How to randomize multiple segments equally? | XM Community
Skip to main content

Hello!  

I am writing a survey where I’ll have 4 segments by device type and will create a variable to categorize them.  There will be 25 per segment.   

I also need to show two different question sets to half of all respondents.  I’d like each device type segment to be equally divided (can be off one since it’s an odd number) so that I don’t end up with the entire sample being equally divided while the different device types are not.  

I know how to create the variables for the device types and I know how to randomize blocks, but not sure how I can assure each segment is equally divided 50/50 between the question sets.  

Appreciate any thoughts! 

Thanks! 

Hi @josiwax2020 ,

To ensure each device type segment is equally divided between the two question sets, you can use embedded data variables and randomization. Here's a step-by-step approach:

1. Create Embedded Data Variables: Create embedded data variables to track the assignment of each respondent to a device type segment and a question set. For example, you could create variables like "Device_Type" and "Question_Set" with values indicating the device type and question set assignment.

2. Randomize Block Order: Randomize the order of blocks containing the two question sets. This ensures that each respondent has an equal chance of being assigned to either question set.

3. Use Embedded Data in Branch Logic: Set up branch logic based on the embedded data variables. Create branch conditions that assign respondents to question sets based on their device type. For example:
   - If Device_Type is "Desktop" or "Tablet", assign to Question_Set A.
   - If Device_Type is "Mobile" or "Other", assign to Question_Set B.

4. Ensure Equal Distribution: To ensure an equal distribution of respondents within each device type segment, you can set up additional branch logic within each block. For example, if each device type segment consists of 25 respondents, you can create branch conditions that allow the first 13 respondents in each segment to be assigned to Question_Set A and the remaining 12 to Question_Set B.

5. Test Survey Logic: Before deploying the survey, thoroughly test the survey logic to ensure that respondents are correctly assigned to device type segments and question sets according to the desired distribution.

 


Leave a Reply