Proposals, not results.
Seven studies we would like someone to run, each scoped so that a single motivated research group could execute it inside one academic year.
Each names honestly what we can contribute and what we cannot. S1 and S2 are the two halves of the question the research brief opens with. S6 and S7 are newer, and follow from the involuntary signals and the generated items the brief describes.
How the programme is scoped
One faculty, one semester, one honest question.
Nothing here needs a new instrument built. Every study below draws on signals a Diaposi session already records, which is what makes a single academic year a realistic envelope rather than an optimistic one. What none of them can do without is an outcome measure we do not own: an assessment, a validated comparison instrument, an approved protocol. That asymmetry is the whole reason the programme is open.
We can supply the instrument and the rows. We cannot supply the ground truth, and a study that lets us do both is not worth running.
(1)The seven studies
Instrument validation
Do slide-level signals predict end-of-unit assessment beyond what attendance alone explains?
You bringOne faculty, one semester, access to assessment outcomes.
We bringThe full signal stream, per slide and per student, plus raw export.
The closed-loop trial
Randomise sections: the teacher sees the live pulse, or does not. Does the feedback loop itself move outcomes?
You bringRandomisation design, ethics approval, assessment instrument.
We bringFeature-flagged builds, reteach latency, per-slide intervention traces.
The validity triangle
How far apart are dwell traces, self-reported comprehension, and graded outcomes, and when do they diverge?
You bringPsychometric expertise and a validated comparison instrument.
We bringTrace and self-report captured natively, on the same timeline, same session.
Modality comparison
Do in-person, online and hybrid rooms produce structurally different engagement signatures?
You bringCourses running in more than one modality.
We bringModality is already a field on every session. No new instrumentation needed.
Voice and equity
Does anonymity change who participates, not just how much? Compare question authorship against hand-raising baselines.
You bringCohort demographics under an approved protocol.
We bringAuthorship distribution across anonymous questions and upvotes.
The volunteer-bias test
Do the involuntary signals find the students the voluntary ones miss? Look-backs and pace flags need no opt-in; taps do. If the two channels disagree, the disagreement is the finding.
You bringA comprehension instrument and a cohort willing to sit it.
We bringBoth channels, on one timeline, per slide, for every student in the room.
Machine-written items
Are AI-drafted questions as discriminating as teacher-written ones, and does the slide they were drawn from predict how they perform?
You bringItem analysis expertise and a teacher-authored control set.
We bringSource slide, difficulty label, and the full response distribution for every generated item.
Starting one
The shortest route in costs an hour.
Any of the seven can begin with a single instrumented lecture in your own course: you keep every row of the data and co-author the write-up. The partnership routes set out the three ways in, including the doctoral one, where the French CIFRE scheme funds a three-year contract between a laboratory, a company and a candidate14.