India's Voter (Re-)Enrollment Exercise and its Consequences on Institutional Trust, Candidate Choice, and Cognition

The study embeds two randomized interventions within a single survey instrument: (1) Priming experiment — randomized ordering of the election process module to measure impact of electoral process priming on trust, risk aversion and cognitive control; (2) Candidate vignette experiment — respondents make candidate choices to reveal demand for electoral process related activities from local politicians.

The study embeds two randomized interventions within a single survey instrument: (1) Priming experiment — randomized ordering of the SIR process module (Section 4) relative to the trust in institutions module (Section 5), the risk aversion module (Section 7), and the cognitive control module (Section 8); (2) Candidate vignette experiment — respondents make three forced-choice candidate preference decisions for upcoming elections. Each round pairs a Public Goods platform from one cadidiate (constant) against either another Public Goods, a Voter Deletion or Voter Addition platform from the second candidate. Candidate religious identity (Hindu/Muslim) is independently randomized per round using pre-specified name pools. For the cognitive control task in Experiment 1, a computerized spatial response task involving arrows is embedded. Tasks are randomized to either a congruent or incongruent condition.

2026-05-05

2026-05-31

Trust in the Election Commission of India (ECI); trust in the EVM process (5-point scale); risk aversion, candidate preference in vignette experiment (forced choice); accuracy and reaction time in incongruent arrow trials relative to congruent arrow trials.

Trust items are drawn from Section 5.1 of the survey instrument (5-point scale: great deal / some trust / neutral / no trust at all / no response). Candidate preference is the forced-choice outcome in Section 6 (Candidate A / Candidate B / no preference). Cognitive outcomes are accuracy (0/1) and reaction time (ms) from the main block of the Section 8 task, comparing congruent and incongruent trials.

Trust in local BLO, Councilor/Sarpanch, MLA, MP, State Government, Central Government, and Judiciary; change in trust in institutions relative to 2024 Lok Sabha elections; perceived fairness of electoral process; worry about roll revision process; strength of candidate preference; awareness and knowledge of SIR process; prior voting behavior.

Cross-sectional survey with two embedded randomized experiments administered to voters in Rajasthan, India. Respondents are randomly sampled from 400 randomly selected polling stations across 28 randomly selected Assembly Constituencies. The sample of 4,000 respondents are drawn from the list of registered voters.

Experiment 1 (Priming): Between-subjects randomization of module order. The electoral process module (module E) is administered either before or after the trust/risk preference/cognitive control module. Either all three module appear before the module E or all three appear after. The before-after ordering is randomized with 0.67 probability of E first. The ordering of the three modules within a before or after treatment are also randomized.

Experiment 2 (Candidate Vignette): Within-subject design. All respondents complete three forced-choice rounds; the order of the Public Good only, and two electoral process frames is randomized across rounds. Candidate religious identity (Hindu name / Muslim name) is independently randomized per respondent from pre-specified name pools, and kept fixed across the rounds for a given respondent.

Cross-sectional survey with two embedded randomized experiments administered to registered voters and deleted voters in Rajasthan, India. Respondents are sampled from 400 polling stations across 28 Assembly Constituencies. The sample of approximately 4,000 respondents includes both currently enrolled voters and voters deleted during the 2026 SIR exercise, identified via the SIR 2026 final rolls and list of deleted voters released by ECI as part of the SIR process.

Experiment 1 (Priming): Between-subjects randomization of module order. The SIR process module is administered either before or after the trust/risk preference/cognitive control module. Either all three module appear before the SIR module or all three appear after. The before-after ordering is randomized with 0.67 probability of SIR first. The ordering of the three modules within a before or after treatment are also randomized. This ensures that about 22% of sample have one of the outcome module coming right after SIR module, which would maximize impact on that outcome.

Experiment 2 (Candidate Vignette): Within-subject design. All respondents complete three forced-choice rounds; the order of the Public Good only, Deletion and Addition frames is randomized across rounds. Candidate religious identity (Hindu name / Muslim name) is independently randomized per respondent from pre-specified name pools, and kept fixed across the rounds for a given respondent.

SurveyCTO built-in randomization at the individual respondent level, conducted at survey initiation.

Individual respondent

No

200 polling stations across 28 Assembly Constituencies

4000 voters

~2000 per arm (Exp. 1); ~2000 for each pair of candidate types (Exp. 2)