Fixed item sets
Each skill-and-seed identity appears in one Wave 1 form only. Generator changes require a new content version and wave manifest.
VERSION-FROZEN PILOT CONTROL
Same evidence.
Fairer forms.
Assign students without names or accounts. Every six consecutive participant slots distribute one student to each fixed form; item order and answer position then counterbalance reproducibly across cycles.
STEP 01 · ASSIGN
Use the next anonymous slot.
Record only the slot in your private roster. Do not enter a student name, email, school, or other identifier into DAT TRAIN.
ASSIGNMENT
FORM ASlot 1
PAT-W1-2026A · cycle 1
https://dattrain.com/?pilot=PAT-W1-2026A&participant=1The slot is not included in the exported pilot CSV.STEP 02 · CONTROL
Keep the wave auditable.
The manifest freezes every form’s item identities. The blank 120-slot roster distributes 20 students per form and completes the answer-position counterbalance cycles.
DESIGN LOGIC
What is controlled—and what is not.
Balanced assignment
Slots rotate A through F. Sequential recruitment therefore keeps form sample sizes within one participant.
Counterbalanced display
Within-family item positions rotate by cycle. Correct answer positions cycle deterministically while stable option IDs preserve scoring.
No false equivalence
Forms cover matched observable structures, but no form is called easier, harder, or score-equated before expert review and pilot evidence.
WAVE MANIFEST
Six forms. No repeated item identities.
Every form follows official PAT family order. Expand a card to inspect its non-secret skill and seed manifest.
FORMA24 items · 16 min
Keyholes1 · 2 · 3 · 4
TFE20 · 23 · 30 · 42
Angle ranking42 · 46 · 39 · 44
Hole punching58 · 59 · 60 · 62
Cube counting77 · 81 · 78 · 80
Pattern folding96 · 97 · 98 · 100
FORMB24 items · 16 min
Keyholes6 · 5 · 11 · 12
TFE21 · 24 · 43 · 31
Angle ranking58 · 53 · 40 · 45
Hole punching61 · 63 · 67 · 64
Cube counting86 · 83 · 79 · 84
Pattern folding99 · 105 · 101 · 102
FORMC24 items · 16 min
Keyholes7 · 9 · 14 · 22
TFE22 · 28 · 44 · 32
Angle ranking60 · 54 · 41 · 52
Hole punching65 · 66 · 69 · 74
Cube counting97 · 88 · 82 · 98
Pattern folding109 · 112 · 103 · 108
FORMD24 items · 16 min
Keyholes15 · 29 · 10 · 17
TFE27 · 39 · 46 · 34
Angle ranking76 · 55 · 43 · 57
Hole punching70 · 68 · 76 · 75
Cube counting99 · 89 · 85 · 107
Pattern folding115 · 117 · 104 · 110
FORME24 items · 16 min
Keyholes8 · 13 · 18 · 23
TFE29 · 47 · 33 · 52
Angle ranking84 · 61 · 47 · 65
Hole punching71 · 72 · 84 · 77
Cube counting91 · 93 · 87 · 108
Pattern folding119 · 125 · 107 · 114
FORMF24 items · 16 min
Keyholes16 · 42 · 44 · 24
TFE35 · 40 · 50 · 36
Angle ranking86 · 63 · 48 · 66
Hole punching78 · 73 · 97 · 79
Cube counting106 · 102 · 95 · 90
Pattern folding120 · 131 · 111 · 127
PILOT CONDUCT
Protect the evidence.
- Assign once.Give each participant one slot and one first attempt unless your protocol explicitly studies retesting.
- Use comparable conditions.Record device class and interruptions outside DAT TRAIN without collecting unnecessary personal information.
- Export promptly.After submission, open My progress and download the pilot CSV before clearing browser history.
- Keep decisions human.Use the analysis workspace to flag items for expert review—not to auto-retire questions or manufacture a DAT score.