How to construct Conjoint Experiment in Qualtrics | XM Community
Skip to main content

Dear dedicated users of Qualtrics

 

Before everything else, thank you very much for your attention to my inquiry.

I am a doctoral student who is working on my dissertation, and I found out that I need to conduct ‘conjoint experiments’ to explore my research questions. 

I heard from the Qualtrics Sales Team that the version of Qualtrics that my university has a license for does not support conjoint analysis. Only “Qualtrics DesignXM” supports conjoint analysis.

 

At the same time, I see that some researchers constructed conjoint experiments in Qualtrics, and they say that they had to ‘hard code’ using HTML and/or Javascript to construct conjoint experiments in Qualtrics.

I am not sure whether 1) these researchers used “Qualtrics DesignXM” but still had to hard code to install in their experiments some functions that cannot be easily chosen by clicking some buttons in the DesignXM platform; or 2) they used a regular, plain version of Qualtrics that does not support conjoint experiments but still managed to construct the experiments by hard coding.

 

My question is:

Does “Qualtrics DesignXM” allow to construct conjoint experiments pretty much without the need to use hard coding? Or will there still be occasions where I need to hard code the functions that I want?

 

I apologize if my questions are a bit too vague. It is my first encounter of conjoint experiments, and I am figuring out how to construct them a bit by bit like a detective! ^.~

 

Thank you.

 

Hi,

I don’t  know if the Sales team provided you with this information, but you can find information about Qualtrics’ out of the box Conjoint and MaxDiff projects here : https://www.qualtrics.com/support/conjoint-project/getting-started-conjoints/getting-started-choice-based/getting-started-with-conjoint-projects/

@vgayraud vgayraud, thank you very much for your reply. Yes, I read the information in the link. 

 

What I want to know is whether most of the needs for hard coding are eliminated if I use Design XM. If I still have to hard code some of the functions for conjoint analysis in the Design XM platform, then it would not make sense to purchase a license for it. 

 

I know that some people know how to construct conjoint experiments in the regular, plain Qualtrics by hard coding. In that case, I may as well just hire someone who know how to hard code conjoint experiment in the regular Qualtrics platform rather than purchasing a license for Design XM. 

@oopssori3,

You don’t need to do any ‘hard coding’ if you use the Qualtrics Conjoint module. You enter the attributes and parameters and it does the rest. It is very locked down so you can’t change anything even if you wanted to. Think of it as ‘Conjoint for Dummies’.

It is possible to do conjoints without using the Qualtrics conjoint module and without ‘hard coding’ using a database and web services.  We’ve been doing it for a long time (before Qualtrics had a conjoint module). We continue to do it this way because it provides the flexibility we require that the Qualtrics conjoint doesn’t provide.

@oopssori3,

You don’t need to do any ‘hard coding’ if you use the Qualtrics Conjoint module. You enter the attributes and parameters and it does the rest. It is very locked down so you can’t change anything even if you wanted to. Think of it as ‘Conjoint for Dummies’.

It is possible to do conjoints without using the Qualtrics conjoint module and without ‘hard coding’ using a database and web services.  We’ve been doing it for a long time (before Qualtrics had a conjoint module). We continue to do it this way because it provides the flexibility we require that the Qualtrics conjoint doesn’t provide.

@TomG - Hi, can you please throw some light on “using a database and web services”? I want to know how can that be done.
Thanks.

 

@TomG - Hi, can you please throw some light on “using a database and web services”? I want to know how can that be done.
Thanks.

The database contains the conjoint ‘design’. A web service assigns a respondent to a design ‘block’ and returns all the associated data as embedded data fields.  Those fields are then piped into a loop & merge block, and presented in a multiple choice question. Using the design, you can lookup what was selected based on the block, loop, and choice.

@TomG

Thank you very much for your reply. It is super helpful. 

But then, I still have one part about constructing conjoint experiments in the regular Qualtrics version that makes me confused, Tom.

When we have many attributes and many levels per attribute, you cannot show all the possible product bundles to respondents. In that case, we need to “randomly select one level from each of the all attributes” and show these selected product bundles only to the respondents. 

You said that I do not need to hard code when I construct conjoint experiments in the regular Qualtrics platform. But, my question is, how can you install this “random selection" algorithm in your experiment, when you have exponential numbers of possible product bundles, if you do not hard code? 

 

Can you please answer this question, Tom?

 

Thank you very much for your attention!

 

@TomG

Thank you very much for your reply. It is super helpful. 

But then, I still have one part about constructing conjoint experiments in the regular Qualtrics version that makes me confused, Tom.

When we have many attributes and many levels per attribute, you cannot show all the possible product bundles to respondents. In that case, we need to “randomly select one level from each of the all attributes” and show these selected product bundles only to the respondents. 

You said that I do not need to hard code when I construct conjoint experiments in the regular Qualtrics platform. But, my question is, how can you install this “random selection" algorithm in your experiment, when you have exponential numbers of possible product bundles, if you do not hard code? 

 

Can you please answer this question, Tom?

 

Thank you very much for your attention!

 

You first create a statistical design that defines what bundles of products respondents will see. Each combination of bundles is assigned a block number. Then the web service randomly and evenly assigns respondents to blocks. 

Leave a Reply


Terms & Conditions