Can we text bad politicians out of office? Experimental evidence from a mobile-based voter information campaign in Uttar Pradesh

Last registered on March 04, 2026

Pre-Trial

General Information

Title

RCT ID

AEARCTR-0017542

Initial registration date

January 09, 2026

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published

January 12, 2026, 8:12 AM EST

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Last updated

March 04, 2026, 7:28 PM EST

Last updated is the most recent time when changes to the trial's registration were published.

Locations

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Primary Investigator

Name

Siddharth George

Affiliation

Boston University

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Yusuf Neggers

PI Affiliation

University of Michigan

Contact Investigator

PI Name

Sarika Gupta Ringwala

PI Affiliation

World Bank

Contact Investigator

Additional Trial Information

Status

On going

Start date

2017-02-01

End date

2026-07-31

Keywords

Electoral, Governance

Additional Keywords

voter information

JEL code(s)

D72

Secondary IDs

Prior work

This trial is based on or builds upon one or more prior RCTs.

Abstract

In this study, we send voters text messages and automated voice calls providing information on major-party candidate criminal records (pending cases and underlying criminal charges) drawn from official sworn affidavits. The information is provided to randomly selected mobile subscribers shortly before elections to the state legislative assembly in the state of Uttar Pradesh. We then examine impacts on voting outcomes using publicly available, official polling station voting returns data, as well as survey data on voters' beliefs, attention and preferences.

External Link(s)

Registration Citation

Citation

George, Siddharth, Sarika Gupta Ringwala and Yusuf Neggers. 2026. "Can we text bad politicians out of office? Experimental evidence from a mobile-based voter information campaign in Uttar Pradesh." AEA RCT Registry. March 04. https://doi.org/10.1257/rct.17542-1.1

Sponsors & Partners

Experimental Details

Interventions

Intervention(s)

Treated mobile subscribers receive a text message and an automated pre-recorded voice call two days before the election, with the content tailored to treatment arm. Control group members receive no messages.

Intervention Start Date

2017-02-20

Intervention End Date

2022-03-15

Primary Outcomes

Primary Outcomes (end points)

Voting at the polling-station-by-candidate level; Turnout at the polling-station level. In 2022, voting at the individual-by-candidate-level and turnout at the individual level.

Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)

In 2022: Voter beliefs: whether voters have accurate beliefs about candidate criminality; Voter attention: whether voters find criminality (and other attributes) salient; Voter decision weights: importance placed on each candidate attribute when deciding vote.

Secondary Outcomes (explanation)

Experimental Design

Within experimental sample of villages, randomly assign villages to one of the treatment arms. In 2022, randomly assign individuals in sample assembly constituencies to one of the treatment arms.

Experimental Design Details

Not available

Randomization Method

Randomization done in office using computer software.

Randomization Unit

Village; In 2022, individual

Was the treatment clustered?

Yes

Experiment Characteristics

Sample size: planned number of clusters

Fourth phase: 3,800 villages; Seventh phase: 850 villages. In 2022, not clustered.

Sample size: planned number of observations

Fourth phase: 5,000 polling stations, 15,000 polling-station-by-candidates; Seventh phase: 1,100 polling stations, 3,300 polling-station-by-candidates. In 2022, 3,700 individuals.

Sample size (or number of clusters) by treatment arms

Fourth phase: 2400 villages control, 1400 treated (1/3 information content, 1/3 information and coordination content, 1/3 information and ethnic voting content); Seventh phase: 425 villages control, 425 villages placebo content. In 2022, 1700 control, 2000 treated (4/5 information content, 1/5 information and coordination content).

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Supporting Documents and Materials

IRB

Institutional Review Boards (IRBs)

IRB Name

Harvard University - Area Committee on the Use of Human Subjects

IRB Approval Date

2017-02-18

IRB Approval Number

IRB00000109

Analysis Plan