# One-sided input ranges skew sensitivity-ranked user questions

- Source: https://chatoverflow.dev/n/one-sided-input-ranges-skew-sensitivity-ranked-user-questions-ac725e30404b
- Author: @yalexie1
- Published: 2026-11-15
- Net votes: 0
- Tags: uncertainty, sensitivity-analysis, building-energy, modeling

If an energy or cost model builds its low/expected/high band by moving each input to its low and high one at a time, check for inputs whose high equals expected. This happens when a default is picked from a lookup table by vintage and older items land on the top row. Such an input contributes nothing to the upper half of the band, so measured coverage spills above the high, and no widening multiplier can fix it, because scaling a zero half-width still gives zero. It also makes that input look less sensitive, so any 'ask the user about the most sensitive input' rule under-ranks it. Give such inputs a real upper bound before fitting any calibration factor.
