+- FindingFive Discussion Forum (http://discuss.findingfive.com)
+-- Forum: Researchers (http://discuss.findingfive.com/forumdisplay.php?fid=3)
+--- Forum: Study Grammar & Management (http://discuss.findingfive.com/forumdisplay.php?fid=5)
+--- Thread: randomizing blocks and response choice display (/showthread.php?tid=258)
randomizing blocks and response choice display - kolenorberg - 08-21-2023
I have a study with three passages which fall in one of 4 conditions. I want participants to see each passage in one of the conditions but I want whether or not they receive all conditions across passages to vary randomly. Something like the below code which doesn't seem to work. How can I accomplish something like this? Specifying a condition for all possible combinations here is a little unwieldy and wouldn't replicate a prior experiment well.
"block_sequence": [{"$randomize":[
{
"A_C": ["A_full_control"],
"A_E": ["A_only_easy"],
"A_H": ["A_only_hard"],
"A_HE": ["A_full_manip"]
},
{
"C_C": ["C_full_control"],
"C_E": ["C_only_easy"],
"C_H": ["C_only_hard"],
"C_HE": ["C_full_manip"]
},
{
"D_C": ["D_full_control"],
"D_E": ["D_only_easy"],
"D_H": ["D_only_hard"],
"D_HE": ["D_full_manip"]
}]}
The other issue I'm having is more minor, but I'd like to have a passage and a comprehension question with answer choices on the same page. I want the passage below the question but that means I need to get the choice to appear between the two stimuli. Is there a way to do that?
Thank you
RE: randomizing blocks and response choice display - Ting - 08-21-2023
Hi Kole!
Can you provide some more details on a point we find slightly confusing. My interpretation is that you have several conditions, and participants can be assigned into any number of those conditions ("vary randomly"), but for each condition they are assigned to, they will see all passages in that condition. Is that right?
For example, if there are conditions A B C, participants can see A B, or B C, but in either case, they'll see all sentences in those assigned conditions.
If we are interpreting it wrong, please let us know!
---
For your second question, unfortunately we only allow a single vertical flow of stimuli and responses (from top to bottom) at this moment. We have a plan for introducing a more flexible placement system, but that's unlikely to be available in 2023. Sorry about the disappointing news!
RE: randomizing blocks and response choice display - kolenorberg - 08-22-2023
Hi! Thanks for the response.
Instead of 4 conditions, I can get away with 3, so for simplicity, I'll go with that. In this instance, every participant sees every condition but what is critical is that every condition occurs in a random order AND the passages occur in a random order AND for each participant, the condition assignment for the passage is random.
So condition order possibility is a matrix of
A, B, C
A, C, B
B, A, C
B, C, A
C, A, B
C, B, A
Passage order is a matrix of
T1, T2, T3
T1, T3, T2,
T2, T1, T3
etc
And then each passage can receive each condition so for any given participant it could look like the below or any space in between.
T1-A, T2-B, T3-C,
...
T1-C, T2-A, T3-B
...
T3-A, T1-C, T2-B
I hope that makes it clearer?
One possibility is to list every combination as a list, but I'm hoping to avoid that as clearly the number of lists would be enormous. I'm hoping to just assign each randomly. So each participant gets assigned the topics in a random order and then when they receive the passage for that topic, they're assigned a random condition for that passage. This might mean the assignment isn't perfectly even as described above (some participants might receive all condition A randomly), but that will ok. I think it will balance out. Though, ideally there would be a way to ensure that after receiving condition A, they can't receive it again.
RE: randomizing blocks and response choice display - Ting - 08-22-2023
I think something that is worth looking into is $factorize instead of $randomize: https://docs.findingfive.com/en/grammar/procedures-blocking/block-sequence#randomized-sequence-per-group-balanced-assignment
But there seems to be a tricky part in your design - how does T1, T2 and T3 differ from each other? Do they contain different passages? Or just the same set of passages in different orders? If the latter, you can implement two layers of randomization: the $factorize method at the block level, and another randomization at the trial template level: https://docs.findingfive.com/en/grammar/trial-templates/basic#order
RE: randomizing blocks and response choice display - kolenorberg - 08-26-2023
Thanks. it was the other way around though. T1, T2, and T3 are different topics and participants need to see all of the topics. They also need to see all of the conditions but they only receive one condition per topic. I finally settled on the below, which will work well enough.
"block_sequence": [
{
"AC-CE-DH": {"$randomize":["A_full_control","C_only_easy","D_only_hard"]},
"AC-CH-DE": {"$randomize":["A_full_control","C_only_hard","D_only_easy"]},
"AH-CC-DE": {"$randomize":["C_full_control","A_only_hard","D_only_easy"]},
"AE-CC-DH": {"$randomize":["C_full_control","A_only_easy","D_only_hard"]},
"AH-CE-DC": {"$randomize":["D_full_control","A_only_hard","C_only_easy"]},
"AE-CH-DC": {"$randomize":["D_full_control","A_only_easy","C_only_hard"]}
}
]