Please fill out this short user survey of only 3 questions in order to help us improve the site. We appreciate your feedback!
Electing to Vote: Strengthening the Credibility of New Kenyan Institutions
Last registered on April 20, 2014


Trial Information
General Information
Electing to Vote: Strengthening the Credibility of New Kenyan Institutions
Initial registration date
Not yet registered
Last updated
April 20, 2014 8:45 PM EDT
Primary Investigator
MIT Sloan School
Other Primary Investigator(s)
PI Affiliation
PI Affiliation
Additional Trial Information
On going
Start date
End date
Secondary IDs
This study analyzes the impact of different sets of text messages (SMS) on voter turnout in Kenya. Kenya held a general election on March 4, 2013, where 6 different officials were elected. The six different positions were President, Member of National Assembly, Senator, Governor, Ward Representative, and Women's Representative. The content of the text messages was varied across treatment groups.
External Link(s)
Registration Citation
Marx, Benjamin, Vincent Pons and Tavneet Suri. 2014. "Electing to Vote: Strengthening the Credibility of New Kenyan Institutions." AEA RCT Registry. April 20. https://doi.org/10.1257/rct.30-1.0.
Sponsors & Partners

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

Request Information
Experimental Details
This study analyzes the impact of different sets of text messages (SMS) on voter turnout in Kenya during the March 2013 election. There were three main treatments in terms of the content of the text messages as described below. In each group, voters received an SMS every day for the five days leading up to Election Day, and one on Election Day itself (thus six messages were sent in total). The randomization was conducted at the level of the polling station. For each treatment, there were two different sub-groups of polling stations: (i) polling stations where all individuals who had registered at the polling station and had provided a phone number were treated, and (ii) polling stations where a random half of the individuals who had provided a phone number at registration were treated (this adds some individual randomization within this set of polling stations).

Treatment 1: Voters received generic mobilization messages encouraging them to vote on Election Day.

Treatment 2: Voters received a message about each of the five positions aside from President (Member of National Assembly, Senator, Governor, Ward Representative and Women's Representative) and were reminded of the significance of each of the positions.

Treatment 3: Voters received information about the monitoring role and the independence of the electoral commission (the IEBC) responsible for tallying the votes and announcing the results.
Intervention Start Date
Intervention End Date
Primary Outcomes
Primary Outcomes (end points)
voter turnout, vote shares, political attitudes, political participation, information, trust in organizations, implementation outcomes
Primary Outcomes (explanation)
The outcomes come from two sources. The administrative data comes from the IEBC on voter turnout and vote shares for each position. The data on attitudes, information, trust, etc. comes from a survey that is currently ongoing (started in Dec 2013). We may use indices (or Likert scales) to aggregate the questions on political attitudes.

In addition to these two sources of outcomes data, we tracked how many text messages were, in fact, received. We will report these outcomes as well.
Secondary Outcomes
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
12,160 polling stations (out of 24,560 polling stations in the country) were selected to be part of the experiment. Half of these polling stations were randomly picked to receive get-out-the-vote text messages. The recipients of the text messages were voters registered in these “treatment” polling stations. The group of “treatment” polling stations was further subdivided into 3 groups, as already described. All messages were sent during the week leading to the election on March 4, 2013. Only individuals who provided their phone number during the voter registration were contacted.
Experimental Design Details
Randomization Method
Polling stations were stratified by county. The randomization code was written in Stata.
Randomization Unit
Polling station level
Was the treatment clustered?
Experiment Characteristics
Sample size: planned number of clusters
The experiment was conducted with 6,086 polling stations in the control group and 2,016, 2,035 and 2,023 polling stations in the three treatment groups.
Sample size: planned number of observations
Approximately 5 million people
Sample size (or number of clusters) by treatment arms
A cluster was a polling station. The number of clusters was as follows:
Control: 6,086 polling stations
T1: 2,016 polling stations
T2: 2,035 polling stations
T3: 2,023 polling stations

In half of the polling stations in each treatment group, only a random half of the individuals were treated. Therefore the number of individuals in the experiment was as follows:

Control: 2,447,876 phone numbers registered at these stations
T1: 626,703 phone numbers treated and 212,333 phone numbers not treated
T2: 617,365 phone numbers treated and 210,181 phone numbers not treated
T3: 599,294 phone numbers treated and 195,223 phone numbers not treated
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
1.5 percentage points in voter turnout
IRB Name
Innovations for Poverty Action, USA (for the survey)
IRB Approval Date
IRB Approval Number
IRB Name
Massachusetts Institute of Technology
IRB Approval Date
IRB Approval Number
Post Trial Information
Study Withdrawal
Is the intervention completed?
Is data collection complete?
Data Publication
Data Publication
Is public data available?
Program Files
Program Files
Reports, Papers & Other Materials
Relevant Paper(s)