Sexual Assault, Victim-Blaming Social Norms, and Willingness to Report

I designed a 2x1 online survey experiment in which subjects answer two direct elicitation questions on their sexual assault victimization status and complete an incentivized social norm elicitation exercise. In the social norm elicitation exercise, subjects are presented with three statements: two “neutral” statements regarding sexual assault and a third statement of interest on victim-blaming. They indicate their own agreement with each statement and are then asked to predict the percentage of previous male survey respondents who agreed with each statement. To investigate how perceptions of social norms impact reporting behaviors, I vary whether the respondent receives a belief correction regarding the actual percentage of previous survey respondents who agreed with the three statements. Subjects see the two “neutral” statements, one regarding whether newspapers should release the names of sexual assault victims to the public and the other regarding whether all women should have access to self-defense classes. The “victim-blaming” statement suggests that women who were sexually assaulted are partly responsible for what happened. Subjects see these three statements in a fixed order: 1) the newspaper statement; 2) the victim-blaming statement; and 3) the self-defense statement. I vary individual’s perception of the social norms by experimentally manipulating whether subjects receive information on the actual percentage of male respondents who agreed with the three statements after they complete the social norm task. To proxy for reporting behaviors, subjects are then presented with a brief description of Callisto, a non-governmental organization with an anonymous sexual assault reporting platform and are provided with a link to create an account.

2025-05-28

2025-06-25

Callisto link clicks to create an account

Individuals are provided with a description of Callisto, a non-governmental organization with an anonymous sexual assault reporting platform. To isolate individuals’ interest in learning more from their interest in creating an account to actually report, subjects are provided with two separate links. The first is specifically for creating an account (and thus potentially filing a report in their system). The second link is for learning more about Callisto.

A link click is any left or right click on the respective link. While the survey is set up so that link clicks automatically open in new tabs, participants may right click and manually select “Open in a New Tab” or “Open in a New Window” to guarantee they do not disturb their progress in the survey. Thus, any left or right click is considered a link click.

Callisto account creation
Callisto link clicks to learn more

I will also receive administrative data from Callisto on the number of accounts created each day, for two weeks prior to launching the experiment and two weeks following the conclusion of the experiment. Importantly, Callisto does require an .edu email address for account creation. Thus, Callisto also knows the institution associated with each account creation and plans to provide the number of accounts created each day within the study period, by university.

To provide suggestive evidence that the link clicks in my experiment did or did not result in actual account creations, I will use information collected in my online experiment on institutional characteristics (like the state and size of the undergraduate institution) to match my survey respondents to actual account creations. I will use these suspected account creations as a secondary outcome, to investigate whether the belief correction treatment had impacts on real-world behavior.

Subjects who are interested in creating an account—or who are interested in filing a report—may want to learn more about Callisto before committing to account creation. In addition, the belief correction treatment may change an individual’s willingness or interest in learning more about a survivor-friendly organization. Thus, I will use Callisto link clicks to learn more as a secondary outcome.

The survey experiment will involve women aged 18-28 that indicate being either current undergraduate students or recent college graduates. These individuals are recruited via Prolific.

The social norm exercise contains questions on individual agreement with social norm statements and predictions of what men from a prior study said. The predictions are incentivized, and subjects will be paid $0.25 for each correct guess. The experiment is a 2x1 design in which I vary whether subjects receive the belief correction.

Thus, I randomly assign participants into one of two treatment conditions:
• T1: No Belief Correction
• T2: Belief Correction

All subjects complete demographic questions and a social desirability scale at the beginning of the survey. They also answer two direct elicitation questions on their victimization status. They first see a brief definition on non-consent. Then, they are presented with two binary questions on their experience with: 1) rape, and 2) all other kinds on nonconsensual sexual contact.

Subjects then complete a social norm elicitation exercise. Participants view the three statements in a fixed order. First, all subjects see a “neutral” sexual assault statement as follows: “Newspapers should not release the name of a sexual assault victim to the public.”

The second statement shown is the social norm statement of interest. This statement is: “If a woman experiences unwanted sexual contact, she may be at least somewhat responsible for what happens.”

The last statement shown is a second “neutral” statement: “All women should have access to self-defense classes.”

Respondents indicate their agreement with each statement on a 5-point Likert scale, from strongly disagree to strongly agree. After indicating their own agreement with all three statements, survey participants are informed that we asked these questions to male current undergraduate students and recent college graduates, aged 18-27, in a prior study. They are then asked to guess what percent of these men they think agreed or strongly agreed with each statement. Respondents receive $0.25 for each correct (within 2 point) guess.

After submitting their guesses, subjects then answer three questions regarding how confident they are in the accuracy of their guess.

Subjects in the belief correction treatment then receive information on the actual percent of men from the prior study that agreed with each statement. At the bottom of this page, subjects see an “attention check” to verify that they read the table. This question is as follows:

• “This is an attention check. To verify that you have read the above information, please answer the following question: Based on the table above, what percent of men from our prior study agreed or strongly agreed with the following statement? If a woman experiences unwanted sexual contact, she may be at least somewhat responsible for what happens.”

All subjects then proceed to the brief passage on Callisto. In this passage, Callisto and its anonymous reporting platform are described. Subjects are provided with two separate links. The first is to create an account. The second link is to learn more.

Randomization is done by the software system.

Individual

No

2,200 women

2,200 women

1,100 women per treatment arm:

• T1: No Belief Correction
• T2: Belief Correction

While this study is targeting a final sample size of 2,200 women, this ultimately is constrained by the response rate of Prolific users. Thus, the final sample size may be smaller.