Derivative step sensitivity grid
StatMedAnalysis | 14 November 2026

Compare finite differences across step sizes.
Keep the evaluation method fixed, halve the step and repeat. Check domain boundaries. Agreement across steps is a sensitivity check, not proof of exactness.
Use synthetic or aggregate values only. Duplicate the blank row for additional records.

Location | Step | Left evaluation | Center evaluation | Right evaluation | Forward | Centered | Change
________ | ________ | ________ | ________ | ________ | ________ | ________ | ________

Fictional example: For x squared at x=2 with step 0.1, evaluations are 3.61,4,4.41; centered slope is 4.

Completed by / version / date: _____________________
Unresolved discrepancies and action: _____________________