Research
Digital mental health has an engagement problem it can't solve, partly because it can't agree on what "stopped using it" means. My work sits on that fault line, between how people actually behave and how we count it.
Primary
The acceptance, use and avoidance of mobile technologies in support of mental health and wellness, with particular attention to what happens after adoption.
Related
Computational and quantitative text analysis of user-generated content, reviews, comments, and public discourse around mental health and suicide prevention.
Related
Mobile device user experience, digital affordances, and the design of community crisis infrastructure that people can actually reach.
The dissertation
Investigating Post-Adoption Abandonment of Mental Health Mobile Applications Among Young Adults. A two-study, mixed-methods dissertation completed at the University at Albany in 2025 and awarded the program's Distinguished Dissertation Award.
Study 01
Topic modeling and semantic text classification applied to a corpus of user-generated MHMA reviews to surface the prominent inhibitors and enablers of sustained engagement, in users' own language, at scale, outside a lab.
Study 02
A cross-sectional survey testing the relationships between inhibitors, enablers, user satisfaction, perceived usefulness and abandonment, including a mediation model, and a "graduation effect" in which users stop because the app worked.
Three findings that change the practical advice
01
They're orthogonal dimensions, not two ends of one satisfaction scale. Removing friction and adding value are separate jobs, and doing one well does not substitute for the other.
02
Inhibitors retain a direct effect on abandonment even after controlling for satisfaction and perceived usefulness. A user can look satisfied on every dashboard you have and still be leaving because of accumulated friction.
03
The "graduation effect": people who discontinue because their condition improved and they no longer need the app. In standard engagement metrics they are indistinguishable from a frustrated exit. They should not be.
Four user trajectories Derived taxonomy
The outcome the product was designed for. Low inhibitor load, sustained use.
The graduation effect. Leaves because the intended outcome was achieved, a success that looks like churn.
Habit and inertia offset real friction. Retained, but fragile, and quietly costly to the experience.
The genuine product failure. The only quadrant most retention dashboards are actually built to catch.
The taxonomy is independently recoverable from survey items that were never used in the original hypothesis testing, which makes it a finding in its own right, not a narrative gloss on the mediation results.
Working papers
Under development · 2026
Operationalizing Post-Adoption Abandonment in Mental Health Mobile Applications: A Cross-Sectional Survey Study.
The same phenomenon carries at least six distinct labels across five scholarly traditions, and the same labels get applied to behaviors that differ in clinically meaningful ways. This paper maps that mess, then demonstrates its cost empirically: five defensible operationalizations applied to a single sample of 314 people produce abandonment prevalence estimates ranging from 35.7% to 66.9%, a 31-point spread driven entirely by measurement choice, not behavior.
It closes with a proposed consensus definition that separates continuous disengagement from discrete termination, excludes non-adoption and goal-completion cessation, and evaluates use against both developer-intended and user-intended thresholds.
Why it matters
If the number you report depends more on your definition than on your participants, the literature can't accumulate. Studies stop being comparable, meta-analyses inherit the noise, and product teams get benchmarks that mean nothing.
Practical recommendation: report your operationalization explicitly, use at least two measures capturing both empirically distinguishable forms, and collect intended use prospectively at adoption.
Peer-reviewed
Harris, D. (2025). Electronic Theses & Dissertations (2024–present), 282. University at Albany.
Scholars ArchiveHarris, D. (2024). International Journal of Human–Computer Interaction, 1–17.
View publicationHarris, B. R., Harris, D., Flanagan, E., Mariani, A., & Hay, T. A. (2024). Community Mental Health Journal, 1–8.
View publicationHarris, D., & Krishnan, A. (2023). Journal of Health Communication, 1–10.
View publicationPublication links point to publisher DOI pages where available. Full texts of open-access work are also available on request.
Presentations & invited lectures
National conferences, federal policy academies and state convenings, usually on digital engagement, crisis system design, or the gap between what research shows and what practice does with it.
Teaching
Adjunct Instructor, University at Albany, Spring 2021 – Spring 2023. Building and evaluating arguments, gathering and weighing evidence, and presenting conclusions in the context of emergency preparedness, homeland security and cybersecurity.