Millisecond Forums

Balancing Randomization of Blocks?

https://forums.millisecond.com/Topic26246.aspx

By am2922 - 1/10/2019

Hello!

I am currently running an experiment in which we are randomizing the order in which participants see different blocks. Here is the expt code that I've been using:

<expt >
/ preinstructions = (Intro)
/ postinstructions = (End)
/ blocks = [1 = StartTone; 2 = panas_baseline; 3-4 = noreplace(human, dog) ]

The randomization works great, but I've come across a problem where it seems like the human block is being shown first much more often than the dog block (human has been shown first 37 times vs. 22 times for dog first). Also, there are periods in which human is shown first 7x in a row, and then the dog block is shown first the next 3x. I'm just wondering if there is any way to balance the randomization, so that the amount of times each block is shown first is a little more even and less clustered? 

I appreciate any help/advice you might be able to provide with this issue!
By Dave - 1/10/2019

am2922 - Thursday, January 10, 2019
Hello!

I am currently running an experiment in which we are randomizing the order in which participants see different blocks. Here is the expt code that I've been using:

<expt >
/ preinstructions = (Intro)
/ postinstructions = (End)
/ blocks = [1 = StartTone; 2 = panas_baseline; 3-4 = noreplace(human, dog) ]

The randomization works great, but I've come across a problem where it seems like the human block is being shown first much more often than the dog block (human has been shown first 37 times vs. 22 times for dog first). Also, there are periods in which human is shown first 7x in a row, and then the dog block is shown first the next 3x. I'm just wondering if there is any way to balance the randomization, so that the amount of times each block is shown first is a little more even and less clustered? 

I appreciate any help/advice you might be able to provide with this issue!


If you truly want to balance the block order, then your best bet is to treat the block order as a between subjects condition and assign half your subjects to one order, and the other half to the other order:

<expt >
/ preinstructions = (Intro)
/ postinstructions = (End)
/ blocks = [1 = StartTone; 2 = panas_baseline; 3 = human; 4 = dog ]
/ subjects = (1 of 2)
/ groupassignment = groupnumber
...
</expt>

<expt >
/ preinstructions = (Intro)
/ postinstructions = (End)
/ blocks = [1 = StartTone; 2 = panas_baseline; 3 = dog; 4 = human ]
/ subjects = (2 of 2)
/ groupassignment = groupnumber
...
</expt>
By am2922 - 1/10/2019

Okay, so just to clarify, if I used this then I would control the order by either typing '1' or '2' in as the group number right before starting the inquisit task program, correct? Or does the program generate the group number and assign each participant to a group randomly?

Thank you for your help!
By Dave - 1/10/2019

am2922 - Thursday, January 10, 2019
Okay, so just to clarify, if I used this then I would control the order by either typing '1' or '2' in as the group number right before starting the inquisit task program, correct? Or does the program generate the group number and assign each participant to a group randomly?

Thank you for your help!

In Inquisit Lab, you'd type either 1 (for the first order: human -> dog) or 2 (for the second order: dog -> human) into the "Enter the group number:" field.



When using Inquisit Web, you would instead select an suitable group number generation method in the web experiment's settings.