Testing the Mandela Hypothesis: An Experimental Analysis of Language and Politics in Côte d'Ivoire

How to cite

Doces, John. 2022. "Testing the Mandela Hypothesis: An Experimental Analysis of Language and Politics in Côte d'Ivoire." AEA RCT Registry. AEARCTR-0008983. https://doi.org/10.1257/rct.8983-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

John Doces*

* Primary investigator

Registry ID
AEARCTR-0008983
DOI
https://doi.org/10.1257/rct.8983-1.0
First registered
Last updated
Status
In development
Trial dates
2022-02-17 – 2022-04-02
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/8983

Abstract

Nelson Mandela said that, "If you talk to a man in a language he understands, that goes to his head. If you talk to him in his language that goes to his heart." This paper tests Mandela's claim using a survey experiment conducted in the Ivory Coast amongst the Dioula, a Mande ethnic group. The Dioula are known as traders and their language is spoken widely in the region and is a major source of pride for them. Like most Ivorians, they also speak French which is the language of the colonizer. Does this matter? Does language affect their political views? To find out, I will be conducting a randomized experiment in which Dioula speakers will be asked to do a survey with questions related to their wellbeing, ethnic identity, support for the current President, and government corruption. The expectation is that when respondents are presented the survey in Dioula that they will (i) report higher levels of wellbeing (ii) a relatively more ethnic and less national identity (iii) more support for the President who is also Dioula. There should be no effect on government corruption from the treatment. Surveys will be randomly assigned using a random number generator. Enumerators are all fluent in Dioula and French and using a random walk they will randomly select respondents in a section of Abidjan that is densely populated with Dioula speakers. Research has important implications for survey research in general as groups like the Afrobarometer conduct their surveys in one language like French in countries such as the Ivory Coast or Senegal. If language affects responses to these questions which are taken from the Afrobarometer then there could be a need to adjust survey methodologies accordingly.

Intervention

The intervention is the language of the survey: half of the surveys will be given in French and the other half in Dioula.

Experimental design

Experiment will be conducted in Abobo, an area of Abidjan that is known to be populated by Dioula speakers. Enumerators will use a random walk sampling method and surveys will be randomly assigned with one in French and the other in Dioula.

Design details

Randomization method
Random assignment will be done using a computer. Surveys will be ordered according to the pre-determined randomization.
Randomization unit
Unit of randomization will be the individual.
Clusters
One cluster
Observations
The total observations will be 200 with the possibility of a second round raising it to 400.
Arms
Each treatment arm will have 100 observations with the possibility of 200 per arm.
Primary outcomes
There are four outcomes: self-reported wellbeing; ethnic vs national identity; support for President; government corruption.

Cited by

No citation data available for this registration yet.

Keywords