For 2024, VoteCast puts Kamala Harris's share of the Asian American two-party vote at 68.7%, 2.7 points below Biden's 2020 share in the same survey (90% interval for the change −8.0 to +2.6). VoteCast cannot say whether South Asian voters moved more or less than other Asian Americans.

Figure 2

South Asian voters gave the Democrat 10 to 13 points more of the two-party vote than other Asian American voters in 2016 and 2020

Democratic candidate's share of the two-party presidential vote among self-reported voters, with 90% intervals. The dashed tick in each row is the national result.

• South Asian voters
• Other Asian American voters
• All Asian American voters (VoteCast)
• All voters nationally

Show data

Pakistani and Bangladeshi voters were the most Democratic

Only the 2016 NAAS has enough Pakistani and Bangladeshi voters to estimate each group separately. Clinton won 94.7% of the two-party vote among Pakistani American voters and 95.5% among Bangladeshi American voters, against 84.6% among Indian American voters. The combined Pakistani and Bangladeshi share was 10.4 points above the Indian share (90% interval 5.9 to 14.9). In the 2020 CMPS the 65 Pakistani and Bangladeshi voters gave Biden 92.2%, 14.5 points more than Indian American voters (7.2 to 21.8).

Origin and religion overlap. Most Pakistani and Bangladeshi respondents are Muslim, and the Muslim category also includes Indian American Muslims. These surveys are too small to separate the two.Religion shows a similar split in 2016. Muslim South Asian voters in NAAS gave Clinton 96.9% and Hindu South Asian voters 83.6%, a difference of 13.3 points (7.8 to 18.8). In the 2020 CMPS the difference was 4.7 points (−5.8 to 15.2), with 82.3% for Muslim voters and 77.6% for Hindu voters, too close to distinguish with 77 Muslim respondents.

Figure 3

Pakistani and Bangladeshi American voters gave the Democrat an estimated 95% of the two-party vote in 2016 and 92% in 2020

Democratic share of the two-party presidential vote among South Asian voters by national origin and religion, with 90% intervals. Rows with fewer than 50 voters are not shown.

• NAAS 2016 (Clinton)
• CMPS 2020 (Biden)

Show data

The gap with other Asian Americans appears in every age, sex, education and birthplace group

The 2020 CMPS is large enough to compare groups within the South Asian electorate. South Asian women gave Biden 84.6% of the two-party vote and men 76.6%, a difference of 8.0 points (90% interval 0.7 to 15.3). Voters born in the US gave him 83.1% and immigrants 79.4%, a difference the survey cannot distinguish from zero (−3.5 to 10.9). By age, the share ran from 83.3% among voters under 40 to 79.0% among those 60 and over.

In every one of these groups the South Asian estimate was higher than the estimate for other Asian American voters, though the 90% intervals overlap for men, voters under 40, US-born voters and voters without a bachelor's degree. The difference was largest among immigrants: 79.4% against 65.0% for immigrant voters of other Asian origins, compared with 83.1% against 76.5% among the US-born. In the sample, 88% of South Asian voters had a bachelor's degree, so the 53 without one give only a rough estimate.

Figure 4

South Asian voters' Biden share was higher than other Asian American voters' in every age, sex, education and birthplace group in 2020

Biden's share of the two-party vote among self-reported voters in the 2020 CMPS, with 90% intervals.

• South Asian voters
• Other Asian American voters

Show data

Party identification shows the same gap

Party identification gives a fourth source, the Pew survey, which interviewed Asian Americans in 2022 and 2023. Among registered Indian American citizens, 72.4% of those who identified with or leaned toward a party chose the Democrats (90% interval 64.2 to 80.6), against 62.9% of all other registered Asian Americans. Republicans and Republican leaners were 27.1% of registered Indian Americans and 35.6% of other registered Asian Americans.

The South Asian Democratic margin over other Asian Americans was 9.5 to 14.0 points across the four surveys. The levels are not comparable from one survey to the next. The NAAS telephone interview let respondents say they do not think in terms of parties, and 24.2% of South Asian registered voters gave no party or lean; in Pew the figure was 1.7%.

Figure 5

South Asian registered voters were 9.5 to 14.0 points more Democratic in party identification than other Asian Americans in each survey

Democrats and Democratic leaners as a share of registered voters who identify with or lean toward a party, with 90% intervals. Pew's public file identifies Indian Americans only; its comparison group is all other Asian Americans.

• South Asian (Pew: Indian)
• Other Asian American

Show data

Evidence on 2024 comes from pre-election polls of Indian Americans

The 2024 report's summary gives Trump 32% among registered voters; its results section gives 31%, used here.No public post-election microdata identify South Asian voters in 2024. The closest evidence is the Carnegie Endowment's Indian American Attitudes Survey, an online poll of Indian Americans that the same researchers ran with YouGov in 2020 and 2024. In September 2020, 72% of registered Indian American voters intended to vote for Biden and 22% for Trump [9]. In September and October 2024 the figures were 61% for Harris and 31% for Trump [10]. As two-party shares, that is a fall from 76.6% to 66.3%, or 10.3 points.

In 2020 the poll's two-party figure was close to what Indian American voters reported after the election in the CMPS (77.7%). The 10-point fall is larger than VoteCast's 2.7-point fall among all Asian American voters, though the two surveys differ in design and timing. Neither survey covers Pakistani or Bangladeshi Americans, and neither does the 2024 Asian American Voter Survey, which was fielded before Biden withdrew [11].

Figure 6

Published estimates of Indian American presidential preference, 2016 to 2024

Democratic and Trump shares as published by each source, the Democratic share of the two-party vote computed from them, and respondents

Table of five estimates of Indian American presidential preference: NAAS 2016 post-election report, Democrat 77%, Trump 16%, two-party Democratic 82.8%; Indian American Attitudes Survey 2020, Democrat 72%, Trump 22%, two-party Democratic 76.6%; CMPS 2020, this chapter, Democrat 76%, Trump 22%, two-party Democratic 77.7%; Asian American Voter Survey 2024, Democrat 46%, Trump 29%, two-party Democratic 61.3%; Indian American Attitudes Survey 2024, Democrat 61%, Trump 31%, two-party Democratic 66.3%.

Source (Indian Americans)	Fielded	Population	Measure	Democrat, %	Trump, %	Democratic, % of two-party	Respondents
NAAS 2016 post-election report	November 2016 to March 2017	Voters	Reported vote, Clinton vs Trump	77	16	82.8	504
Indian American Attitudes Survey 2020	September 1 to 20, 2020	Registered voters	Vote intention, Biden vs Trump	72	22	76.6	842
CMPS 2020, this chapter	April to August 2021	Voters	Reported vote, Biden vs Trump	76	22	77.7	398
Asian American Voter Survey 2024	April 4 to May 26, 2024	Registered voters	Vote intention, Biden vs Trump (14% did not know)	46	29	61.3	447
Indian American Attitudes Survey 2024	September 18 to October 15, 2024	Registered voters	Vote intention, Harris vs Trump	61	31	66.3	700

What these surveys cannot show

All of these figures are self-reports. Respondents who did not vote may say they did, and the 2020 CMPS asked about the election months after it happened. In the NAAS, 17.2% of South Asian respondents who said they voted declined to name a candidate, against 11.5% of other Asian American voters; if those who declined leaned Republican, the Democratic shares here are too high.

The estimates also depend on the weights. The CMPS draws much of its sample from opt-in panels and relies on weighting to the American Community Survey. Weighting moves the 2016 CMPS South Asian estimate from 77.2% unweighted to 86.0%, and the 2020 estimate from 83.6% to 80.2%. The NAAS weights are very uneven within the South Asian sample, so its 728 South Asian voters carry the precision of about 165 equally weighted respondents. The intervals capture sampling error and the variability of the weights. They do not capture error from how opt-in samples are recruited, so they are lower bounds.

The group definition leaves out South Asian Americans whom these surveys cannot classify, including Nepali, Sri Lankan and Indo-Caribbean voters, and respondents of mixed origin who gave another ethnicity as primary. A post-election estimate for 2024 would need the CMPS 2024 file, which asks Asian American respondents their national origin. It is available only to the study's contributors until a public release through ICPSR planned for December 2026 [12].

Figure 7

Weighting changes the South Asian estimates by up to 9 points

Democratic share of the two-party presidential vote among voters under the main weight, the survey's alternative weight, and no weight.

Table of the Democratic two-party share under three weightings: 2016 NAAS, South Asian: main weight 86.0 (PIDADJWEIGHT), alternative weight 86.1 (NWEIGHTNATIVITY), unweighted 88.0; 2016 NAAS, Other Asian American: main weight 72.9 (PIDADJWEIGHT), alternative weight 73.0 (NWEIGHTNATIVITY), unweighted 71.4; 2016 CMPS, South Asian: main weight 86.0 (NAT_WEIGHT), alternative weight 86.0 (WEIGHT), unweighted 77.2; 2016 CMPS, Other Asian American: main weight 74.2 (NAT_WEIGHT), alternative weight 74.2 (WEIGHT), unweighted 68.7; 2020 CMPS, South Asian: main weight 80.2 (WEIGHT), alternative weight 79.8 (OS_WEIGHT), unweighted 83.6; 2020 CMPS, Other Asian American: main weight 69.7 (WEIGHT), alternative weight 69.1 (OS_WEIGHT), unweighted 75.2.

Survey and group (voters)	Main weight	Alternative weight	Unweighted	Effective sample size
2016 NAAS, South Asian	86.0 (PIDADJWEIGHT)	86.1 (NWEIGHTNATIVITY)	88.0	165 of 728
2016 NAAS, Other Asian American	72.9 (PIDADJWEIGHT)	73.0 (NWEIGHTNATIVITY)	71.4	88 of 1,892
2016 CMPS, South Asian	86.0 (NAT_WEIGHT)	86.0 (WEIGHT)	77.2	91 of 162
2016 CMPS, Other Asian American	74.2 (NAT_WEIGHT)	74.2 (WEIGHT)	68.7	441 of 1,094
2020 CMPS, South Asian	80.2 (WEIGHT)	79.8 (OS_WEIGHT)	83.6	351 of 452
2020 CMPS, Other Asian American	69.7 (WEIGHT)	69.1 (OS_WEIGHT)	75.2	1,127 of 1,886

Appendix tables

The published shares are of all NAAS voters naming a candidate; "Asian American" in the published report [13] also includes groups this chapter excludes. The public file has 358 Indian voters naming any candidate, of whom 331 named Clinton or Trump.

Figure 8

The NAAS public file reproduces the published 2016 vote by origin

Clinton, Trump and other shares of NAAS voters naming a candidate, as published by the producers and as computed from the public file

Table comparing published and computed NAAS 2016 vote shares: Asian American: Clinton 69 published and 71.5 in the file, Trump 25 and 22.6, other 6 and 5.9; Indian: Clinton 77 published and 79.3 in the file, Trump 16 and 14.4, other 7 and 6.2; Pakistani: Clinton 88 published and 88.3 in the file, Trump 5 and 4.9, other 7 and 6.8; Bangladeshi: Clinton 90 published and 89.7 in the file, Trump 4 and 4.2, other 6 and 6.1.

NAAS 2016 voters	Clinton, published	Clinton, this file	Trump, published	Trump, this file	Other, published	Other, this file
Asian American	69	71.5	25	22.6	6	5.9
Indian	77	79.3	16	14.4	7	6.2
Pakistani	88	88.3	5	4.9	7	6.8
Bangladeshi	90	89.7	4	4.2	6	6.1

---

Data sources

• National Asian American Survey 2016, post-electionNAAS 2016 post-election survey, public-use file, from ICPSR [1]Years: 2016 · Weights: PIDADJWEIGHT, the producers' weight adjusted for party differences in re-interview cooperation
• Collaborative Multiracial Post-Election Survey 2016CMPS 2016 public-use file, from ICPSR [2]Years: 2016 · Weights: NAT_WEIGHT
• Collaborative Multiracial Post-Election Survey 2020CMPS 2020 public-use file, from ICPSR [3]Years: 2020 · Weights: WEIGHT, the primary sample weight
• Pew Research Center Survey of Asian Americans, 2022 to 2023Pew Research Center 2022–23 Survey of Asian Americans, public-use file [5]Years: 2022 to 2023 · Weights: P_RAKEDW0, with its jackknife replicate weights
• AP VoteCastAP VoteCast 2020 and 2024 general election public-use files [6, 7]Years: 2020 and 2024 · Weights: FINALVOTE_NATIONAL_WEIGHT
• Federal Election Commission, official election resultsnational popular vote from the Federal Election Commission [4]Years: 2016, 2020 and 2024 · Weights: full counts

Methods

Groups. South Asian respondents gave Indian, Pakistani or Bangladeshi as their primary ethnicity or ancestry (NAAS RETHNIC; CMPS S8). Pew's public file identifies Indian Americans only (ASIANID_MOD_SUM_1_RECODE = Indian only). Other Asian American respondents named any other Asian origin; Pacific Islander, Iranian, write-in and refused answers are excluded from both groups.

Voters and vote choice. Voters said they voted in the presidential election (NAAS Q2_4; CMPS 2016 C12; CMPS 2020 Q12, "certain" or "think I voted"). The 2020 CMPS asked every respondent the presidential question, including non-voters, so the voter filter matters: without it the South Asian estimate is 82.9%. VoteCast voters follow AP's likely-voter classification. Shares are of voters naming the Democratic or Republican candidate; refusals and "don't know" are excluded.

Party identification. Identifiers plus leaners among self-reported registered voters (NAAS Q10_0A with Q10_0D/Q10_0E; CMPS 2016 C25/C27; CMPS 2020 Q21/Q23; Pew PARTYSUM among citizens certain or probably registered).

Weights. NAAS PIDADJWEIGHT, the producers' weight adjusted for differences by party in re-interview cooperation; CMPS 2016 NAT_WEIGHT; CMPS 2020 WEIGHT, the primary sample weight; Pew P_RAKEDW0; VoteCast FINALVOTE_NATIONAL_WEIGHT. With those weights VoteCast reproduces the national two-party result (52.2% in 2020 and 49.2% in 2024), and the NAAS public file reproduces the producers' published 2016 vote by origin within 2.5 points (see the appendix tables).

Uncertainty. Intervals are 90% (1.645 standard errors). Pew intervals use its 100 JK2 replicate weights. The other surveys have no replicate weights, so intervals come from a respondent-level Poisson bootstrap of the weighted estimate (1000 replicates). Differences within a survey use the same replicates for both groups; changes between CMPS waves treat the samples as independent. Cells with fewer than 50 respondents are not shown.

Reproduction. south-asian-voters/chapters/03-vote-choice/build_data.py reads the zipped survey files in data/asian-american/raw, computes every survey number and writes data.json and claims.json; build_page.py assembles this page and holds the published figures quoted from other sources.

References

1. Ramakrishnan, S. K.; Lee, J.; Lee, T.; Wong, J. (2020). National Asian American Survey (NAAS) Post-Election Survey, [United States], 2016 (ICPSR 37380, version 1) [dataset]. Inter-university Consortium for Political and Social Research. doi:10.3886/ICPSR37380.v1
2. Frasure, L.; Wong, J.; Vargas, E.; Barreto, M. (2022). Collaborative Multi-racial Post-election Survey (CMPS), United States, 2016 (ICPSR 38040, version 2) [dataset]. Inter-university Consortium for Political and Social Research. doi:10.3886/ICPSR38040.v2
3. Frasure, L.; Wong, J.; Vargas, E.; Barreto, M. (2024). Collaborative Multi-racial Post-election Survey (CMPS), United States, 2020 (ICPSR 39096, version 1) [dataset]. Inter-university Consortium for Political and Social Research. doi:10.3886/ICPSR39096.v1
4. Federal Election Commission (2025). Federal Elections 2016 (December 2017) and Federal Elections 2020 (October 2022); Official 2024 Presidential General Election Results (January 2025). Federal Election Commission. 2016: https://www.fec.gov/documents/1890/federalelections2016.pdf; 2020: https://www.fec.gov/documents/4227/federalelections2020.pdf. www.fec.gov/resources/cms-content/documents/2024presgeresults.pdf
5. Pew Research Center (2023). 2022–23 Survey of Asian Americans, public-use file, codebook and methodology report (fielded July 5, 2022 to January 27, 2023) [dataset]. Pew Research Center. First report: Ruiz, N. G., Noe-Bustamante, L., and Shah, S. (2023), Diverse Cultures and Shared Experiences Shape Asian American Identities, Pew Research Center, May 8, 2023. www.pewresearch.org/race-and-ethnicity/2023/05/08/diverse-cultures-and-shared-experiences-shape-asian-american-identities/
6. Associated Press; NORC at the University of Chicago (2024). AP VoteCast 2020 General Election, public-use file (files dated December 2024) and codebook (May 2021) [dataset]. Associated Press and NORC at the University of Chicago.
7. Associated Press; NORC at the University of Chicago (2025). AP VoteCast 2024 General Election, public-use file and codebook (March 2025) [dataset]. Associated Press and NORC at the University of Chicago.
8. Hartig, H.; Keeter, S.; Daniller, A.; Van Green, T. (2025). Voting patterns in the 2024 election. In Behind Trump's 2024 Victory, a More Racially and Ethnically Diverse Voter Coalition. Pew Research Center, June 26, 2025. www.pewresearch.org/politics/2025/06/26/voting-patterns-in-the-2024-election/
9. Badrinathan, S.; Kapur, D.; Vaishnav, M. (2020). How Will Indian Americans Vote? Results From the 2020 Indian American Attitudes Survey. Carnegie Endowment for International Peace, October 14, 2020. carnegieendowment.org/research/2020/10/how-will-indian-americans-vote-results-from-the-2020-indian-american-attitudes-survey
10. Badrinathan, S.; Kapur, D.; Vaishnav, M. (2024). Indian Americans at the Ballot Box: Results From the 2024 Indian American Attitudes Survey. Carnegie Endowment for International Peace, October 28, 2024. carnegieendowment.org/research/2024/10/indian-american-voters-election-survey-us
11. AAPI Data; APIAVote; Asian Americans Advancing Justice–AAJC; AARP (2024). 2024 Asian American Voter Survey, report and crosstabs. July 10, 2024. apiavote.org/wp-content/uploads/2024-AAVS-Full-Report_July15.pdf
12. Collaborative Multiracial Post-Election Survey (2026). 2024 Collaborative Multiracial Post-Election Survey, survey page with data access terms. cmps.ss.ucla.edu, read November 6, 2026. cmps.ss.ucla.edu/2024-survey/
13. Ramakrishnan, K.; Wong, J.; Lee, T.; Lee, J. (2017). 2016 Post-Election National Asian American Survey. National Asian American Survey and AAPI Data, May 16, 2017. naasurvey.com/wp-content/uploads/2017/05/NAAS16-post-election-report.pdf

How to cite

@misc{singh2026votechoice,
  author = {Singh, Vishal},
  title = {How South Asian Americans voted, 2016 to 2024},
  year = {2026},
  url = {https://vishalsingh.org/south-asians/vote-choice},
  note = {South Asians in America, chapter 9. Published 7 November 2026.}
}

Acknowledgments

Survey data are from the National Asian American Survey and the Collaborative Multi-racial Post-election Survey, distributed by ICPSR; the Pew Research Center; and the Associated Press and NORC at the University of Chicago. None of them is responsible for the analysis here.