Randomly display a question in two different locations | XM Community
Skip to main content

Hi all.

I am building a survey and want to test a specific question in different locations to see if the question position in the survey influences the results.

So I want 50% of respondents to see the question at the start of the survey and the other 50% to see the question at the end of the survey.

Is this possible? Has anyone done something like this?

I can just send the survey out to 50% of the sample with the question at the top. Reposition the question at the bottom and then send it out to the other 50% of the sample. But this won’t guarantee an even number of responses for both conditions and I was hoping for something more automated.

Thanks.

Hi Steve,

 

You should be able to do this.

First copy the question and place it where you need it in the survey.  If you can set the embedded data in the email lists, using that you can apply the display logic to both the questions and display the questions as you need .

 

 

Hello ​@SteveJ78,

Create two blocks, place the question at the top in the first block and at the bottom in the second block. Then, go to the survey flow and add a randomizer. Put both blocks under the randomizer and set it to show 1 of the 2 blocks to each respondent. If you’d like to ensure an even split, enable the “Evenly Present Elements” option. Please see the screenshot below for your reference:

Let me know if this helps.

Hi,

I’d use a solution similar to ​@Sachin Nandikol ‘s, but without using separate questions sets. I find it’s often easier to analyse the data afterwards if they are in the same field from the beginning.

Use a randomizer and the same questions blocks but change the display order. Something like this:

 

You could choose to set embedded data that gets randomised through a randomizer, with evenly present elements selected at the top of the survey flow. Then, show the block in the beginning if start gets selected and show the block in the end if end gets selected.

However, this method will cause the back button to disappear for certain portion of the survey due to the branch logic.

Hence, I would advise to use vgayraud’s method.

​@vgayraud ...your solution worked perfectly! Thank you. The only thing I added was an embedded data field in each Group to more easily identify which version of the survey the respondent received (i.e., Group A or Group 😎. This will help with analysing results to determine if respondents answer differently based on the order of questions they received. 

 

Again, thank you!

  

Hi,

I’d use a solution similar to ​@Sachin Nandikol ‘s, but without using separate questions sets. I find it’s often easier to analyse the data afterwards if they are in the same field from the beginning.

Use a randomizer and the same questions blocks but change the display order. Something like this:

 

 

Leave a Reply


Terms & Conditions