Using Financial Incentives and Screening to Increase Engagement with Mental Health Services among College Students in Chennai, India

Treatments are randomized at the individual level, stratified by categories of PHQ-ADS scores. Treatment status will be revealed via the survey administered through students' mobile phones. Participants will be randomized to four conditions:

1. Control: the control group will (a) watch a video introducing them to the mental health services available in the college, including introductory videos of the counselors, (b) have access to therapy at no cost for the duration of the study via their college counselors and additionally at SCARF, a mental health center in Chennai, and (c) be given links to book a therapy session or download a mental health app.

2. Screening: in addition to the interventions in the control group, students in the screening group will receive feedback on their PHQ-ADS score and recommendations based on that. In particular, they will be informed whether their responses to the PHQ-ADS questions fit in the standard risk categories of no, mild, moderate, or severe distress and corresponding recommendations for whether to make an appointment with a counselor/therapist.

3. Rewards: in addition to the interventions in the control group, students in the rewards group will be provided with a cash payment of Rs.\ 500 (USD 5.70 at market exchange rates) for attending a (first) therapy session.

4. Screening and rewards: students in this group will receive both interventions, i.e., both the screening information and the reward described above.

For ethical reasons, participants with PHQ-ADS scores of 30 or above (indicating severe risk of depression and anxiety) in the non-screening treatment arms will also be informed of their scores at the end of the survey. For outcome variables measured within the survey, we will treat these participants as though they are part of the group to which they are randomly assigned. One outcome variable (completing the process to make an appointment for therapy) is measured after the conclusion of the participant survey. For this outcome variable, we will separately run regressions both including and excluding these individuals.

2025-01-24

2025-04-30

We have two primary outcomes:

1. Expressing interest in therapy. This consists of students answering ``yes'' when asked in the survey if they would like to book an appointment with a counselor, or if students answer ``yes'' when asked in the survey if they would like to be contacted by a study counselor to discuss mental health and therapy.

2. Making an appointment. If a student actually makes an appointment with a counselor (measured using the administrative records from the counseling offices)

The secondary outcomes of the study include the following:

1. Take-up of mental health app. This is measured in two ways: participants expressing interest in accessing a mental-healthcare app in the survey, and participants actually clicking a link to download the app.

2. Demand for information about mental health. This is measured in two ways: participants expressing interest in receiving more information about mental health, and participants actually clicking a link to access the information.

3. Attendance of therapy sessions. We will collect data from the administrative records of the counselling office on whether participants actually attended a therapy session, as well as the number of sessions attended.

4. Other variables. We will also report exploratory analyses and mechanism tests including: (i) whether participants report in the survey if they would consider seeking therapy in future; (ii) what steps they report they would take to take care of their mental health in future; (iii) reported motivating/inhibiting factors for taking up therapy and perceptions of therapy.

Individuals will be randomly assigned to treatment arms.

Treatments are randomized at the individual level, stratified by categories of PHQ-ADS scores. Treatment status will be revealed via the survey administered through students' mobile phones.

Individual

No

We will attempt to visit all classrooms (i.e., all majors and years) over the course of the study. However, we rely on the college administration to make students available and scheduling constraints on the part of the college may keep us from visiting all classrooms. In addition, attendance rates on a given day are well below 100\% (and as low as 50% during our scoping visits). Thus, the final experimental sample will be substantially smaller than the full college student population of approximately 3000.

Approximately 3000 students, subject to the discussion above

Equal shares in each treatment group

Assuming 3000 students in total, and a 10% level of take-up in the control group, the minimum detectable effect size is 3.28 percentage points

Pre-Analysis Plan