Low cost Conjoint analysis services in India - write us at info@surveytabs.com

ARCADIA, Fla. - May 9, 2014 - PRLog -- @visitors: In this article we would discuss in detail complete conjoint design and analysis . All the R codes are also provided.

Now I want to make a questionaire for conjoint analysis commonly known as Choice Modelling. Choice modelling is useful for various purposes like Business,Marketing Science, Economics, Pharmaceutical research, Pricing Analytics etc.  Choice Modelling is useful to identify the user preferences or demand curve(Economics) over various alternatives.

Let us assume that there are 4 different product characterstics . Let these characterstics be named as

characterstic_1 (Frequency)
characterstic_2 (Color)
characterstic_3 (Size)
characterstic_4 (Age group)

These characterstics can be a quantitative variable like price which can also be taken into account in a similar way.

Let us also assume that there are 2 levels of choices in characterstic_1 - High and Low , 2 levels in characterstic 2 - Pink and Red,2 levels of size - Large and Small and 2 levels of age group - <20 and >=20.

Now there can be 2x2x2x2=16 product possibilities. Now if the researcher is interested in finding out that which product is the most preferred product within the targeted user group. The most common thing and the simplest thing that can be done is to make the user to rate or rank the preferred products.

But this is often biased and can be erroneous as it is difficult for the user to compare the 16 products at a go while ranking.Similarly while rating the user may give similar ratings to the different products. This is more of problem if there are more than 16 products.

Hence choice modelling is a method in which the user compares 2 to 3 alternatives. There are often less number of product combinations/questions that are compared than requiredto exact estimate the choice structure.The choice structure is deduced with the limited responses obtained by comparison of questions/user preferences reflected in some of the questions.Choice Modelling has many different variants. One of the ways and most common way of enquiring a question something like this.We would only discuss this variant
in detail.

Which of the following do you prefer most among the following alternatives

a) Product 1 ( combination of characterstic_1 to characterstic_4)
b) Prdouct 2 ( combination of characterstic_1 to characterstic_4)
c) None of the above

Inclusion of the None of the above is very Important if you think the None of the above is always an option and can be a better alternative than option a) or b)

None of the above option can be omitted if the you feel that it always a least prefered option than any of the alternatives.Although an indicator for it would be taken into account in our calculations of utility functions. All the utility functions would be calculated on the basis of assuming the "None of the above option" as zero utility.

In all if we need to compare each alternative with each other . It would require as 16 combinations .This is called a full factorial design.

If all these combinations are answered, then we do not need a conjoint analysis. The preference structure can itself be analysed by looking the order in preferences.But practically , making the user to answer all the sets of choices is virtually impossible due to the tediousness of the questionaire.

Hence the beauty of the conjoint analysis is that the user need not answer all the questions but instead needs to answer a few questions which help empirically derieve the preference structure. Let us suppose we decide to ask the 10 questions, say 12 or 6 in the above example instead of 16. We can also have more or less than 8 questions. More the number of questions -lesser would be the variance and error from actual preference structure but more would be the tediousness of the questionaire.

Hence Let us keep the number of questions to be 10. While creating a design of the preference structure, we want to extract maximum information about the choice structure from these 10 quesions.

If all these combinations are answered, then we do not need a conjoint analysis. The preference structure can itself be analysed by looking the order in preferences.But practically , making the user to answer all the sets of choices is virtually impossible due to the tediousness of the questionaire.

Hence the beauty of the conjoint analysis is that the user need not answer all the questions but instead needs to answer a few questions which help empirically derieve the preference structure. Let us suppose we decide to ask the 8 questions, say 10 or 6 in the above example instead of 16. We can also have more or less than 8 questions. More the number of questions -lesser would be the variance and error from actual preference structure but more would be the tediousness of the questionaire.

Hence Let us keep the number of questions to be 10.This is called a fractional design. While creating a design of the preference structure, we want to extract maximum information about the
choice structure from these 8 quesions. Hence the most efficient way of designing is called the orthogonal deign.

For designing 10 orthogonal questions of format QF. we have to use a particular library in R. One such library is Algdesign.


Outsource Conjoint analysis to SurveyTabs

For more details and help write us at info@surveytabs.com

One of the leading market research company in India (www.surveytabs.com)