How to conduct a 2x2 between subjects experiment design to create 4 conditions with even number of people? | XM Community
Skip to main content

I’m conducting a 2x2 between-subjects experiment for my dissertation where the independent variables are culture (American vs. Chinese) and the anthropomorphism of virtual influencers (human like and cartoon-like) and I want to test their perceptions of them. 

 

I’m wondering how to construct this survey so that based on what their culture is (American vs. Chinese) they will be allocated randomly to a virtual influencer who is either human-like or cartoon-like, therefore creating a total of 4 conditions. I also want the numbers to be relatively even across conditions. What would the survey flow look like? 

 

I also want to make sure that when I analyse the results, it’s structured according to the 4 conditions I set, is embedded data responsible for this?

 

 

In the survey flow you would create block display randomization, on the right hand side you can select to evenly present the elements.

You could add the embedded data of American or Chinese and then nestle the randomizers accordingly under them. 

One more way you could go about this is by randomly assigning numbers and then creating display based off their assigned number. I would then say for the 4 variations that if their number is 1-25 they would see 1 variation, if their number is 26-50 then they see variation number 2, etc...

 

Thanks @kgillis I appreciate your help. 

 

Thing is, I haven’t collected any data yet so when I actually run my experiment, I will be getting American and Chinese respondents for the first time. I therefore want them to be randomly allocated to the different virtual influencer conditions based on whether they answer as being American or Chinese in a previous question. From my understanding, embedded data helps assign participants to conditions but I’m not sure how… Here is what I have so far in my survey flow:

 

Do you think this could work?

That should work! And I like that you’ve embedded their condition number in there for easier analysis on the back end. I would recommend testing and testings (and having others test too) to ensure it’s working as anticipated. 

Leave a Reply


Terms & Conditions