Description of Methods to Capture Perceived Attribute Values

Full Profile ConjointChoice ModelingCompositional Conjoint Profiling (Simalto)Rating Model
Task Rank ordering of a set of product descriptions -usually personal interview / Mall Intercept or phone-mail -phone are used.Select choice of purchase of a number of cases (8-10 cases) each - often mail or phone-mail is used.Rank a set of attribute levels including price in order of importance - often by mail , phone-mail or interview)A series of exercises of selecting, ranking and identifying attribute levels - usually personal interview (mall Intercept)A series of attribute ratings of competitive products and concepts and an importance scale (preferably a constant-sum).
LimitsIn practice should not handle more than seven attributes or more than 27 cards which limits levels. Only single situation can be tested.Usually only a couple of attributes can be included due to the interaction with brand. Only single situation can be tested.While as many as 83 attribute levels have been used, we suggest less than 20. Multiple situations can be tested.Large number of attributes and attribute levels (> 100). Multiple situations can be tested.Only a few attributes can be considered with a few competitors, typically < 16 attributes and <6 competitors.
Simulated DecisionBroad product selection simulationPurchase process choice simulationNegotiation simulationNegotiation or a design decisionSatisfaction model
Underlying AssumptionsConsistent trade-off respondent behavior based on primary effects. It is assumed that respondent can better select complete offerings than attribute levels.Consistent trade-off decision behavior by respondents with interaction of attributes (particularly price) with brand.Respondent is able to trade-off attribute levels. There is some explicit value of attributes.Ability of respondent to describe past and future purchases in regards to attribute levels and values.Ratings and importance of attributes are comparable. Ratings and importance are ratio scale values.
ResultsAttribute level value & market simulationMarket ModelAttribute level value & market simulationMarket simulation and attribute valuesProduct, attribute position, market simulation.

Full Profile ConjointChoice ModelingCompositional Conjoint Profiling (Simalto)Rating Model
AggregationEither on a respondent basis or aggregated.Due to the large number of cases needed, only aggregated data is used.Done on a respondent basisDone on a respondent basisEither on a respondent bases or aggregated.
Analysis Regression on the respondent and averages. Monotonic Regression is sometimes used. Rankings are usually scaled to dollars.Regression on the aggregated data. Logit transformed share and maximum likelihood loss regression.Rankings of attribute levels scaled to dollars and segment clusters usually also done.Represented as tabular data and through market simulation.Position maps (Factor Maps and MDS) for distance, segment clustering, value modeling, ratings and importance presented as tabulations.
Market ModelMarket simulations based on value, share based on either winner takes all or proportional.Share (stochastic) market model is inherent in this method. Market simulations based on value.Multiple simulation models, based on: value, ideal case, last purchase and satisfaction.Model based on Fisher model: (Ratings x importance)
Fault ToleranceFault intolerantFault intolerantSomewhat fault tolerantFault tolerantFault tolerant
ProblemsProne to decision model problems as well as design problems. Major problems are interaction and unrealistic cases. Task size is another problem.Aggregation, design and task size are the major problem. There is no ability to examine market details.Inability of the respondent to rate prices against other attributes. Alternatively, there is a problem of selecting appropriate price levels.Complexity and cost are the major fielding difficulties in this method. Selection of an appropriate market model is also a major problem.Usually not accurate. No evaluation of attribute level - No direct recommendations for action.

Full Profile ConjointChoice ModelingCompositional Conjoint Profiling (Simalto)Rating Model
AdvantageSimulates a selection process without an explicit buyer model. It has industry credibility.Simulates the buying process. It is also being presently "hyped"Low costs, ability to provide multiple situations and easy integration with other surveys.Flexibility and in-depth analysis.Simple and widely undertaken as customer satisfaction.
Relative CostsFairly high costs if personal interview is used, otherwise medium cost range.Usually expensive due to large sample size required.Usually very inexpensive and can be coupled to existing survey.Usually very expensive due to the need for personal interviews.Usually inexpensive and coupled to existing survey.
Successful CasesOrganizational and group decision making, packaged goods, big ticket consumer products, pharmaceuticals, etc.Competitive pricing studies.Agricultural chemicals, industrial and medical products, services.Industrial and medical products, consumer soft-goods, and big ticket consumer purchases.Consumer, industrial and service products.
RulesMin. # of cards = 1.5 x ( attribute levels -1)+1Min. # of exercises cards = 1.5 x ( attribute levels -1)+1