Improving Immigrant Mental Health through a Digital App

This study will experimentally assess the effects of offering access to an emotional well-being and mental health phone application, which includes an AI-powered chatbot, among Hispanic immigrants in the United States. Additionally, we will examine the impact of incentives in promoting adoption of the app.

This study will experimentally assess the effects of offering access to an emotional well-being and mental health phone application, which includes an AI-powered chatbot, among Hispanic immigrants in the United States. Additionally, we will examine the impact of direct incentives or a lottery in promoting adoption of the app. In addition, participants will be randomized into 2 or 4 months of access to the app.

2024-05-20

2025-01-01

Symptoms of depression, anxiety, and stress, measured by the PHQ-8, GAD-7, and PSS.

1. Parental expectations and aspiration about children’s educational attainment.
2. Migrant remittances.
3. Social connectedness and integration: integration into the U.S. (questions on social links with natives, general knowledge about US, knowledge about how to use public services), revealed-choice measures of their demand to invest in learning English (by giving them a choice between access to language apps or cash).
4. Labor outcomes: employment status,earnings, and, absenteeism

Access to the premium version of the app and the duration of access will be randomized among eligible individuals.

Individuals will be recruited through social media and/or other online advertisements. The target population will be: immigrants to the U.S. of Hispanic origin whose native language is Spanish, who have been living in the U.S. (intermittently or continuously) for less than 10 years and who are over 18 years of age.

A short screening questionnaire will be implemented with interested participants, to assess their interest and ensure they meet the inclusion criteria. Those who do will be invited to complete a baseline online survey. After completing the baseline survey, individuals will be randomized to a treatment and a control arm. Treated individuals will receive a code that they can use to access the premium version of the app.

We will also cross-randomize sub-treatment arms, but to ensure statistical power, the primary analysis will pool all treatment arms.
For exploratory hypotheses, we will randomize within the treatment arm: app access duration – T1( app access for 2 months) and T2 ( app for 4 months); app use incentives – T3 (no incentives for app use), T4 (direct incentives for app use), and T5 (lottery incentives for app use). We will cross randomize app access (T1 and T2) with incentives (T3,4,5).

We will conduct follow up online surveys at around the one month, 3-month and 6-months mark.

Randomization to be done in office by a computer

individual

No

2,100 individuals

2,100 individuals

We will work with a total sample of 2,100 individuals. Individuals will be assigned to a treatment group (1,400) or a control group (700). Within the treatment arm, individuals will be cross-randomized to either receive access to the app for 2 months or 4 months (700 per group) and to receive incentives for app use (932) or not (468).

For pooled specifications with 1,400 individuals in the treatment arm and 700 in the control, we are powered to detect MDEs of 0.13 s.d. between treatment and control arms (power= 0.8, alpha value = 0.05). With a take-up of 90%, this would translate to MDEs of 0.14 s.d. and with a take-up of 80% to MDEs of 0.15 s.d. For comparisons between subtreatment arms for length (each n=700) we will be able to detect differences of 0.15 s.d. and for comparisons between incentives and no-incentives we will be able to detect differences of 0.16 s.d.