Do minorities hide their ethnicity to avoid being discriminated? Experimental evidence from Georgia

Last registered on June 28, 2018

View Trial History

Pre-Trial

Trial Information

General Information

Title

Do minorities hide their ethnicity to avoid being discriminated? Experimental evidence from Georgia

RCT ID

AEARCTR-0002522

Initial registration date

October 11, 2017

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

October 11, 2017, 1:33 PM EDT

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

Last updated

June 28, 2018, 4:48 PM EDT

Last updated is the most recent time when changes to the trial's registration were published.

Locations

Country

Georgia

Region

Tbilisi

Primary Investigator

Name

Nikoloz Kudashvili

Affiliation

CERGE-EI

Contact Primary Investigator

Other Primary Investigator(s)

Additional Trial Information

Status

Completed

Start date

2017-10-16

End date

2017-12-31

Keywords

Education, Post-conflict

Additional Keywords

Discrimination, Trust, Minorities

JEL code(s)

C91, C93 D04

Secondary IDs

Abstract

The prevalence of discrimination against minorities is well documented in the econom-ics literature. But how do minorities react to the threat of discrimination? In this project, we investigate whether minorities misrepresent their ethnicity to avoid being discriminated and how such behavior affects efficiency. We run a modified version of trust game on discrimination in the country of Georgia. In Georgia, first names serve as unambiguous identifiers for ethnic affiliation (Georgian or Armenian). We implement four between-subject treatments: In treatment G-G, a Georgian first-mover (trustor) plays the trust game with a Georgian second-mover (trustee). In G-A, the first-mover is Georgian and the second-mover is Armenian. Following Fershtman and Gneezy (2001), ethnicity is revealed through first names in the two treatments. Comparing first-mover behavior between G-G and G-A yields the basic level of discrimination in the society. In the third treatment, G-A_signal, we implement a signaling stage before the first-mover takes her decision. In the signaling stage, the Armenian second-mover has the option to signal her ethnic affiliation to the Georgian first-mover. Comparing first-mover behavior between G-A_signal and G-A shows whether the signaling stage increases efficiency by mitigating discrimination. Finally, in the fourth treatment, A-A_signal, first- and second movers are Armenians. Comparing second movers’ signaling behavior between G-A_signal and A-A_signal informs us about the extent to which second movers’ signal choice is strategic.

External Link(s)

Registration Citation

Citation

Kudashvili, Nikoloz. 2018. "Do minorities hide their ethnicity to avoid being discriminated? Experimental evidence from Georgia." AEA RCT Registry. June 28. https://doi.org/10.1257/rct.2522-2.0

Former Citation

Kudashvili, Nikoloz. 2018. "Do minorities hide their ethnicity to avoid being discriminated? Experimental evidence from Georgia." AEA RCT Registry. June 28. https://www.socialscienceregistry.org/trials/2522/history/31308

Sponsors & Partners

Experimental Details

Interventions

Intervention(s)

Intervention (Hidden)

We first describe our general procedure, and then provide details about our treatments.
We conduct a lab-in-the-field experiment with adolescents aged between 13 and 16 years in Tbilisi, Georgia. In Tbilisi, schools are segregated by ethnicity (Georgian and Armenian schools) and the first names of individuals are clear identifiers of their ethnicity. Following Fershtman and Gneezy (2001), we use the trust-game as our basic paradigm to study discrimination. The trust game is a two-person (first-mover and second-mover), two-stage game. Each person takes one decision and both decisions together determine payoffs. The first-mover is endowed with 10 tokens, the second mover does not receive initial endowment for the game. In the first stage, the first-mover decides how many tokens to send to the second-mover. Each token sent is tripled. In the second stage, the second-mover decides how many tokens to return to the first-mover. The returned tokens are not tripled. Tokens sent by the first-mover is referred to as “trust”, tokens sent by the second-mover is referred to as “trustworthiness”. Since matching needs to be done after decisions are collected, the strategy method is used to collect second-mover’s decisions. Pair-matching is always across different schools. In addition to choices, we measure first-order beliefs about the opponents’ number of tokens sent in an incentive-compatible way. We incentivize choices with tokens which can be exchanged for a variety of small gifts after the experiment. More tokens buy more gifts. All participants also receive one present after the sessions as a show-up fee.

We have four treatments:
1. Treatment “G-G”: The game is played with Georgians in both roles (first- and second movers). Participants in both roles know the ethnicity of their opponents (following Fershtman and Gneezy (2001), we use first names as identifiers for ethnicity.
2. Treatment “G-A”: In that treatment, the first-mover is Georgian and the second mover is Armenian. Participants in both roles know the ethnicity of their oppo-nents
3. Treatment “G-A_signal”: In that treatment, the first-mover is Georgian and the second mover is Armenian. Before the first mover makes his move, we imple-ment a signaling stage in which the second mover can send a signal about her ethnicity to the first-mover.
4. Treatment “A-A_signal”: The game is played with Armenians in both roles (first- and second movers). Before the first mover makes his move, we imple-ment a signaling stage in which the second mover can send a signal about her ethnicity to the first-mover.

We are interested in the following treatment contrasts: First, we compare first mover’s choices in G-G and G-A to establish the extent of discrimination present in Georgia. Second, we compare first mover’s behavior between G-A to G-A_signal to investigate whether the option to signal increases overall efficiency (i.e., the number of tokens produces). Third, we compare second mover’s signaling behavior between G-A_signal and A-A_signal to investigate whether signal choice depends on the recipient (i.e., whether it is strategic).

Intervention Start Date

2017-10-16

Intervention End Date

2017-12-31

Primary Outcomes

Primary Outcomes (end points)

1. Overall efficiency: That is the number of total tokens “produced” in the game (number of tokens sent by first mover times three).
2. Signaling behavior: The message chosen second movers in “G-A_signal” and “A-A_signal”.

Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)

First- and second-mover’s beliefs about each other’s behavior.

Secondary Outcomes (explanation)

Experimental Design

We conduct a lab-in-the-field experiment with four treatments: “G-G”, “G-A”, “G-A_signal”, “A-A_signal”.

Randomization Method:
- Assigning participants to different treatments by blindly drawing a numbered card out of a bag of several cards.
- Georgian Schools: Randomization between G-G (first- and second-mover), G-A (first-mover), G-A_signal (first mover). Armenian schools: Randomization be-tween G-A (second-mover), G-A_signal (second-mover), A-A_signal (first- and second mover).

Experimental Design Details

Randomization Method

Numbered Cards drawn from non-transparent bag

Randomization Unit

individual

Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters

700 (350 Georgian participants; 350 Armenian participants)

Sample size: planned number of observations

700

Sample size (or number of clusters) by treatment arms

There will be four treatments.
G-G: 160 observations (80 first-movers, 80 second-movers)
G-A: 160 observations (80 first-movers, 80 second-movers)
G-A_signal: 220 observations (110 first-movers, 110 second-movers)
A-A_signal: 160 observations (80 first-movers, 80 second-movers)

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Continuous outcomes. For continuous outcomes, we rely on Johnson and Mislin’s (2011) Meta-Study on trust game behavior. They find that, across 161 independent trust games, the share of tokens sent by the first-mover was 0.5 with a standard deviation of 0.12. Based on these fig-ures, we perform our power calculation for our first primary outcome, overall efficienty as measured by first-mover choices. Comparing first movers between treatments G-G (80 observations) and G-A (80 observations), our minimal detectable effect size is 5 percent of initial endowment. Comparing first movers between treatments G-A (80 ob-servations) and G-A_signal (110 observations) yields a minimal detectable effect size of 5 percent of the initial endowment. Binary outcomes. For our second set of primary outcomes, the dependent variable is binary (e.g., 1=signaling Armenian ethnicity, 0 else). Assuming that in treatment G-A_signal 30 percent of second-movers reveal their Armenian identity, the minimal detectable effect size for comparing second-mover’s signals between G-A_signal and A-A_signal is 19 percentage points.

Supporting Documents and Materials

IRB