Variants of Conjoint Analysis

The most common variant of conjoint is choice-based conjoint; the same technique is also known by other names. There are a variety of closely related techniques.

Jargon

What is referred to in this guide as choice-based conjoint is also known variously as:

• stated preference choice modeling
• choice modeling
• discrete choice experiments
• choice experiments
• conjoint analysis
• conjoint

Unfortunately, there is no great consistency in terms of how these terms are used, so sometimes when somebody says they have “done a conjoint” they may mean choice-based conjoint, but they may also mean something a bit different.

A statistical analysis that understands preferences using choices by consumers is known as a choice model. When a choice model is created using hypothetical choice questions like the one shown in What is Choice-Based Conjoint (CBC)? is known as a stated preference choice model. A choice model can be created by analyzing historic data, such as in-store purchases at supermarkets, or last week’s transport choices. Such choice models are known as revealed preference choice models.

Other variants of conjoint and choice modeling

The original meaning of conjoint analysis in market research referred to experiments where people were typically given around 16 descriptions of hypothetical products, on cards, and were asked to rank these according to preference. Today, when a person says that they used “conjoint”, they may mean they used this original technique (better known as ranking-based conjoint), or they may mean that they used choice-based conjoint.

There are many other versions of conjoint analysis, including:

• Ratings-based conjoint, which gives people a series of descriptions of hypothetical products and asks them to provide a rating of how likely they are to buy them.
• Best-worst conjoint, which gives people choice tasks and asks which they would buy and which they would not buy.
• Adaptive choice-based conjoint analysis, which is a variant of choice-based conjoint, where the questions that a person is asked change depending on their answers to earlier questions.
• Hybrid conjoint analysis, which involves combinations of two or more of choice questions, ratings, and rankings.
• Menu-based conjoint, where people are given choices for different aspects of a product (e.g., choosing the burger, side, and drink when choosing at a fast-food restaurant).