. The project will test changes to Ticket notices including the types of notices sent, the language used, and the timing of the notices.

Primary outcome: Beneficiary “assigns a Ticket”, which constitutes an agreement between
the beneficiary and a service provider of their choice for employment support services
1
(9
months post mailing)
● Secondary outcome: Beneficiary (or representative payee) ever calls the Ticket Help Line
(9 months post mailing)
We will be drawing from four major SSA administrative datasets (i.e., the Disability Control File
(DCF), the internet Ticket Operations Provider Support System (iTOPPS), the Supplemental
Security Record (SSR) and the Master Beneficiary Record (MBR)
2
) and contractor data to collect the following information:
● Data to determine whether an individual assigns a Ticket: ENs and VRs assign Tickets using the Ticket Portal, which is a secure portal that feeds directly into iTOPPS.
● Data to determine whether an individual callsthe Ticket Help Line: SSA’s Ticket Program
Manager (TPM) tracks this information via iTOPPS. TPM will provide a monthly report to SSA’s Office of Research, Demonstration, and Employment Support that tracks whether a beneficiary in our study has called the Help Line and which notice they received.

We will estimate the causal effect of the treatment (more specifically, intention to treat) using differences of means calculated using ordinary least squares (OLS) regressions of two binary outcomes (Ticket assignment and Help Line calls) on treatment assignment. We will estimate the differences in means for two comparisons for each experiment—one for each experimental factor, the so-called main effects. For each of the three experiments, we conduct a separate analysis, each looking at different exposure to the intervention based on cohort. This means we will test for instance, the 1-year cohort’s response to a 1-year optimized notice that is separate from the test of the start-up cohort’s response to a 1-year optimized notice.
We will be estimating main effects of the optimized notices (factor #1) and the inclusion of a cardstock Ticket (factor #2) in the first two cohort-based experiments. In the 3rd experiment, we will examine the receipt of an optimized notice plus cardstock Ticket (as a single factor) versus no notice or cardstock Ticket. In the following, we differentiate our confirmatory analyses by each of the three experiments.

We anticipate no exclusion of beneficiaries. Medical improvement not expected (MINE) diary beneficiaries (other than those with deafness and blindness diagnoses) as well as beneficiaries that have met listing codes for neoplasia and certain neurological conditions are excluded from receiving the TTW notices all together.

If beneficiaries die within 9 months of receiving a respective intervention, they will be excluded from our analyses. Other than that, there should be no missing data because all outcomes, including death, are observed by SSA. Additionally, we do not anticipate significant missing data in the imported variables.