Conjoint Analysis

Description

Estimates how much users value each feature and level by having them choose between systematically varied product bundles, yielding part-worth utilities and trade-offs.

Conjoint analysis infers how people value the attributes of a product from the choices they make between profiles that vary those attributes systematically. In choice-based conjoint, respondents see sets of two to four hypothetical products defined by attribute levels (features, price, service terms) and choose one, repeated over about 10 to 15 tasks. An experimental design ensures attributes are balanced and uncorrelated; hierarchical Bayesian or multinomial logit models then estimate part-worth utilities for each level. Because respondents must trade off, the method avoids the 'everything is important' problem of rating scales. Rigour lies in attribute definition grounded in qualitative research, efficient designs, holdout tasks to validate predictions, and adequate sample sizes (often 300+).

Attributes

Loss type
Active
Effort
Weeks

Steps

  1. DCE design:
  2. Select 4–8 design attributes and 2–4 levels each.
  3. (2) Generate an efficient choice design (D-optimal or orthogonal array).
  4. (3) Present participants with 8–12 choice sets, each comparing 2–3 design alternatives.
  5. (4) Fit conditional logit or mixed logit model.
  6. (5) Extract part-worth utilities for each attribute level.
  7. (6) Calculate willingness-to-trade-off: how much of attribute A is the user willing to sacrifice to gain attribute B? Software: R (mlogit, logitr) or NLOGIT.

Sources

  • Stated Choice Methods — Louviere et al.
  • Stated Choice Methods
  • Discrete Choice Methods with Simulation
  • Getting Started with Conjoint Analysis (Orme)
  • Factor and Discriminant Analysis (Sauro & Lewis, MeasuringU training)
  • Feature Prioritization Methods (slide deck)
  • Veal, A. J. & Darcy, S., Research Methods in Sport Studies and Sport Management (2014)

Also known as

  • Choice-based conjoint
  • Discrete choice experiment
  • Stated preference modelling
  • CBC
  • Trade-off analysis
  • Adaptive conjoint
  • Conjoint Analysis for Feature Valuation
  • Ratings-based conjoint
  • Metric conjoint

Categories

  • Analytics & Behaviour
  • Survey & Psychometrics