Mislearning from Silence and Misperceived Social Norms

2023-02-01

2023-06-01

We will recruit college students as participants and the key outcome variables include:
1. private beliefs and second-order beliefs on the a set of socioeconomic and political topics
2. public expression of views in group chats.
3. willingness to take certain political expression actions (signing a petition and donating to a charitable cause)

We will conduct a lab-in-the-field experiment through Xlab with UC Berkeley students.

Our study consists of two parts, “first movers” (10% of participants) and “second movers” (90% of participants), which allows us to study the dynamic effects of increasing attention on silence on political expression and action. Specifically, all respondents will participate in Zoom sessions where they have the opportunity to express their views on various political and socioeconomic topics. However, “first movers” will participate in these Zoom sessions without any additional information while “second movers” will first receive anonymized summaries of the views expressed by “first movers” before participating in their own Zoom discussions. To isolate the causal effects of increasing attention to silence, we will randomly assign "second movers" to different information treatment groups where they see different summary statistics about "first movers". We will also ask all participants to complete a baseline survey (prior to the Zoom sessions) about their private views on these topics and their beliefs about the views of other participants as well as an endline survey (after the Zoom sessions) that again elicits beliefs about the views of other participants as well as willingness to engage in other types of political expression and action.

Specifically, our experimental design is as follows:

1. Baseline survey: We will ask participants about their private beliefs on various socioeconomic and political topics. Participants can choose “agree”, “disagree” or “prefer not to say” for each statement. Participants will also be incentivized to guess how other study participants privately answer each topic. We will also collect demographic data, including gender, ethnicity, school year, major, and self-identified political affiliation.

2. After participants complete the baseline survey, we will randomly split participants into three subsamples: 1) First movers (10% of participants); 2) Second movers - Control group (45% of participants), and 3) Second movers - Treatment group (45% of participants).

3. First movers - public expression: The “first-movers” will be asked to attend a Zoom session. Each session will have 12 participants. During the Zoom sessions, we will read participants a statement related to a specific topic and participants will send a private chat message to the moderator indicating whether they ”agree” or ”disagree” with the statement if they would like to share their views publicly. If they do not want to share their views, they simply do not send any messages to the moderator. All participants who respond in the chat will be called upon to share their views with the rest of the Zoom room. The order in which participants are called upon will be randomly chosen to shut down potential dynamic effects within a Zoom session.

4. Summary Statistics about the previous discussion: We randomly assign the remaining 90% of the participants to two treatment groups where we show them different versions of summary statistics about the Zoom expression decisions of the first movers.
- Control participants will be provided with information about how many “first movers” said “agree” and “disagree” respectively for each topic. For example: “10 publicly said agree, 6 publicly said disagree.”
- Treatment participants will be provided with the same information as the control group plus additional information about how many “first movers” stayed silent for each topic. For example, “10 publicly said agree, 6 publicly said disagree, and 8 stayed silent.”

5. Control/Treatment groups - public expression: All second movers will then attend a Zoom session with other participants assigned to the same information treatment group (i.e. “control” Zoom sessions consisting of only participants assigned to the control group or “treatment” Zoom sessions consisting of only participants assigned to the treatment group). Upon entering the Zoom sessions, we will send a survey link in the chat which includes the corresponding summary statistics above and elicits their second order beliefs about other participants’ views. The protocol for the Zoom sessions is otherwise identical to the “first movers”. At the end of each Zoom session, participants will complete an endline survey where we will ask them again about their beliefs about the distribution of others’ views, their recollections of the Zoom discussions they participated in, and their willingness to take certain political actions, such as signing a petition, donating to a cause, or attending a public meeting.

Randomization done in office by a computer

The randomization will be done at the individual level.

No

We plan to recruit 500 Berkeley students. Assuming a 20% attrition rate, we estimate that about 400 students will complete the study.

Recruit 500 students, estimate that we will have 400 observations in the endline.

We will recruit 50 first-movers, 225 students in the control group and 225 students in the treatment group.