Elicitation
What was asked, and what was answered, when weights were elicited by the swing method, together with the ranges the answers were given against.
The ranges are the point of keeping this. A weight in an additive model is a scaling constant tied to a range: saying cost matters twice as much as delay means the swing from the worst cost to the best cost is worth twice the swing from the worst delay to the best delay. Change either range afterwards and the answer no longer means what the person who gave it meant, even though the number is unchanged. Recording the ranges is what makes that detectable rather than silent.
Parameters
the metrics from the most valuable swing to the least
each metric's swing rated against the top one, which is fixed at 100
the range each metric was measured over when the ratings were given
the metrics whose range the study permits to be refitted, and whose weights are therefore the ones at risk of being invalidated later
Constructors
Functions
A description of why these weights no longer apply to snapshot, or null when they still do. Written to be shown to whoever gave the answers.
Indicates whether these weights still mean what they meant, given what snapshot evaluated over. When this is false the weights should be discarded rather than applied to ranges they were not elicited against.
The metrics whose range has moved since the weights were elicited, comparing against the ranges snapshot actually evaluated over.