Back to History Current Version

Political Discrimination in a Polarized Society

Last registered on June 06, 2023

Pre-Trial

Trial Information

General Information

Title
Political Discrimination in a Polarized Society
RCT ID
AEARCTR-0011495
Initial registration date
May 28, 2023

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
June 06, 2023, 3:29 PM EDT

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

Locations

Region

Primary Investigator

Affiliation
The Hebrew University of Jerusalem

Other Primary Investigator(s)

Additional Trial Information

Status
In development
Start date
2023-05-29
End date
2024-09-30
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
The proposed project aims to study how political polarization manifests itself in the willingness of individuals to conduct economic transactions across vs. within the political divide.
The project will do so using a combination of a field experiment and a follow-up telephone survey. In the field experiment, the research team will post on a website for classified ads a large number of fictitious ads for used cars in different localities throughout the country. To signal the political orientation of the sellers, we will use bumper stickers to manipulate the photographs included in the ads. Specifically, using the stickers ads will be randomly assigned either: (a) a right-wing orientation; (b) a left-wing orientation; or (c) a neutral orientation (the ads in this “control” group will have no sticker). The research team will then collect for each ad several main outcome variables: the number of hits on the ad on the website, the number of incoming inquiries, and the number of unique phone numbers associated with these inquiries.
In the second step, the research team will use the telephone numbers included in the logs of incoming inquiries to contact all prospective buyers and solicit their participation in a survey. These prospective buyers will not be alerted to the fact that they took part in the field experiment. The survey's main goal is to collect information on buyers’ political orientations.
By comparing outcomes across the three types of ads (right-wing, left-wing, and neutral), the field experiment will reveal the direction and strength of political discrimination in this market. To shed light on the sources of discrimination, we will merge the results of the field experiment with data on the political orientation (and other characteristics) of the relevant localities and with the results of the follow-up survey and then study the association between patterns of discrimination and political orientation both at the locality level and at the individual level.
External Link(s)

Registration Citation

Citation
Zussman, Asaf. 2023. "Political Discrimination in a Polarized Society." AEA RCT Registry. June 06. https://doi.org/10.1257/rct.11495-1.0
Experimental Details

Interventions

Intervention(s)
The project uses a combination of a field experiment and a follow-up telephone survey.
In the field experiment, the research team will post on a website for classified ads a large number of fictitious ads for used cars in different localities throughout the country. To signal the political orientation of the sellers, we will use bumper stickers to manipulate the photographs included in the ads. Specifically, using the stickers ads will be randomly assigned either: (a) a right-wing orientation; (b) a left-wing orientation; or (c) a neutral orientation (the ads in this “control” group will have no sticker).
In the second step, the research team will use the telephone numbers included in the logs of incoming inquiries to contact all prospective buyers and solicit their participation in a survey. These prospective buyers will not be alerted to the fact that they took part in the field experiment. The survey's main goal is to collect information on buyers’ political orientations.
Intervention (Hidden)
Background
Israel has been engulfed in an unprecedented political crisis since January 2023. The attempt of the newly elected right-wing government to push through the parliament major legal reforms has encountered strong resistance, with numerous mass demonstrations and protests taking place throughout the country. While at the time of writing this proposal, the government has paused the disputed legislation, tensions still run high.
The proposed project aims to study how political polarization manifests itself in the willingness of individuals to conduct economic transactions across vs. within the political divide.
The proposed project will address the research question using a combination of a field experiment and a follow-up telephone survey.
Field Experiment
This article studies discrimination in an online market. Such markets have dramatically grown in importance in recent years with the widespread use of the internet and have almost fully replaced more traditional forms of connecting sellers with buyers. The specific platform that will be used in the project is Israel’s leading website for classified ads, which receives millions of hits a month. On any given day there are tens of thousands of ads for used cars on the website.
The research team will post hundreds of fictitious ads for used cars on the classified ads website. The posting will be done in three rounds, each with the same set of 120 car models. The models have been chosen, based on a preliminary analysis, for being representative of the market (in terms of manufacturer, model year, engine size, price range, etc.). Each week we will post 12 ads. All ads will be posted on the website on Monday and removed on Friday. Each of the ads posted in a given week will be randomly assigned a unique mobile telephone number.
The key feature of the field experiment is the random assignment of implied political orientations to sellers. This will be achieved by manipulating the photographs included in the ads. The website encourages sellers to include photographs in their ads, claiming that this significantly raises buyer interest. The vast majority of sellers follow this advice and include in their ads several photographs of the car taken from different angles. To signal political orientation we will rely on bumper stickers attached to our advertised cars. The decision to signal political orientation in this way is motivated by the fact that bumper stickers, including those conveying political messages, are popular in Israel. For each car model, we will produce three sets of photographs. Each set will contain four photographs. The main photograph will have three versions: (1) a version with a sticker associated with the pro-reform camp (“The People Have Chosen Legal Reform”); (2) a version with a sticker associated with the anti-reform camp (“Loyal to the Declaration of Independence”); and (3) a neutral version with no sticker. Potential buyers will see the main photograph on the search results page. In addition, all ad versions will include the same set of three additional neutral photographs of the car, taken from different angles, which buyers will be able to view after clicking on the ad on the search results page.
The website asks sellers to input their name and locality of residence. Since the proposed project focuses on the effect of political tensions on discrimination within the Jewish community, and since the market for used cars is dominated by men, we chose, based on data on first names from the Israeli Central Bureau of Statistics, 12 popular Jewish male first names that will be randomly assigned to the ads. In choosing the 12 localities that will be part of the experiment, we applied several criteria: (a) the localities are either Jewish or integrated; (b) the localities are relatively large, thus ensuring the existence of thick used car markets; (c) the localities are in different parts of the country; (d) the localities vary in terms of their political orientation.
When posting an ad, the website requires the seller to input additional information beyond the specific car model and model year. We will use a common standard for the additional information, e.g. number of previous owners (one, including the current owner), kilometers traveled (1,000 km per month on the road), type of current owner (private individual).
The posted price will be determined based on the website’s built-in car valuation calculator. The calculator takes into account not only the specific car model and model year but also additional factors such as the number of previous owners and the number of kilometers traveled.
In the first round of the field experiment, which will be conducted over 10 weeks, we will randomly assign the pro-reform identity, the anti-reform identity, and the neutral identity to the 120 used car models, with each identity assigned to a third of the total. Each week we will post one ad in each of the 12 cities, where four ads will have a pro-reform identity, four an anti-reform identity, and four a neutral identity. In the second and third rounds, we will use the same car models but switch the assignment of political orientation (e.g. a specific model that was assigned a pro-reform orientation in the first round will be assigned a neutral orientation in the second round and an anti-reform orientation in the third round). Thus, overall the experiment will involve posting 360 ads over 30 weeks.
We will track prospective buyers’ interest in the posted ads using several measures. First, when buyers search the website for a specific car model, model year, etc., they are supplied with a list of cars that match the search terms. For each matching ad, the list includes the main photograph of the car. A hit occurs when a buyer clicks on a specific car in the list. The website allows sellers to see the number of hits on their ads and also the number of times the ads was saved, which are measures of buyer interest. However, it is important to note that on the search results page, buyers see only a small version of the main photograph of each car, which may lead some buyers to miss our political orientation signal. This implies that comparing the number of hits (and the number of times the ad was saved) across ads with different implied political orientations may provide an underestimate of the extent of politically-based discrimination,.
When buyers click on a specific car in the search results page they see a larger version of the main photograph of the car (which makes our political orientation signal more salient) and additional information about the car. Buyers need to click on this photograph to see an even larger version of it and large versions of the additional photographs of the car. They also need to click on a hyperlinked text to see the seller’s phone number.
Our main measures of the intensity of prospective buyers’ interest in a given ad are derived from the log of incoming inquiries (WhatsApp and SMS messages and phone calls). For each inquiry, the log records the telephone number from which it came (unless caller ID is blocked by the prospective buyer). We are mainly interested in two statistics from the log: the total number of incoming inquiries and the number of unique phone numbers associated with these inquiries. The number of inquiries can be further broken down into WhatsApp messages, SMS messages, and phone calls. We can also track the number of incoming e-mails.
Contacting prospective buyers
We will use the phone numbers recorded in the logs of incoming inquiries to contact all prospective buyers and solicit their participation in a survey. Buyers will not be alerted to the fact that they took part in the field experiment. The survey will be presented to participants as studying the attitudes of Israeli citizens. The questions in the first part of the survey will focus on the socio-demographic and other personal characteristics of the participants. In the second part of the survey, we will ask participants about their support for the legal reform and their political orientation.
Intervention Start Date
2023-05-29
Intervention End Date
2023-12-29

Primary Outcomes

Primary Outcomes (end points)
The research team will collect for each ad several outcome variables: the number of hits on the ad on the website, the number of times the ad was saved by prospective buyers, the number of incoming inquiries (WhatsApp and SMS messages and phone calls), the number of unique phone numbers associated with these inquiries, and the number of incoming e-mails.
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Tone of incoming messages
Secondary Outcomes (explanation)
Given sufficient information, we may be able to use methods like "sentiment analysis" to classify the emotional tone of incoming messages as either positive, negative, or neutral.

Experimental Design

Experimental Design
In the field experiment, the research team will post on a website for classified ads a large number of fictitious ads for used cars in different localities throughout the country. To signal the political orientation of the sellers, we will use bumper stickers to manipulate the photographs included in the ads. Specifically, using the stickers ads will be randomly assigned either: (a) a right-wing orientation; (b) a left-wing orientation; or (c) a neutral orientation (the ads in this “control” group will have no sticker).
Experimental Design Details
The specific platform that will be used in the experiment is Israel’s leading website for classified ads. The research team will post hundreds of fictitious ads for used cars on the classified ads website. The posting will be done in three rounds, each with the same set of 120 car models. The models have been chosen, based on a preliminary analysis, for being representative of the market (in terms of manufacturer, model year, engine size, price range, etc.). Each week we will post 12 ads. All ads will be posted on the website on Monday and removed on Friday. Each of the ads posted in a given week will be randomly assigned a unique mobile telephone number.
The key feature of the field experiment is the random assignment of implied political orientations to sellers. This will be achieved by manipulating the photographs included in the ads. The website encourages sellers to include photographs in their ads, claiming that this significantly raises buyer interest. The vast majority of sellers follow this advice and include in their ads several photographs of the car taken from different angles. To signal political orientation we will rely on bumper stickers attached to our advertised cars. The decision to signal political orientation in this way is motivated by the fact that bumper stickers, including those conveying political messages, are popular in Israel. For each car model, we will produce three sets of photographs. Each set will contain four photographs. The main photograph will have three versions: (1) a version with a sticker associated with the pro-reform camp (“The People Have Chosen Legal Reform”); (2) a version with a sticker associated with the anti-reform camp (“Loyal to the Declaration of Independence”); and (3) a neutral version with no sticker. Potential buyers will see the main photograph on the search results page. In addition, all ad versions will include the same set of three additional neutral photographs of the car, taken from different angles, which buyers will be able to view after clicking on the ad on the search results page.
The website asks sellers to input their name and locality of residence. Since the proposed project focuses on the effect of political tensions on discrimination within the Jewish community, and since the market for used cars is dominated by men, we chose, based on data on first names from the Israeli Central Bureau of Statistics, 12 popular Jewish male first names that will be randomly assigned to the ads. In choosing the 12 localities that will be part of the experiment, we applied several criteria: (a) the localities are either Jewish or integrated; (b) the localities are relatively large, thus ensuring the existence of thick used car markets; (c) the localities are in different parts of the country; (d) the localities vary in terms of their political orientation.
When posting an ad, the website requires the seller to input additional information beyond the specific car model and model year. We will use a common standard for the additional information, e.g. number of previous owners (one, including the current owner), kilometers traveled (1,000 km per month on the road), type of current owner (private individual).
The posted price will be determined based on the website’s built-in car valuation calculator. The calculator takes into account not only the specific car model and model year but also additional factors such as the number of previous owners and the number of kilometers traveled.
In the first round of the field experiment, which will be conducted over 10 weeks, we will randomly assign the pro-reform identity, the anti-reform identity, and the neutral identity to the 120 used car models, with each identity assigned to a third of the total. Each week we will post one ad in each of the 12 cities, where four ads will have a pro-reform identity, four an anti-reform identity, and four a neutral identity. In the second and third rounds, we will use the same car models but switch the assignment of political orientation (e.g. a specific model that was assigned a pro-reform orientation in the first round will be assigned a neutral orientation in the second round and an anti-reform orientation in the third round).
Randomization Method
Randomization was done in office by a computer.
Randomization Unit
The unit of randomization is the individual used car. As explained in the "Experimental Design (Hidden)" section above, randomization applies to the first round. In the second and third rounds, we will use the same car models but switch the assignment of political orientation
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
Overall, 120 used car models will be used in the experiment. Each will appear in three different versions: neutral, right-wing, and left-wing. Thus, in total, we will post 360 ads.
Sample size: planned number of observations
In total, we will have 360 observations.
Sample size (or number of clusters) by treatment arms
Overall, 120 ads will be in the "control group" (neutral political orientation), 120 in the "right-wing treatment group" , and 120 in the "left-wing treatment group".
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB

Institutional Review Boards (IRBs)

IRB Name
The Hebrew University of Jerusalem: Faculty of Social Sciences Ethics Committee
IRB Approval Date
2023-05-09
IRB Approval Number
2023-09051

Post-Trial

Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials