What is Conjoint Analysis?
Conjoint analysis is a statistical technique, originated in mathematical psychology, that is used to determine how people value different features that make up an individual product or service.
This popular research technique was initially developed by psychologists in the early 70s, interested in understanding how people make decisions. By directly asking how and why – assuming a conscious decision making process – people might respond in line with what is top of mind or what they believe the interviewer wants to hear (politically / socially correct answers). The answers don’t necessarily reflect what one would actually do, choose or buy. Choices involve trade-offs and compromises.
The key characteristic of conjoint analysis is that a product is composed of multiple conjoined elements (attributes or features). Based on how the combined elements (product concepts) are evaluated, the underlying preference structure can be determined.
Over time, various forms of conjoint analysis have been developed: from Conjoint Value Analysis and Adaptive Conjoint Analysis to Choice Based Conjoint and Adaptive Choice Based Conjoint to Menu Based Conjoint and Preference Based Conjoint.
Conjoint Analysis examples
Below is an example of how a Conjoint Analysis exercise looks like. You can also try it online here.
What are the advantages and disadvantages of Conjoint Analysis?
There are a few advantages that you can benefit from when doing a conjoint analysis:
- Replicates consumer choice and trade-off behavior:
- Selection of a product among multiple alternatives,
- Test objective not obvious for respondents,
- Can include compensatory and non-compensatory models.
- Flexible experimental research designs,
- Optimal product configuration and item prioritization from a business or consumer standpoint, and
- Predicts preference shares
However, there are also some disadvantages that you need to take into account before conducting the analysis:
- It doesn’t predict Market Shares exactly. Some common overestimations:
- New product versus just a few concepts (or only a none) since the market has so much more to offer,
- Expensive options (e.g. for a car) since people do not have to pay.
- Calibration can be used to correct and / or make the results better understandable for clients.
Why SKIM does it better than competition
- An expert in understanding and discrete choice modeling since 1979,
- Provides consultancy to other research and consultancy firms on how to optimally design the technical part of their study with either conjoint or other advanced analytics,
- A preferred development, training, and support partner of Sawtooth Software, the global leader in conjoint analysis software solutions,
- With almost 40 years experience we know exactly when to use which method, and how to best apply it. No matter how complex your business question is, our methodologists gladly use their creativity to come up with novel – yet solid – research approaches to solve it.
How to do a Conjoint Analysis
Conjoint analysis is usually done via a respondents survey. One needs to define the attributes and levels to test having the end goal in mind: for instance does one want to optimize product management or product development or does one want to test an online product or service by replicating the purchase decision? Or is one mainly interested in the brand price trade-off? The right set up is necessary to make sure that the final market simulators can be used to test different scenarios and deliver the answers to the business questions.
Many factors play a role when determining how to set up ones conjoint analysis survey. See below for an instruction video
What software can you use to conduct a Conjoint Analysis?
Within SKIM we use various software packages. First of all, for scripting the surveys and setting up the conjoints we use Lighthouse, the latest update of Sawtooth Software. The extended packages have possibilities to run Hierarchical Bayesian analysis and segmentation via Latent Class or Combined Cluster Ensemble Analysis.
The first package has quite some interesting settings for the more experienced researcher, e.g. to estimate utilities part worth or linear. The output of the regular questionnaire data is in SPSS format.