How can I use Qualtrics to randomise participants into 4 conditions while keeping gender ratio even? | XM Community
Solved

How can I use Qualtrics to randomise participants into 4 conditions while keeping gender ratio even?

  • 12 February 2020
  • 5 replies
  • 526 views
A
Is there any way to use Qualtrics to assign research participants into four between-subjects conditions WHILE ALSO keeping the gender ratio for each condition roughly balanced?

Assuming I collect gender information as the first question on a survey, is there some way to use quotas or randomiser features or display logic to make this happen? I have tried to think it through but can't think of any way to do this.
icon

Best answer by Akdashboard 12 February 2020, 15:18

View original

5 replies

PeeyushBansal
Rank
Userlevel 7
Badge +33
if it is only single variable that is age.. you can put randomizer in condition to select 1.. and copy these conditions.

Like if(age=18-24)... Randomizer(select 1 and equaly present)--4 elements
Best Regards, Peeyush
Akdashboard
Rank
Userlevel 7
Badge +6
This seems pretty straightforward. You need 2 branches in your survey flow. 1 IF male and 1 IF female (however you define that in your data).

Then nest a randomizer block under both branches and use the randomizer to assign 1 of the 4 conditions evenly (make sure to check "equal present" 1). Both branches should look identical, with the only difference being the Male/Female branching logic.

This will make sure to assign each condition evenly among the genders, so that, assuming you have an even gender split, your responses will be evenly represented.

Note, if you have something like 40 males and 60 females, then you will still get an even distribution of the conditions within the genders, but your results will skew slightly higher for females.

If you want to control this further, you can use Quotas to cap the number of responses from each condition and gender.
PeeyushBansal
Rank
Userlevel 7
Badge +33
hi @Akdashboard , How to control balancing if say i have to consider 7 screener variables like age, gender, income...
Best Regards, Peeyush
A
Hi @Akdashboard, thanks for your suggestion! I did consider that methodology, but unfortunately as you note in the latter part of your message, this will not be as random as is ideal. Similar problem when it comes to using quotas to balance the number of males and females across conditions. That said, this is indeed the best possible way to approach randomness, so thanks for sharing your thoughts!

> @Akdashboard said:
> This seems pretty straightforward. You need 2 branches in your survey flow. 1 IF male and 1 IF female (however you define that in your data).
>
> Then nest a randomizer block under both branches and use the randomizer to assign 1 of the 4 conditions evenly (make sure to check "equal present" 1). Both branches should look identical, with the only difference being the Male/Female branching logic.
>
> This will make sure to assign each condition evenly among the genders, so that, assuming you have an even gender split, your responses will be evenly represented.
>
> Note, if you have something like 40 males and 60 females, then you will still get an even distribution of the conditions within the genders, but your results will skew slightly higher for females.
>
> If you want to control this further, you can use Quotas to cap the number of responses from each condition and gender.
Akdashboard
Rank
Userlevel 7
Badge +6
> @PeeyushBansal said:
> hi @Akdashboard , How to control balancing if say i have to consider 7 screener variables like age, gender, income...

You'd really just be doing the same thing, just multiple times, right? If X or Y or Z (etc.) then randomly assign correct blocks. Controlling the maximum number of responses from a single group with quotas.

Leave a Reply


Terms & Conditions