Adaptive Conjoint Analysis

ACA is a hybrid, adaptive interview that starts from self-explicated importance and desirability ratings and then asks graded paired comparisons chosen to be near-ties, updating each respondent's utilities in real time. Choice-based conjoint instead shows fixed sets of full profiles and asks respondents to pick one, analyzed with discrete-choice logit. ACA's advantage is handling many attributes and producing individual-level utilities from a short interview; its disadvantage, highlighted by Green, Krieger, and Agarwal, is that its additive, self-explicated foundation models price and strong interactions less faithfully than the choice format, which is why CBC is preferred for pricing.

The self-explicated stage gives ACA a complete initial estimate of every respondent's utilities cheaply, which is what allows it to scale to many attributes and to choose informative trade-off pairs from the very first paired comparison. Without these priors the adaptive selection would have nothing to adapt to. The cost, as Green, Krieger, and Agarwal noted, is that stated importances are imperfect, so ACA anchors but does not fix the utilities to the priors, letting the subsequent paired comparisons revise them. The balance between trusting the priors and trusting the trade-offs is a key design choice in any ACA study.

Generally no. Because ACA builds utilities additively from self-explicated components and graded trade-offs, it tends to understate the strong, sometimes non-compensatory role that price plays in real purchase decisions. Green, Krieger, and Agarwal explicitly cautioned against using ACA for pricing, and practitioners typically turn to choice-based conjoint, Gabor-Granger, or Van Westendorp methods when price is central. ACA's sweet spot is many-attribute product-design studies where the goal is to map feature preferences rather than to estimate a precise demand curve or willingness to pay.