Gender-neutral language, persuasion, and policy preferences

Last registered on July 17, 2024

Pre-Trial

Trial Information

General Information

Title
Gender-neutral language, persuasion, and policy preferences
RCT ID
AEARCTR-0014012
Initial registration date
July 13, 2024

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
July 17, 2024, 1:46 PM EDT

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

Locations

Region

Primary Investigator

Affiliation
Princeton University

Other Primary Investigator(s)

PI Affiliation
Bank of Spain
PI Affiliation
INSEAD

Additional Trial Information

Status
In development
Start date
2024-07-15
End date
2024-12-31
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
This is a pilot study for a project studying how the use of inclusive language (in particular gender-neutral language) affects support for different public policies and the persuasiveness of messages aiming to boost support for certain policies.

In Spanish, like many gendered-grammar languages spoken by 39% of the world population (Jakiela and Ozier, 2018), all nouns are assigned to a male or female gender. The traditional default is to use the masculine form as a “generic” when referring to an unspecified sex. A growing (and controversial) movement has challenged this norm, making it a political issue. For example, "employee" in Spanish is either the male noun "empleado" or the female noun "empleada," and historically there has been no word for a non-gendered "employee." Gender-neutral language in our context consists of stating "empleado o empleada" instead of only "empleado" when referring to an "employee.”

In an online survey in Spanish, Peruvian citizens will be provided with short briefs and policy proposals on two context-relevant topics: a social issue related to diversity (bilingual education) and an economic issue (labor informality). The research design cross-randomizes two aspects of the briefs and proposals that respondents see: i) whether the proposed policy is progressive (left-wing) or conservative (right-wing); and ii) whether gender-neutral language is used in the briefs.

This research design allows us to estimate the effect of gender-neutral language on the persuasiveness of the briefs (i.e., if using it makes the respondent more or less likely to support the proposal) separately for progressive and conservative messages and for each of the issues we study. For example, our design allows us to test if gender-neutral language makes a progressive (left-wing) message on an economic issue less persuasive, but would have smaller effects if the message was conservative or if it was about a social issue.

The research design also randomizes the salience of gender-neutral language (whether the respondent sees both proposals with gender-neutral language, both proposals without it, or one proposal with or without gender-neutral language).
External Link(s)

Registration Citation

Citation
Del Carpio, Lucia, Thomas Fujiwara and Carlos Sanz. 2024. "Gender-neutral language, persuasion, and policy preferences." AEA RCT Registry. July 17. https://doi.org/10.1257/rct.14012-1.0
Experimental Details

Interventions

Intervention(s)
Intervention (Hidden)
Intervention Start Date
2024-07-15
Intervention End Date
2024-08-15

Primary Outcomes

Primary Outcomes (end points)
There are two outcomes: respondent support for the proposal she read (measured on a 0-10 scale) and whether the respondent is willing to provide a name for a potentially public online petition with text similar to the proposal she read.
Primary Outcomes (explanation)
Variable construction is straightforward: one is a 0-10 scale and the other is a dummy =1 if and only if a name is provided.

Secondary Outcomes

Secondary Outcomes (end points)
After reading both proposals and answering the questions that generate the primary outcomes, respondents are asked four questions that can, presumably, be affected by the different treatments. These are: i) support for gender quotas in elected office, ii) support for the use of gender-neutral language, iii) who they voted for president in the past election, and iv) how credible/trustworthy they found the proposals shown to them.

We do not expect (i)-(iii) to be affected by treatment. These are asked to be used as covariates and to test for heterogeneous effects by political preferences. However, they are asked after the primary outcomes so as to not “prime” respondents about the study being about gender-neutral language (or gender in general).

We will use (iv) to address potential mechanisms at play (e.g., whether the use of gender-neutral language reduces the trustworthiness and credibility of messages, especially among more conservative respondents).
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
See "Experimental Design (Hidden)" field and the pre-analysis document.
Experimental Design Details
The entire study consists of a short online survey expected to take less than 10 minutes for respondents to complete. Respondents will be adults based in Peru.

The survey begins with a brief set of questions (gender, age, education, language spoken, political ideology on a left-right scale).

Then, each respondent is provided with two policy briefs. One regards bilingual education (school children in Peru being taught in both Spanish and an indigenous language). The second regards informal workers (many Peruvian workers do not have a formal contract with their employer). Whether they see the bilingual education or informality brief first or second is randomly assigned. Below we describe a case where bilingual education is shown first.

At the time respondents see the bilingual education policy brief, the Qualtrics platform randomly assigns respondents to either a "progressive message" (arguing in favor of bilingual education) or a "conservative message" (arguing against it). Moreover, we cross-randomize the use of gender-neutral language in the messages. Messages can be: i) not gender-neutral, ii) 50% gender-neutral, or iii) 100% gender-neutral status. The difference between the “50%” and “100%” versions is whether the nouns are explicitly mentioned in both male or female form whenever possible or a generic masculine noun is substituted by a non-gendered noun 50% of the time. For example, when referring to a set of employees, the gender-neutral version of the brief would use “empleados”, the 100% gender-neutral would always use “empleados y empleadas”, but the 50% version would use “personal”, the Spanish version of “personnel” which is a word that does not connote gender and (in traditional Spanish grammar) could refer to a set of female-only employees.

Then the respondent will be asked whether or not they support the policy and would be willing to sign an online petition in favor of it.

After this, the respondents will also be provided with a brief on informal workers, that will have an analogous set of (randomly assigned) variations, both in the progressive and conservative message dimensions and in the use of gender-neutral language or not. In this context, the progressive message suggests the government should have more resources to enforce labor laws, and the conservative message argues that labor laws should be simplified to lower labor force informality. Again, these messages are based on factual and research-based arguments.

Similarly, respondents will be asked if they support the policy position they just read about, and whether they would be willing to sign an online petition in favor of it.

A brief final set of questions is then asked and the survey concludes.

Randomization Method
The Qualtrics platform will perform the randomization.
Randomization Unit
Randomization is at the "brief" level: each respondent sees two briefs (one on each topic). Each time a brief is shown to her, a variation of the brief of a given topic is chosen. The order of the topics of briefs shown is also randomly determined.
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
1,000 respondents (design is not clustered)
Sample size: planned number of observations
1,000 respondents (design is not clustered)
Sample size (or number of clusters) by treatment arms
When a brief is shown to the respondent, there is a 50% chance it will be not gender-neutral, a 25% chance it will be partially gender-neutral, and a 25% chance it will be 100% gender-neutral. There is also a (independently and cross-randomized) 50%-50% chance the brief will have the "conservative" or "progressive" position on the topic. Which topic a respondent sees first is also randomized with a 50%-50% chance.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB

Institutional Review Boards (IRBs)

IRB Name
Princeton University
IRB Approval Date
2024-07-08
IRB Approval Number
16999
Analysis Plan

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

Request Information

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