Harming to Signal

How to cite

Haenni, Simon; Lichand, Guilherme. 2018. "Harming to Signal." AEA RCT Registry. AEARCTR-0002856. https://doi.org/10.1257/rct.2856-4.1

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

Simon Haenni* , Guilherme Lichand

* Primary investigator

Registry ID
AEARCTR-0002856
DOI
https://doi.org/10.1257/rct.2856-4.1
First registered
Last updated
Status
Completed
Trial dates
2018-06-12 – 2019-12-31
Countries
Malawi ()
Registry record
http://www.socialscienceregistry.org/trials/2856

Abstract

Girls and boys in developing countries are often exposed to a number of traditional practices that have strong, negative effects on their future, like for example child marriage and sexual initiation rituals. The measurement of the prevalence of these harmful traditional practices is particularly difficult due to social desirability biases and thus, reliable evidence is scarce.

This research project measures the prevalence of different harmful traditional practices in Malawi, while explicitly accounting for social desirability bias by using a combination of advanced elicitation techniques. This not only provides us with an individual-specific measure for susceptibility to social pressure, but also allows us test interventions intended to mitigate these social forces.

Further, we shed light on underlying behavioral mechanisms behind harmful traditional practices. In particular, we test the hypothesis that families take part in harmful traditional practices for the purpose of social signaling.

Intervention

Signaling Intervention:
Introducing a new signaling language, i.e. a less harmful signaling opportunity, might decrease support for other, more harmful signals.

Self-esteem priming:
Higher self-esteem is expected to decrease reputational concerns and thereby reduce the willingness to contribute to goods with status signaling component, like local traditions. At the same time, self-esteem may have an effect on social desirability bias.
Therefore, we use a self-affirmation task (Steele 1988, Cohen et al. 2009, Hall et al. 2013), where we ask treated individuals to reflect on a recent experience or achievement that made them feel proud. Control individuals are asked to talk about their favorite dish. We use the 10-question Rosenberg self-esteem scale as manipulation check of the self-esteem priming.

Experiment on perceived public image:
In order to learn whether people use harmful traditional practices for signaling their pro-sociality/status, we want to find out how people perceive a hypothetical person who does (not) engage in harmful traditional practices. We therefore randomly assign people to one of two conditions.

List experiments:
List experiments (Raghavarao and Federer, 1979) are a standard method to account for social desirability bias in survey questions. We adapted the method to work under constraints regarding illiteracy.

Experimental design

Signaling Intervention:
Assignment to treatment and control group is done at the village level. Incomplete take-up is treated as intention-to-treat. Thus, villages where chiefs do not allow for the chosen treatment are still included in the survey – if permission is given.
2x2 Design with cross-randomized variation in manager of the intervention:

Self-esteem priming:
Assignment to treatment and control group is done at the household level. Half the households within each village are randomly assigned to the treatment condition while the other half is assigned to the control condition.

Experiment on perceived public image:
Assignment to treatment and control group is done at the household level. Half the households within each village are randomly assigned to the treatment condition while the other half is assigned to the control condition.

List experiments:
Assignment to treatment and control group is done at the household level. Half the households within each village are randomly assigned to treatment condition 1 while the other half is assigned to treatment condition 2.
In condition 1, subjects answer 3 sub questions in List experiments 1-3 and 4 sub questions in List experiments 4-6.
In condition 2, subjects answer 4 sub questions in List experiments 1-3 and 3 sub questions in List experiments 4-6.
Additionally, individuals answer 3 sub questions in List experiment 7 and 4 sub questions in List experiment 8, or vice versa, with equal proportions in both treatment conditions of the self-esteem intervention.

Design details

Randomization method
Randomization is done by sorting on values of the runiform() function in Stata 14.
Randomization unit
The signaling intervention is randomized at the village level, while self-esteem priming, List experiments, and image perception are randomized at the household level.
Clusters
413
Observations
8260 households >>> ~25,000 individuals
Arms
118 villages: donation+bracelets 117 villages: donations 89 villages: bracelets 89 villages: control
Primary outcomes
-Attitudes towards and planned future engagement in harmful traditional practices -Public perception of an individual that engages in harmful traditional practices -Prevalence of harmful traditional practices -Village chiefs’ characteristics and self-perception

Cited by

No citation data available for this registration yet.

Keywords