close× Call Us
close×
Title Information and Politicians’ Beliefs About Election-Related Violence
Post date 02/15/2019
C1 Background and Explanation of Rationale Recent research on election-related violence in Kenya has shown that politicians may overestimate the benefits of violence and underestimate its costs (Rosenzweig 2018). Specifically, politicians underestimate the strength and breadth of voter backlash against the use of violence in their efforts to win office, including among their core coethnic supporters. Thus, elite misperceptions about the electoral effects of violence may explain the persistence of election-related violence in Kenya, and correcting these misperceptions may reduce the use of violence associated with electoral competition. The experiment described here randomizes the provision of information about the negative effects of violence on voter support to a sample of Kenyan politicians, evaluating the effect of this information on their perceptions about the efficacy of violence and violent rhetoric as electoral tactics, and assessing whether the effect of the information depends on politicians' pretreatment levels of certain cognitive biases or the length of their engagement in electoral politics. The study seeks to shed light on why politicians misperceive the effects of violence on voting, and whether providing systematic information about these effects can reduce violent electoral tactics by reducing politicians' misperceptions about their efficacy.
C2 What are the hypotheses to be tested?

If elite misperception is due to a lack of information about voter backlash against violence
and violent rhetoric, then:
H1: Politicians in the treatment group will rate politicians with allegations of employing
violence or violent ethnic rhetoric as less likely to win an election than those in the control
group.
H1a: This effect will be most pronounced among politicians with less education and less
experience in electoral politics.

If elite misperception is due to cognitive biases that affect politicians' interpretation of
the evidence about the efficacy of violence and violent rhetoric, then:
H2: The effect of the treatment will be greater for politicians with lower pretreatment

The study will analyze the effects of the treatment on the following outcomes:

1) Respondents' perception of the likelihood that a candidate with a history of violence will win election;

2) Respondents' perception of the likelihood that coethnics with vote for a candidate with a history of violence;

3) Respondents' perception of the likelihood that non-coethnics with vote for a candidate with a history of violence;

4) Respondents' perception of the likelihood that a candidate that has employed hostile ethnic rhetoric will win election; and

5) Respondents' perception of the likelihood that coethnic voters will vote for a candidate that has employed hostile ethnic rhetoric.

All outcomes are measured on a 1 to 10 scale, 1 indicating least likely and 10 indicating most likely.

C3 How will these hypotheses be tested? *

Half of respondents will be randomized in the information treatment and half into a control group that does not receive the information. Respondent will then answer outcome questions related to their perceptions of the likely electoral success of two candidates for office describes in short vignettes, one of which has an alleged history of instigating violence, the other of which has employed hostile ethnic campaign rhetoric.

The data will be analyzed in a regression framework using OLS. Each outcome question will be regressed on an indicator for the experimental treatment. The null hypothesis of no effect will be tested using a one-tailed test, with variance estimated using Huber-White robust standard errors. To test for heterogenous treatment effects on the basis of education and electoral experience or levels of cognitive bias, I will estimate an interaction effect, interacting the treatment indicator with either 1) pretreatment measures of education and electoral experience or 2) pretreatment measures of cognitive bias (the latter both individually and in combination).

C4 Country Kenya
C5 Scale (# of Units) 584
C6 Was a power analysis conducted prior to data collection? No
C7 Has this research received Insitutional Review Board (IRB) or ethics committee approval? The research was approved for exemption by the IRB at Boston University.
C8 IRB Number n/a
C9 Date of IRB Approval December 12, 2018
C10 Will the intervention be implemented by the researcher or a third party? Researchers
C11 Did any of the research team receive remuneration from the implementing agency for taking part in this research? not provided by authors
C12 If relevant, is there an advance agreement with the implementation group that all results can be published? not provided by authors
C13 JEL Classification(s) not provided by authors