The study needed a credible workflow.
This was research infrastructure, not a one-off study setup. Before the team could evaluate AI reading support, it needed a credible way to recruit, verify, schedule, and support the participants the study was designed for.
DITAVA Lab was preparing a study with DHH adults fluent in ASL to evaluate 18 AI-based text styling designs for skimming support. Prior work focused on hearing participants; this study tested whether those directions held for a DHH population.
Prolific could filter for self-reported "hearing impairment," but the study needed more than a broad population flag. It needed evidence of ASL comprehension, a way to route participants by communication preference, and an IRB-compliant path from first contact to session.
No verified DHH/ASL participant pool, no screener, no interpreter routing workflow, and no standardized participant messaging.
Define what evidence was sufficient for study inclusion and what access conditions had to be in place before a session could run.
The wrong pool would weaken the evidence.
If the pool did not match the intended population, the team could not interpret DHH preferences with confidence. If access logistics were handled too late, eligible participants could still be blocked from participating.
Confident interpretation required knowing participants were DHH adults fluent in ASL, not just anyone who self-reported hearing loss. The lab also needed a repeatable system for future DHH studies, not a one-time recruiting push.
Separate eligibility from access.
Recruiting participants, verifying eligibility, and making sessions accessible are different jobs. Treating them separately made the workflow easier to audit, defend, and reuse.
First, Prolific surfaced people who self-identified as DHH or reported hearing loss. That identified the right broad population segment, but it was not enough for inclusion.
Next, the Qualtrics screener tested receptive ASL comprehension with two short videos and open-ended questions. Submissions then moved into a review tracker with Prolific ID, completion status, correctness rating, and a written note for every decision.
Finally, verified participants were routed into the main study. At scheduling, they selected English or ASL, which determined the Calendly path, interpreter lead time, and participant messaging they received.
-
Prolific pre-filter
Broad targeting based on self-reported DHH identity or hearing loss.
-
Qualtrics ASL screener
Behavioral verification using two ASL video comprehension tasks.
-
Manual review tracker
Google Sheets decision log with Prolific ID, completion status, correctness rating, and review notes.
-
Calendly scheduling and routing
Booking flow that routed English and ASL sessions into different lead-time paths.
Eligibility needed low-burden proof.
The workflow held because each decision was tied to clear evidence criteria. The goal was to protect study validity while keeping the process privacy-conscious and manageable for participants.
-
Test receptive ASL comprehension
Expressive verification would have required participants to submit videos of themselves signing. That added privacy concerns and unnecessary participant burden. Receptive comprehension gave us the eligibility signal without requiring participant video.
-
Use open-ended comprehension
The screener asked participants to interpret signed videos in their own words. That created a stronger signal than self-report alone while avoiding assumptions that every ASL-fluent DHH person has the same relationship to Deaf culture.
-
Keep human review for ambiguous responses
Some responses were partial, culturally informed, or inconsistent. I used the Notes column as a decision log, flagged uncertain cases for PI review, and kept the reasoning visible instead of forcing every response into an automated pass/fail rule.
-
Capture access needs before scheduling
ASL interpretation could not be arranged at the start of a session. The workflow used a 1-day scheduling buffer for English sessions and an 8-day buffer for ASL sessions, building interpreter lead time into the study design.
Make access part of protocol.
The recommendation was a five-stage workflow: screen broadly, verify carefully, document decisions, route by communication preference, then run sessions only when access support was in place.
-
Prolific outreach
A screener study was posted on Prolific targeting participants who self-identified as DHH. Those who clicked through were taken to the Qualtrics consent form and screener survey.
-
ASL comprehension verification
Participants watched two ASL videos and answered open-ended questions. I reviewed each response, marked it correct or incorrect, and wrote notes explaining the decision.
-
Pool construction
Verified participants became a named Prolific pool the lab could contact for future DHH studies.
-
Study enrollment and scheduling
The main study was launched to verified participants. At booking, participants selected English or ASL as their session language. That choice routed them to the appropriate Calendly path and triggered interpreter coordination for ASL sessions.
-
Session delivery
English-preference participants completed sessions verbally. ASL-preference participants had interpreter coordination built into scheduling. Participant communication used approved templates from first contact through reminder.
Decision Impact
A verified workflow became reusable.
- Participant pool
- 30+ screener responses reviewed; verified DHH/ASL pool built on Prolific, reusable for future DHH/ASL recruitment
- Session routing
- English and ASL sessions routed at booking; 1-day buffer for English, 8-day buffer for ASL with interpreter coordination confirmed before sessions ran
- Infrastructure
- 5-stage workflow documented: Prolific outreach, ASL screener, manual review, pool construction, scheduling and delivery
The study could run with verified participants and the right communication path in place before each session. The lab also gained a foundation it could reuse for the next DHH study. Access logistics became research infrastructure.
The screener held; the tracker lagged.
Receptive comprehension was the stronger evidence choice
It gave the team an eligibility signal without increasing privacy risk or making the screener more burdensome.
The tracker needed automation earlier
Manual review was accurate but time-intensive across 30+ initial screener submissions. A stronger intake flow would use form logic to surface incomplete responses first, so human review could focus on substantive judgment calls.
Access logistics are research design decisions
When to capture communication preference, how to route sessions, and how to write consent language are not administrative details. They determine who can participate and whether the data reflects the population the study intended to reach.