Estimating In-Group Bias With Fuzzy Identities: Evidence from Regional Bias in Mexico (Online Experiment)

How to cite

Duque, Mauricio Fernandez. 2020. "Estimating In-Group Bias With Fuzzy Identities: Evidence from Regional Bias in Mexico (Online Experiment)." AEA RCT Registry. AEARCTR-0005236. https://doi.org/10.1257/rct.5236-2.0

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

Mauricio Fernandez Duque*

* Primary investigator

Registry ID
AEARCTR-0005236
DOI
https://doi.org/10.1257/rct.5236-2.0
First registered
Last updated
Status
Completed
Trial dates
2015-10-13 – 2015-12-08
Countries
Mexico ()
JEL codes
C91, C92, D03, Z13
Registry record
http://www.socialscienceregistry.org/trials/5236

Abstract

Measuring discrimination is crucial to key questions in political economy, but studies of how group members discriminate against outsiders have focused on instances where membership is sharply defined. To study common and important instances of fuzzy group membership, I propose two approaches: one based on notions of distance, and the other based on measuring subjective perceptions of group membership. Taking regions in Mexico as an example, I use survey, field experiment and administrative data to test for regional bias in donations, in short-term migration, and in long-term migration. The bias appears only if region membership is conceptualized as fuzzy. Subjective perceptions provide the most consistent regionalization for capturing regional bias, and allow for testing several mechanisms. First, national appeals and a stronger attachment to the nation diminish regional bias. Second, the way individuals draw regions exacerbates regional inequality. The novel approaches can be applied to other identities with fuzzy demarcations.

Intervention

The intervention studies the impact of varying a donation petition. We studied the impact of varying the state in Mexico where the beneficiaries of donation petition cards resided, as well as whether the petition included a national prime. We were interested in whether individuals had an in-region bias, whether the bias was affected by the prime, and whether these effects differed according to individuals' attachment to the region and the nation. We measured attachment in a baseline two weeks earlier than the endline.

Experimental design

College students from ITAM, a private university in Mexico, took part in two short online surveys between October 13th and December 8th, 2015.

In the first survey, they selected the state they had the strongest connection to, as well as the region that they thought corresponded to that state. In addition, they were asked about their national and regional attachment (each on a 5-point Likert scale), how they ranked the states by income levels, and sociodemographic characteristics. This information was used to define subjects' regions, select rich and poor states within the in-region and out-region, and randomly choose one state from each of those subsets. It was also used to identify regionalists (those more attached to their region) and nationalists (those not more attached to their region).

A second survey was conducted at least two weeks after the first. Subjects who entered the second survey were assigned to a treatment that was either an in-region or out-region state with or without a nationalist prime. We blocked on regional identity, and randomized whether they were assigned a rich or poor state within their region in order to improve balance.

Subjects in the endline survey were exposed to a donation petition towards helping child development efforts in their assigned state. The donation petition are cards which feature the name of the NGO, three indigenous children and the caption `Thanks for nourishing the children of [Name of state]'. The card varies on two dimensions: the name of the state, and whether or not it contains nationalist colors. The nationalist colors, which are the colors of the flag, replace the teal color of the alternative version in the card's frame and in the letters of the name of the state.

The endline survey had four types of questions. The first type of question was meant to get at whether the prime brought the nation to the top of mind, and whether it was informative. To get at top-of-mindedness, we asked an open ended word association question. We allowed subjects to give five answers. To get at informativeness of the prime, we asked questions regarding the efficiency of the NGO, the political parties ruling the state, organizations tied to the NGO, the poverty level of the state, and the poverty of the beneficiaries, as well as about the perceived attractiveness of the card.

The second type of question was meant to test for allocation decisions. We had two variables to capture this. The first was a dictator game, in which subjects were told that, in case they won a 100 peso prize, they could allocate any part of the prize to beneficiaries from the donation petition. About 10\% of subjects received the prize, which was raffled among all endline participants. The second question was about redistribution. We asked how in favor they were of a policy that would tax the nation and whose benefits were concentrated on the state they were assigned.

The third type of question asked about cognition. We wanted a question that got at whether the prime made subjects think of the state differently with the prime. In order to do so, our main question is `When you think of [Name of state], is it easier for you to think of it as part of a region or as part of Mexico?'. We asked this question not only of the assigned state, but of the three other states that would have been assigned to the subject had she been in alternative treatment arms. A second question we asked was whether they thought the state is similar to other states in Mexico.

The fourth type of question asked the same regional and nationalist attachment questions as in the baseline. We did not want subjects to answer based on their memory of what they answered in the baseline, which is part of why the endline survey came two weeks after the baseline.

Design details

Randomization method
To randomize, at the beginning of the baseline survey we assigned subjects a state in their region or outside their region. We further randomized whether the state was rich or poor. We use this to control for donations based on need, and not identity. The rich (poor) states the subjects were assigned were subjectively identified by the subject as rich (poor), and either widely perceived by others to be rich or objectively relatively rich (poor) using 2013 GDP. There were 13% of respondents whose in-region states did not vary in terms of income (i.e. having both rich and poor states). We faced a trade-off between sticking strictly to the states the subject selected as part of his or her region and balancing on state income. We favored the latter. If a region had a middle-income and rich or poor state, we considered it balanced, if not, we added a state that at least a third of the subjects of the same state considered part of their region. Once we had defined the set of in-region rich and poor states and out-region rich and poor states we randomly chose one state from each set. We then assigned subjects to one of four conditions: in-region with prime, in-region without prime, out-region with prime, and out-region without prime. We block-randomized on regionalism. Regionalism is defined by the difference between their reported attachment to their region and their nation.
Randomization unit
Subjects
Clusters
955 subjects signed up for the baseline. 580 of those signed up for the endline.
Observations
955 subjects signed up for the baseline. 580 of those signed up for the endline.
Arms
There are 142 subjects assigned to a state in their own region and a prime There are 151 subjects assigned to a state outside of their own region and a prime There are 141 subjects assigned to a state in their own region and without a prime There are 146 sbjects assigned to a state outside of their own region and without a prime
Primary outcomes
We are interested in several outcome variables. The key outcome variables are the following: - How much money was donated to an NGO from an endowment of 100 pesos. - Subjects' cognition about the states Our main question was whether subjects found it easier to think of the beneficiary state as part of a region or as part of Mexico. - Subjects attachment to the region and the nation. We asked subjects how attached they were to the region and to the nation, each on a 5-point Likert scale. - Subjects redistributive preferences to the assigned state. We asked subjects how much they would support a redistributive policy that would mostly benefit their beneficiary state. - Whether the prime made the nation salient and whether the prime was uninformative. To get at whether the prime made the nation salient, we asked them to give five words that came to mind after observing the donation petition. To get at whether the prime was uninformative, we asked a battery of questions related to the NGO, the beneficiary state and the attractiveness of the card what would affect the donation decision.
Sponsors
Name: Harvared Experimental Working Group Location: Harvard University Url: https://projects.iq.harvard.edu/hewg/home

Cited by

No citation data available for this registration yet.

Keywords