Is the random number generator sampling from a uniform distribution? | XM Community
Skip to main content
I'm setting up a survey where I need to use Qualtrics's random number generator to assign respondents to different scenarios. I've tried to find information about how the random number generator works, but have found little documentation. I'm assuming that all numbers are drawn from a discrete uniform distribution, but am I wondering if anybody has more information on this or has actually tested this.



As a rough test, I sampled approximately 220 random numbers from the generator and got the following results (apologies for the huge images, I could not figure out how to make them smaller):

!



It looks promising, but I would gladly take part of any more information I can find about this.
If you are concerned about a uniform distribution a better way to assign respondents to scenarios is to use a survey flow randomizer with the evenly present box checked.
> @TomG said:

> If you are concerned about a uniform distribution a better way to assign respondents to scenarios is to use a survey flow randomizer with the evenly present box checked.



That is a very good suggestion. Unfortunately, I need to use the random number generator in order to assign approximately 1/10 of the respondents to a control block, and approximately 9/10 to different treatment conditions within a treatment block. The weighted randomization requires me to use the random number generator.
Hi @lchrstnsn! You can find additional information about our random number generators here and here.
You could still use the randomizer. For example, say you have an embedded data variable named control... Put 10 embedded data blocks under the randomizer with select 1 and evenly present. In the first block set control to 1 and in the other 9 set it to 0. Then continue the flow using control to branch accordingly.

Leave a Reply


Terms & Conditions