Conjoint Attribute Importance
Four attribute spans open from the zero utility rule towards their best and worst levels in turn, and a rank caret drops onto the widest span to name the attribute that matters most.
Four attribute spans open from the zero utility rule towards their best and worst levels in turn, and a rank caret drops onto the widest span to name the attribute that matters most.
An attribute-by-level grid of four rows and five cells holds the design space, and one cell in each row takes an accent chip in turn until the profile on test stands complete, before the lower three release.
Two brand lines run across the feature axis close to parallel until the lower one pivots on its no-feature end and climbs past the other, and the crossing point is ringed.
A new product rises into the empty slot at the right of the simulated market and the three incumbent share columns each give up height in turn to make room for it.
A marker runs along the price axis and the share-of-preference step curve is uncovered behind it, each price point landing as the sweep reaches it until the whole demand schedule stands.
Comparison arrows fill in row by row between two profiles and every one points the same way, so the losing profile is struck through and dimmed out of the design as a task that teaches nothing.
An audit frame steps across the attribute columns of a design matrix, and as it leaves each column that column's level count settles onto the target rule until the whole design stands balanced.
The last task row drops out of the estimation stack into the reserved holdout slot, and the model's predicted share then rises beside the observed one until a match tick lands between them.
A scan frame visits three profile cards in turn, declines all of them and lifts away, and the wide opt-out tile beneath is framed and ticked while the rejected profiles dim.
Part-worth bars grow out of a central zero utility line, some above it and some below, attribute group by attribute group, until the estimated set stands complete and then falls back to the line.
A cloud of respondent points opens away from the mean utility rule in both directions while the rule itself holds still, and two segment bars fill in beneath the axis as the split resolves.
One option is taken in each task down the column of paired profiles, and the two tally bars at the foot step outward from the centre line as the votes land until the preference has separated.
A lever swaps one attribute level on the first product and the market bar answers, widening that product's share while the two rival segments slide right and give ground to the simulated split.
Two profile cards stand side by side with three attribute rows each, and the selection frame moves from the left card to the right one, locking the choice with a tick while the declined profile dims before the next task opens.
A bracket on the utility scale holds the part-worth gap between two levels, and a marker runs out along the notched money rule until the span beneath it matches that gap and a lock tick confirms the price.