Conjoint analysis is a popular technique in the area of marketing. It gives valuable information about the voice of the consumer. We are often faced with situations in which we have to take decisions (purchase of goods, selection of a tourism package etc.) and we are required to evaluate multiple criteria while we arrive at the final decision. In almost all MCDM (Multiple Criteria Decision Making) situations, conjoint analysis comes in handy. Let us take an example – We may be aware of the 4Ps of Marketing (now called the 7 Ps) – Product, Price, Place and Promotion. How does a marketer evaluate what P to focus more to achieve sale? Conjoint analysis would then come to his rescue and help him decide what P to focus more on as per the consumer’s preference.
Pre-requisites for Conducting Conjoint Analysis
a. Business Problem or a Problem Statement
b. Multiple criteria
c. At least two levels or sub-criteria per criterion
d. Any statistical analysis tool like R, SPSS and even MS Excel.
The approach to any conjoint analysis problem is two-fold. First of all we must do some qualitative study and then secondly comes the quantitative part.
This stage of the analysis deals with extensive literature review. This is quite relevant as it helps to find the possible criteria. After this, we have to see what criteria are the most relevant. Simple online surveys, telephonic discussions and interviews can help to find the most relevant criteria. The analyst must then select the top five or six criteria (along with their respective levels) as a result of the qualitative study.
Quantitative study comprises of formulating the orthogonal matrix. Suppose, you had 6 criteria each having two sub criteria and you were to generate an orthogonal matrix. As per the fundamental principle of counting you’d have 26 = 64 different combinations. This is called a full factorial design which is rarely practised. The conventional approach deals with creation of an orthogonal matrix based on 16 to 18 combinations, which is called a fractional factorial design. The matrix thus generated is then administered to the consumers/respondents wherein they are asked to rank these 16 to 18 combinations based on their preference.
Application: HR analytics
Portfolio design for total rewards:
Problem Statement: What is the preferred compensation component out of a given set of components?
Elements and their levels: Dr. Jac Fitz-Enz in his book titled “The New HR analytics” mentions six vital elements of any HR compensation package
· Work-life Balance
· Performance and Recognition
· Professional Development
I adopted a 6 criteria, 2 level model for this analysis and used a fractional factorial design of 16 combinations.
Aspiring MBA students ranked their preferences in the survey form provided to them. R analysed these responses and provided some interesting results.
Surprisingly, the students preferred position over salary and work-life balance. The most noteworthy observation is that work-life balance has the lowest importance out of the six elements. Performance and Recognition and Professional Development have almost equal importance. In addition, we can ascertain the attribute-wise level importance also.
Looking at each element, the analysis could find the relative importance or each (highlighted)
Implication: As a result, this analysis provides insights into how HR departments can structure their compensation policies. Consequently, this increases job satisfaction and reduces attrition.