Data Format

Data Format

Standard CSV (crossed GR&R)

Part,Operator,Trial1,Trial2,Trial3
P01,Alice,10.0089,10.0107,10.0097
P01,Bob,10.0116,10.0099,10.0094
P01,Carol,10.0071,10.0083,10.0078
P02,Alice,9.9812,9.9825,9.9819

Required columns: Part, Operator, Trial1, Trial2 (minimum two trials) Column names are case-insensitive.

Supported dimensions

Dimension	Supported
Trials	2, 3, 4, 5
Operators	2, 3, 4, 5
Parts	2–10

AIAG standard design: 10 parts × 3 operators × 3 trials = 90 measurements

---

Attribute CSV (go/no-go)

Trial values must be 0 (fail) or 1 (pass):

Part,Operator,Trial1,Trial2,Trial3
P01,Alice,1,1,1
P01,Bob,1,0,1
P02,Alice,0,0,0

Reference CSV for effectiveness calculation:

Part,Reference
P01,1
P02,0

---

Run order column

Add a RunOrder column and pass --run-order RunOrder:

RunOrder,Part,Operator,Trial1,Trial2,Trial3
1,P01,Alice,10.0089,10.0107,10.0097
2,P01,Bob,10.0116,10.0099,10.0094

---

Study tips

Part coding — Re-code parts (P01–P10). Operators must not see original IDs to prevent memory effects.

Part selection — Span the FULL production tolerance range. FDA, IATF 16949, and NADCAP auditors specifically check for cherry-picked parts.

Trial randomisation — Measure in randomised order for each trial pass.

Operator isolation — Operators must not see each other's readings. Complete all Trial 1 measurements before anyone starts Trial 2.

Environment — Run under production conditions (temperature, vibration, fixturing). Controlled lab conditions produce optimistic results.

Recommended sizes:

Context	Design
AIAG baseline	10 × 3 × 3
Life-sustaining devices	10 × 3 × 5
Aerospace flight-critical	10 × 5 × 3 or larger
Clinical labs (CLSI EP05-A3)	2 replicates × 2 runs/day × 20 days

---

Generating sample data

python grr_tool.py --generate-sample --input sample.csv

Writes a synthetic 10×3×3 CSV with realistic operator biases and gage noise. Use as a template for data collection sheets.