How to perform Conjoint Analysis

<<< Click here to go back 

First of all the categorical variables in data prepared for 18 mobiles in previous shown format is transformed to dummy variables :

Normal data

Initial Format of Data >>>

Data with dummies

Dummy Format of Data >>>

Using the golden rule for creating dummy variables that if there are n variation in a categorical variables, we should make only n-1 dummy variables for it.

Hence the redundant variables highlighted in yellow in "Data with dummies" are removed and we are left with the data shown right >>>

With the use of Data Analysis Tools pack within Excel, we run a Regression analysis on the above data with Preference as dependent (Y) variables and Hamsung, Jokia, Red, $50 and $100 as independent variables (X variables).

Result of Regression in Excel

The beta coefficients of the regression equation are called "Part Worth Utility" (PWU) of the variants of variables. Large part-worth utilities are assigned to the most preferred levels, and small part-worth utilities are assigned to the least preferred levels.

The PWU of $50 is more than that of $100, means that the buyer prefers $50 mobile more that $100 mobile. But I would suggest you to be little patient with interpretation, wait for utility calculations.


If A, B, C are the variants of a variable, and if the dummies A and B are used in model, the coefficient of A and B are both relative to C,  C's part worth utility is consider as "0" and hence the coefficients are part worth utility of A and B.

Now we calculate the utility of each variant using part worth utilities.

As stated above : the coefficient of A and B are both relative to C, we can consider solving following sets of equations:

Set 1:

EQ 1 :    Hamsung - Pineapple = -3.17

EQ 2 :    Jokia - Pineapple = -1.50

EQ 3 :    Hamsung + Pineapple + Jokia  = 0

Set 2:
EQ 1 :    Red - Blue = -1.11
EQ 2 :    Red + Blue = 0

Set 3:
EQ 1 :    $50  - $150 = 4.50
EQ 2 :    $100 -$150 = 2.33
EQ 3 :   $50 + $100 + $150 = 0


Solving above sets of equation gives the utility of individual variant of each variable.

Now you can start interpreting the choice and preferences of buyer.

1. Brand preference of Pineapple > Jokia > Hamsung
2. Color preference Blue > Red
3. Price preference $50 > $100>$150 ...

But one question is still unanswered ? What  ????

Which is more important for buyer : Brand, Color or Price ? How much weightage he gives to these attributes ?

So Now let's calculate Relative Importance Weights :

First we calculate the range of utility using following formula :

Range =  Maximum (Utility of variants) -   Minimum (Utility of variants)

Now relative importance or weighatge :

Importance of Brand  = range of Brand /  (Range of Brand + Range of Color + Range of Price )
Importance of Color  = range of Color /  (Range of Brand + Range of Color + Range of Price )
Importance of Price  = range of Price /  (Range of Brand + Range of Color + Range of Price )

With this I conclude the Linear Conjoint Analysis theoretical part. In subsequent article, I would explain the short and simple method to perform a conjoint analysis in SAS.


Enjoy reading our other articles and stay tuned with ...


Kindly do provide your feedback in the 'Comments' Section and share as much as possible.