Members of CVS/Aetna will receive a direct mailing or receive no outreach at all.

Our design will be a 3-cell between-subjects design:

1. control: this will be a group who will not receive any outreach
2. how: this will be a group that will receive a mailer emphasizing only how to seek treatment for mental health issues
3. how and why: this will be a mailer emphasizing both how to get help and why they should get help for mental health issues (i.e., listing out how mental health treatment can improve their lives).

Within the how and how and why conditions, participants will be further divided into two different groups:

1. incentive ineligible
2. incentive eligible

The incentive ineligible will receive the how or how and why mailer, depending on their condition.

For the incentive eligible, they will again further be divided into the following two groups:

1. incentive holdout
2. incentive receive Those randomized into the incentive holdout group will receive the how or how and why mailer, depending on their condition. Those randomized into the incentive receive group will receive one of those two mailers, but it will also include an incentive if they seek out mental health treatment (if they visit a therapist within 30 days, they will receive a gift card).

Our primary dependent variables will be: 1) number of participants who visit the website to find a mental health appointment and 2) number of participants who submit a claim for an appointment to the insurance company (indicating they did in fact make an appointment and go).

For clicks to the website, we will conduct all analyses at 2 weeks and 4 weeks post mailer.

We will examine a claim for an appointment that was up to 3 months and up to 6 months post mailer. The company we are collaborating with has indicated that claims will appear about 2 months after the person takes an appointment. We will base these analyses on the date the appointment was taken.

We will conduct the same analyses for both of our DVs.

First, we will examine if there are any differences between the incentive receive and incentive holdout conditions depending on if they are in the how and how and why conditions. That is, we will examine if there is a 2(How vs. How and Why) x 2(Incentive vs. No Incentive) interaction. If there is no interaction, we will examine the main effects: How vs. How and Why and also incentive vs. no incentive.

If there is no main effect of incentive and no interaction with incentive condition, we will collapse across the two incentive eligible groups (hold out and receive) for all further analyses.

1. We will then compare the collapsed incentive eligible groups for both the how and how and why conditions to each other and to the control condition.
2. We will also compare the collapsed incentive eligible groups for both the how and how and why conditions to their respective incentive ineligible groups. If there is no significant difference, we will compare the collapsed how and how and why conditions to each other and to the control.

If there are differences, we will keep the two incentive eligible groups separate for all further analyses.

1. We will then run pairwise comparisons between the following seven groups:

1. Control condition
2. How incentive receive
3. How incentive holdout
4. How incentive ineligible
5. How and why incentive receive
6. How and why incentive holdout
7. How and why incentive ineligible

We will run all these analyses as either t-tests/chi square tests or linear/logistic regressions. This will be up to the preferences of the company we are working with as they will be conducting the analyses for us.

Inclusion Criteria:

• Participant must be a health insurance customer with the company we are working with.
• All other eligibility criteria are determined by the company and not disclosed to us. They will pick a sample of 60,000 eligible customers for the study, with participants then being randomly assigned to treatments they qualify for.

Exclusion Criteria:

• To be eligible for the incentive receive and holdout conditions, a participant must live in a state that allows these incentives to be mailed out by the company.

The company we are working with will be analyzing the data and informing us of the results. We will not have data to share.