The first academics who applied the notion of conjoint measurements were Green and Rao inspired by the mathematical psychologists: Lucy and Turkey (1964). Since the first version of conjoint (which was developed in the 1970s), we observe the development of conjoint techniques which change in both design and methods of analysis. To create a conjoint design one can currently use different software packages such as SPSS, SAS, and even Excel but the world leader in conjoint software is Sawtooth Software. Their software contains all existing conjoint analysis methodologies and the two newest conjoint methods; ACBC and MBC were introduced by this company.
Conjoint Value Analysis (CVA)
Traditional Full-Profile Conjoint (called CVA by Sawtooth CBC Software; this version is also offered by SPSS and SAS) is useful for measuring up to six attributes. CVA can design pairwise conjoint questionnaires or single-concept card sort design so it can display either one or two products at a time. It may be used for paper-based or computer-based interviewing.
In the context of conjoint analysis in medical devices markets, we do not recommend the questionnaire design tool to be used for studies with many attributes. With many attributes, the questions are likely to provide too much information to the respondent, who may be overloaded and confused.
Pairwise Comparison example:
Comparative information is all that is necessary for estimating relative part worth and running competitive market simulations.
CVA features relevant for conjoint analysis in medical devices markets:
- CVA does not use any strategy for simplifying the respondent’s task. Every question involves consideration of all the conjoint attributes. With pairwise questions, the amount of information the respondent must consider is twice as great as with single concept methods.
- CVA requires showing respondents a considerable number of questions (tasks) e.g. a study with 5 attributes with 3 levels each would require about 33 tasks.
- CVA is limited by the number of attributes, six is the cut-off point.
- CVA utilities are estimated from rating scales or rank-order data. These scales do not automatically lead to market simulations with appropriate choice probability scaling. One can achieve approximate choice-probability scaling for utilities by adding holdout choice tasks within surveys.
Adaptive Conjoint Analysis Methodology(ACA)
Adaptive Conjoint Analysis (ACA) is a software for conjoint (trade-off) analysis which can also be used in conjoint analysis in medical devices markets. The term “adaptive” refers to the fact that the computer-administered interview is customized for each respondent. Data are analyzed as the interview progresses, and we choose questions likely to reveal the most about the respondent’s values in the shortest time.
The term “adaptive” refers to the fact that the computer-administered interview is customized for each respondent; at each step, previous answers are used to decide which question to ask next, to obtain the most information about the respondent’s preferences. The respondent’s utilities are continually re-estimated as the interview progresses, and each question is chosen to provide the most additional information, given what is already known about the respondent’s values. Respondent utilities are available upon completion of the interview.
The ACA interview has several sections:
- Preference for Levels (respondent rates the levels in terms of relative preference).
- Attribute Importance (where we determine the relative importance of each attribute to this respondent)
- Paired-Comparison Trade-Off Questions (is shown two product concepts; the respondent is asked which is preferred, and also to indicate the strength of preference)
- Calibrating Concepts (Optional Section) (The respondent is asked a “likelihood of buying” composed of products from very unattractive to very attractive for the respondent)
Initial utility estimates are based on the respondent’s desirability ratings for attribute levels together with ratings of attribute importance. Utility estimates are updated during the interview to incorporate the answer to each paired-comparison question. As the respondent progresses through the paired comparison section, the initial estimates become less influential.
- In the context of conjoint analysis in medical devices markets, it’s suitable for studies with more than about six attributes. ACA interviews can study up to 30 attributes, and each attribute can include up to 15 levels. It also works well for studies with less than six attributes but no major advantages to the CVA method.
- More suitable for studies where pricing research is not the goal (ACA often underestimates the importance of price)
- Most of ACA’s questions present only small subsets of attributes, so questions do not necessarily become more complex when there are many attributes in the study
Choice-Based Conjoint (CBC)
CBC (Choice-Based Conjoint) – the respondent sees a series of choice tasks. Each task displays several concepts and asks the respondent which he would choose from that set. Optionally, a “would choose none” option may be offered. Attribute levels in each concept are varied in such a way that values similar to conjoint utilities can be estimated for each attribute level. Analysis can be done at the group level and individual level (if HB estimation of utilities used).
- it mimics the purchase process in competitive contexts, instead of rating or ranking respondents indicate which product they would purchase
- respondents can decline to purchase by choosing option “None”
- not appropriate for studies involving large numbers of attributes – suggested is to use no more than about six attributes (although advanced CBC module enables larger a number of attributes)
- suitable for pricing studies and “off-the-shelf” products
- apart from single choice, CBC also enables chip allocation where respondents can distribute the number of products they would purchase from a set of product concepts (which also mimics the real purchase process)
- it requires larger a sample than other conjoint methods – recommended sample is 200-300 respondents if some level of respondent segmentation is envisaged. In our experience, the optimal minimum for a conjoint analysis in medical devices markets is >30 respondents
CBC: Task example
Adaptive Choice-Based Conjoint (ACBC)
ACBC is a combination of two methods: ACA and CBC. It captures more information at the individual level than traditional CBC surveys and may be used even with small samples. ACBC interviews include the three core sections plus one optional:
- BYO (Configurator)
- Screening Section
- Choice Tasks
- Calibration Section (Optional)
As part of the conjoint analysis in medical devices markets, respondents first configure their preferred product via a BYO (Build Your Own) question. Based on that preferred product, we create a set of similar products for the respondent to evaluate in the Screening section. Respondents indicate which of these similar products they would consider, and reveal non-compensatory cut-off rules. Finally, respondents make a final product selection among products screened into their consideration set. This is done via a Choice Tasks section, using a tournament format. Last – the calibration section – is used only for estimation of a part worth threshold for “None.”
- ACBC questionnaires generally take longer than CBC, especially in the context of conjoint analysis in medical devices markets
- Some ACBC respondents report higher satisfaction and interest in the survey than CBC respondents
- Suitable for studies with more than six attributes
- Suitable for complex products/services
- Requires lower sample than CBC
Menu Based Conjoint (MBC)
MBC is software for analyzing a variety of menu-based and discrete choice problems. The idea behind this method is that respondents can select one to multiple options like a real menu where multiple options are available but only some of them we are interested in.
- respondents make from zero to multiple selections of options on the way to building their preferred choice
- the price of the total solution is shown and is updated as respondents “buy” more items on the menu (it isn’t necessary to show the total price, but it is common to do so)
- selections can be restricted (cannot pick item Y if you pick item X, etc.) or unconstrained
- there are often constraints on pricing (e.g. medium size drinks must be a higher price than small size; the price of the bundle must be less than purchasing the same items a la carte)
- The number of tasks varies from 6 to 12 and prices (and/or availability of items) change each time
- High requirements for sample size (even larger than for CBC) – about 800 for a typical consumer. Our experience for conjoint analysis in medical devices markets shows that the actual minimum for healthcare is 100-150 respondents
MBC Task example (prices vary in each task)
The conjoint analysis methodology in medical device markets differs slightly from the traditional approach (mainly in the sample size researchers are able to obtain). If you wish to read more about a general overview of conjoint analysis in medical devices markets, please go to Introduction to Conjoint Analysis and if you want to learn more about our approach to Unmet Need Analysis, go here.
IDR Medical offers the full range of conjoint techniques choosing and adapting each technique to solve our clients’ issues and questions in the most optimal way. Please contact us if you wish to learn more about the solutions we can provide you with.
http://sawtoothsoftware.com, Technical Papers
“Getting Started with Conjoint Analysis” Bryan K. Orme