Random Variable Number Generation problem
Jan 08, 2024
Hello,
To evaluate our courses in Microsoft Forms, we have created an external Questionnaire link. And since our courses are taken by thousands, then we only want 10% of our userbase to get presented with said Questionnaire.
So what we have done, is created a number Variable that starts at 0.
On the introduction slide, in the beginning of the course, we then have a trigger to sets this Variable to a random number between 1 and 10.
At the end of the course, after passing the final quiz, we have a Continue button that send people to the Questionnaire if the beforementioned Variable is "1" and if not, send them to the final slide.
The problem is, that roughly 30% of all people get the questionnaire, and not 10%. When we have 10.000 people taking a course, then this seems statistically impossible.
So my question is, how is this random number generated through the trigger? How come so many people land on the number 1? And what can we do to increase the random-ness of the random number generator, so we truly get 10% people to take the questionnaire and not 30%?
I can't attach the course, it's classified. But I've checked on the Variable window that there are no additional triggers on other slides that uses this questionnaire variable. It's literally just randomized once on the third slide of the course and that's it.
Best regards,
3 Replies
Hello Asbjorn!
Sorry for the trouble working with random variables; I'm happy to help! I appreciate the detailed explanation of the issue.
You mentioned that you're using a trigger to generate the random variable. Can you try creating a random variable inside the Manage project variables window instead? Here are the steps:
Please let me know if you have any questions.
I'll try to make the variable random that way, that will save me a trigger! Great. :)
I hope it will change the outcome, it seems super weird that we get so many people getting presented with the Questionnaire, statistically unlikely.
I will get back in some months when we have a new batch of data, using this new method. :)
One last thing for now, do you know how the random number is generated? Does it generate a number from a seed dependent on some internal clock of sorts?
I'm not sure what the random seed process is, but I have noticed in the past that the first random number generated seems to repeat more often that random should allow. Eric's method may help that. Otherwise, here are a couple of options you might try.
Choose a number other than 1 to trigger the questionnaire. Or, you might generate a number between 1 and 10, and a second, say of either 4 or 5, and only go to the questionnaire if those two are equal. (I'm just guessing here, as that approach might have the same problem you are already facing.)