Matched pairs, not random labels
The six students form three pairs, and exactly one child in each pair is flagged at random. Without the pairing, a difference in treatment could always be blamed on the student rather than the label.
Scenario module · in production
The Screening
Fifteen minutes. You teach a class of six students. Before the lesson, three of them are flagged by a “Cognitive Potential Screening”. You run the lesson, make the small decisions teachers make, and at the end of the year the flagged students have done better. Then the module shows you why.
- Role
- Design, build, measurement
- Tools
- Storyline 360, JavaScript
- Output
- SCORM 1.2, 15–18 min
The problem
Awareness training about bias has a poor record. Single-session interventions rarely produce measurable behaviour change, the small effects they do produce tend to fade, and in some studies telling people not to be biased made things worse. The dangerous part is what is left behind: participants who now feel informed.
So a module on this subject has two ways to fail. It can be ineffective, or it can be ineffective while feeling effective. I designed this one to make the second failure impossible: it tells the learner, in its own words, that it will not change them, and then hands them something concrete to do instead.
The experience is short. The debrief is long.
The literature is blunt about simulations: the scenario swells and the debrief gets squeezed into whatever time is left. For behavioural subjects that ratio has to be inverted. Here the lesson takes about six minutes and the debrief and practice take thirteen.
Act 1–3
You run the class
~6 min
Act 4
You meet your data
~2 min
Act 5
What the research says
~5 min
Act 6–7
Tools, second round
~5 min
Measurement layer
Seven measures, none of them announced.
Nothing in the lesson says it is being recorded — a visible timer would change the very thing it measures. JavaScript inside the module records each decision, pairs the flagged student with the unflagged one, and the debrief plays the comparison back.
| id | measure | how |
|---|---|---|
| M1 | Turns to speak | who gets picked, five times |
| M2 | Wait time | milliseconds, no visible timer |
| M3 | Task difficulty | who gets the hard worksheet |
| M4 | Feedback type | consoling, explanatory, dismissive |
| M5 | Break-time attention | who you talk to |
| M6 | Attribution | same behaviour, two students |
| M7 | Grading | identical answer, two names |
The sharpest one is M7. Two students hand in the same sentence, word for word, and you grade it twice. Whatever difference appears between those two numbers is difficult to explain away.
Four decisions worth defending
Notes that read both ways
“Finishes tasks before everyone else” can mean quick or careless. “Doodles while listening” can mean focused or absent. Every note is written to carry both readings, so the label in your hand decides which one you see.
Wrong answers a smart adult would give
The misconceptions in the lesson are documented ones — summer is warmer because the Earth is closer, the Moon does not rotate. If the wrong answers were silly, the learner would dismiss the student instead of teaching them.
The module admits its own limits
One screen states plainly that finishing this module will not reduce anyone's bias, and that the only thing which changes is what you do tomorrow. Saying anything else would reproduce the harm the research describes.
Where it stands
Content, script and storyboard are finished in English and Turkish. The module is being built in Storyline now; the first release covers the lesson, the reveal and the data comparison. Theory, tools and the second round are designed and follow after. When it ships, it will be playable on this site with its tracking data visible beside it.