Selection and impact in the sharing of stigmatized information

The experimental intervention consists of the manner in which the counseling helpline is introduced to participants. The four framings are designed to separately alleviate the sender's image concerns regarding sharing stigmatized information, and the recipient's image concerns regarding being targeted to receive information about a stigmatized service.

2023-08-15

2023-09-30

Sender’s choice of whether the recipient will receive information about the helpline from the study. (Constructed as a recipient-level indicator variable taking 1 if the sender chooses the person in the subset of friends who should receive information about the helpline, and 0 for the sender’s friends who the sender does not select.)

The recipient’s take-up of the helpline: Takes 1 if the recipient agrees that she wishes to be contacted by the helpline if found eligible (measured prior to completing the screening to confirm eligibility)

Recipient’s self assessment of her distress: Standardized score of the recipient’s self assessment of her distress on a scale from 1 to 10 with 10 being the most distressed, standardized using the mean and standard deviation of Framing 1.

Recipient’s self-assessment of the benefit she would gain from using the helpline: Standardized score of the recipient’s self-assessment of her distress on a scale from 1 to 10 if she were to start having weekly phone sessions with the helpline for 1 month, standardized using the mean and standard deviation of Framing 1.

Recipient’s assessment of the distress that her friends believe she is experiencing: Standardized score of the recipient’s self-assessment, standardized using the mean and standard deviation of Framing 1.

Recipient’s opinion of the ideal framing to use when communicating the helpline to someone who is uncomfortable with mental health services.
- Indicator variable for preferring compensation framing over “want to share” framing.
- Indicator variable for preferring “targeting my friends” framing over general friends framing.

Accuracy of recipient’s self-assessment of the probability that she has anxiety or depression: An indicator variable taking 1 if and only if the recipient states she believes it is “likely” or “very likely” that she has experienced anxiety or depression in the past 2 months AND her PHQ-9 score or GAD-2 score indicates that she likely has depression or anxiety, using standard cutoffs.

Recipient’s opinion that she should be able to solve her own mental health problems: Standardized score of likert variable, standardized using the mean and standard deviation of Framing 1.

Recipient’s opinion that religious solutions are the best way to solve mental health problems: Standardized score of likert variable, standardized using the mean and standard deviation of Framing 1.

Heterogeneities to be tested
- Sender prior use of MH services
- Sender wants to use the advertised helpline
- Sender treatment status in prior experiment
- Sender’s guess of the recipient’s distress level
- Recipient’s depression and anxiety scores
- Sender’s depression and anxiety scores
- Recipient prior use of mental health services
- Original framing under which the sender selected the recipient to get the information

The experimental design is as follows:
1. Participants surveyed about their social network
2. Participants randomized to a particular framing/wording to introduce the counseling helpline
3. Participants asked which friends they want informed about the helpline if using the assigned framing.
4. Participants offered a financial incentive if they are willing to have the helpline introduced using different framings.
5. Recipients (friends of the original participant) who the original participant agreed to have contacted are contacted and introduced to the helpline using a randomly selected framing from within the framings the original participant consented to.
6. Recipients’ take-up of the helpline is measured.

Participants’ close social network will be elicited by phone survey. Participants will then be randomized to one of the following four framings, which they will be told is how the helpline could be introduced by the study to their friends if they want the study to inform their friends about the helpline:
- F1: We may tell her that the project offered to compensate you financially to help us check if your friends are interested in this helpline, and you mentioned she is one of the people you know.
- F2: We may tell her the project offered to compensate you to help us check if your friends are interested in this helpline, and you indicated she or people she knows might benefit from the information.
- F3: We may tell her you wanted us to check if your friends are interested in this helpline, and you mentioned she is one of the people you know.
- F4: We may tell her you wanted us to check if your friends are interested in this helpline, and you indicated she or people she knows might benefit from the information.

The participant then selects which of her friends, if any, she wants the study to inform about the helpline if using that framing.

Participants are then offered a financial incentive if they are willing for the study to introduce the helpline to their friends in the 3 other ways.

Recipients, that is the original participant’s friends, are then randomized to receive information about the helpline using one of the framings that the original participant agreed to.

Recipients are called, read the introduction, and asked if they want to use the helpline. Conditional on expressing interest, the recipient is screened for eligibility based on her mental health vulnerability, and if found to be eligible then her contact information is shared, with her explicit consent, with the helpline to be contacted for counseling.

Randomization done in office by a computer.

Senders are randomized across all 4 framings at the sender-level.
Recipients are randomized to framings at the recipient level, randomizing across the framings to which the sender consents.

Yes

700 senders, with an expected average of 3 friends named per sender.
The design of sender willingness to each recipient is clustered at the recipient level.
2100 recipients: Recipient treatment is not clustered.

700 senders, 2100 recipients

175 senders per framing arm
525 recipients per framing arm

The binary variable of whether each friend is selected by the sender to receive the helpline information (constructed at the recipient level): MDE is 8.8 percentage point difference on a base sharing rate of 80%, a 10.1 percentage point difference on a base sharing rate of 70%, and an 11.5 percentage point difference on a base sharing rate of 60%. Take up of the helpline by the recipient (binary variable): 6.4 percentage point difference on a base take-up rate of 80%, 8.6 percentage point difference on a base take-up rate of 50%.