Scaling Political Information Campaigns

Last registered on February 01, 2018


Trial Information

General Information

Scaling Political Information Campaigns
Initial registration date
February 01, 2018

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
February 01, 2018, 2:54 PM EST

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



Primary Investigator

Stanford GSB

Other Primary Investigator(s)

PI Affiliation
Department for International Development

Additional Trial Information

In development
Start date
End date
Secondary IDs
How do you move from a low information-low accountability political equilibrium to one where voters are informed and politicians feel pressure to perform well? This paper explores two main constraints to this equilibrium shift in the context of a national scale up of a political information campaign in Sierra Leone. The first captures general barriers to scaling a controlled demonstration pilot to serve a much broader population, and the second asks when politicians have incentives (or not) to participate in initiatives that inform voters.
External Link(s)

Registration Citation

Casey, Katherine and Rachel Glennerster. 2018. "Scaling Political Information Campaigns." AEA RCT Registry. February 01.
Former Citation
Casey, Katherine and Rachel Glennerster. 2018. "Scaling Political Information Campaigns." AEA RCT Registry. February 01.
Sponsors & Partners

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

Request Information
Experimental Details


Many successful pilot projects in international aid fail to scale up to serve the broader population of the host country. We study barriers to scaling from multiple angles in the context of a political information campaign in Sierra Leone.

As background, results from a 2012 pilot implemented by Search for Common Ground (SFCG) suggest that hosting debates between Parliamentary (MP) candidates from rival political parties and disseminating them via large public film screenings was highly effective in educating voters, informing their vote choices, and increasing accountability pressure on elected MPs (Bidwell, Casey and Glennerster 2016). On the strength of these results, SFCG raised funds to scale their debates pilot to cover nearly half of all Parliamentary constituencies nationwide in 2018. The first component of this research explores whether the original model can be stretched to reach more voters at lower cost; voter willingness to incur travel costs to access information; and whether debate dissemination can be brought into the business model of private cinema halls.

While debates have become quite common in elections around the world, it is not always easy to convince candidates to participate in them. Frontrunners, for instance, often refuse the invitation from rival party candidates or news organizations to participate in debates, presumably as the downside risk of a bad performance outweighs any potential benefits. On the other hand, if one viable candidate agrees to participate, and a media outlet commits to broadcasting it, the incentives tip toward broader participation as reluctant candidates fear losing out on the free campaign publicity. The second component of this research tests a simple model of a candidate's decision to participate in an initiative that informs voters and sees how it changes with the introduction of a guaranteed broadcast platform.
Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
Component 1: voter knowledge, candidate campaign expenditure, willingness to incur costly action to enhance accountability, and (later) MP performance in office

Component 2: candidate interest in participating in debates, ability to coordinate across party, occurrence of debates
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Component 1: Random assignment of SFCG video debates across 45 of 90 constituencies. Within constituencies, data collection will occur in the constituency headquarters, and three VRCs randomly selected from the 25th to 75th percentile of size distribution. In treated constituencies, the primary public screening will be held in constituency headquarters, and then the three randomly selected VRCs will host a secondary screening, mobilization only and control, respectively.
Experimental Design Details
Component 2: On demand experiment. All candidates in 72 constituencies delivered information nudge. Then, randomization assigns 15 to video treatment, 15 to pure control, 21 to guaranteed radio platform, and 21 to radio information control. Second information nudge delivered tailored to assignment.
Randomization Method
All randomizations are done in the office on a computer.
Randomization Unit
The unit of randomization is the Parliamentary constituency.
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
Component 1: 45 treated and 45 control constituencies
Component 2: 15 video debates (overlaps with T above), 15 pure controls (overlaps with C above), 21 radio airtime, 21 radio information control
Sample size: planned number of observations
Pooling both components: 132 constituencies in total, with at least 264 candidates (exact number of candidates per constituency TBD from NEC official registration), 1,800 voters
Sample size (or number of clusters) by treatment arms
Component 1: at least 2 candidates (total candidates TBD based on NEC official registration) and 20 voters per constituency, generating at least 90 candidates and 900 voters in control and at least 90 candidates and 900 voters in debate treatment

Component 2: at least 2 candidates per constituency, so at least 144 in overall demand estimation, and then at least 30 in video treatment and 30 in pure control (overlaps with component 1), plus at least 42 in radio platform treatment and 42 in radio information control
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Institutional Review Boards (IRBs)

IRB Name
IRB Approval Date
IRB Approval Number
IRB Name
Stanford University
IRB Approval Date
IRB Approval Number
IRB Name
Sierra Leone Ethics and Scientific Review Committee
IRB Approval Date
IRB Approval Number


Post Trial Information

Study Withdrawal

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

Request Information


Is the intervention completed?
Data Collection Complete
Data Publication

Data Publication

Is public data available?

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials