Randomised intermixed trials


Author
Message
Haniel_Do
Haniel_Do
New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)
Group: Forum Members
Posts: 8, Visits: 37
Hi all,
I am constructing an experiment with trials drawn from three separate conditions (A vs. B, C Vs. D, E Vs. F) based of a previous paradigm. However the previous paradigm stated that the three conditions were randomised and intermixed during the trial, I understand how to randomise within condition, but how can I intermix the conditions while ensuring they remain paired (A Vs. B remains so and not randomised to A Vs. D or C Vs. F etc.). I currently have them all coded as separate item attributes if that helps at all i.e. A is item.attributeA, B is item.atributeB, etc. Thanks

Daniel
Dave
Dave
Supreme Being (923K reputation)Supreme Being (923K reputation)Supreme Being (923K reputation)Supreme Being (923K reputation)Supreme Being (923K reputation)Supreme Being (923K reputation)Supreme Being (923K reputation)Supreme Being (923K reputation)Supreme Being (923K reputation)
Group: Administrators
Posts: 11K, Visits: 68K
Haniel_Do - 4/8/2021
Hi all,
I am constructing an experiment with trials drawn from three separate conditions (A vs. B, C Vs. D, E Vs. F) based of a previous paradigm. However the previous paradigm stated that the three conditions were randomised and intermixed during the trial, I understand how to randomise within condition, but how can I intermix the conditions while ensuring they remain paired (A Vs. B remains so and not randomised to A Vs. D or C Vs. F etc.). I currently have them all coded as separate item attributes if that helps at all i.e. A is item.attributeA, B is item.atributeB, etc. Thanks

Daniel

I don't understand what this is supposed to mean. You need to be more specific and precise.
Haniel_Do
Haniel_Do
New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)
Group: Forum Members
Posts: 8, Visits: 37
Dave - 4/8/2021
Haniel_Do - 4/8/2021
Hi all,
I am constructing an experiment with trials drawn from three separate conditions (A vs. B, C Vs. D, E Vs. F) based of a previous paradigm. However the previous paradigm stated that the three conditions were randomised and intermixed during the trial, I understand how to randomise within condition, but how can I intermix the conditions while ensuring they remain paired (A Vs. B remains so and not randomised to A Vs. D or C Vs. F etc.). I currently have them all coded as separate item attributes if that helps at all i.e. A is item.attributeA, B is item.atributeB, etc. Thanks

Daniel

I don't understand what this is supposed to mean. You need to be more specific and precise.

Hi Dave,

Sorry I was trying to not take too long with my explanation but the structure is a bit tricky. Basically what we're trying to recreate is a memory paradigm which has two phases, encoding and recognition. The encoding phase is the one I am a bit stumped on. Basically in the encoding phase the participant is shown adjectives under one of three conditions. The first is Uppercase vs lower case where the participant must respond based on whether the first letter of the word is capitalised or not. The second condition is social desirability, where the participant must respond based on whether the adjective is seen as socially desirable or not and the final is the desirability but for oneself. Each adjective is a timed, forced choice between the two options. In the previous study the three conditions were randomised but also intermixed with each other, so one trial could be upper vs lower case but the next could be social desirability or desirability for self, etc. Thus while I know how to randomise within condition, I am not sure how to then intermix the three conditions randomly but the choices themselves remain within condition i.e. I don't want an adjective choice being between socially desirable and lower case or not socially desirable and desirable for self, etc.. Part of the issue may be that I coded each part of the condition separately (eg. itemattributeA is uppercase and item attributeB is lower case), but I'm basing my code of the example experiments given. Apologies for the long post, hopefully its a bit clearer.

Daniel
Dave
Dave
Supreme Being (923K reputation)Supreme Being (923K reputation)Supreme Being (923K reputation)Supreme Being (923K reputation)Supreme Being (923K reputation)Supreme Being (923K reputation)Supreme Being (923K reputation)Supreme Being (923K reputation)Supreme Being (923K reputation)
Group: Administrators
Posts: 11K, Visits: 68K
Haniel_Do - 4/8/2021
Dave - 4/8/2021
Haniel_Do - 4/8/2021
Hi all,
I am constructing an experiment with trials drawn from three separate conditions (A vs. B, C Vs. D, E Vs. F) based of a previous paradigm. However the previous paradigm stated that the three conditions were randomised and intermixed during the trial, I understand how to randomise within condition, but how can I intermix the conditions while ensuring they remain paired (A Vs. B remains so and not randomised to A Vs. D or C Vs. F etc.). I currently have them all coded as separate item attributes if that helps at all i.e. A is item.attributeA, B is item.atributeB, etc. Thanks

Daniel

I don't understand what this is supposed to mean. You need to be more specific and precise.

Hi Dave,

Sorry I was trying to not take too long with my explanation but the structure is a bit tricky. Basically what we're trying to recreate is a memory paradigm which has two phases, encoding and recognition. The encoding phase is the one I am a bit stumped on. Basically in the encoding phase the participant is shown adjectives under one of three conditions. The first is Uppercase vs lower case where the participant must respond based on whether the first letter of the word is capitalised or not. The second condition is social desirability, where the participant must respond based on whether the adjective is seen as socially desirable or not and the final is the desirability but for oneself. Each adjective is a timed, forced choice between the two options. In the previous study the three conditions were randomised but also intermixed with each other, so one trial could be upper vs lower case but the next could be social desirability or desirability for self, etc. Thus while I know how to randomise within condition, I am not sure how to then intermix the three conditions randomly but the choices themselves remain within condition i.e. I don't want an adjective choice being between socially desirable and lower case or not socially desirable and desirable for self, etc.. Part of the issue may be that I coded each part of the condition separately (eg. itemattributeA is uppercase and item attributeB is lower case), but I'm basing my code of the example experiments given. Apologies for the long post, hopefully its a bit clearer.

Daniel

Set up separate trials for each kind of choice, have those trials display the appropriate stimulus and reminders, sample those trials randomly at the block level. I don't understand why you believe there could be any kind of condition mix-up,.

<block example>
/ trials = [1-24 = noreplace(a,b,c,d,e,f)]
</block>

<trial a>
/ stimulusframes = [1=a, areminder, breminder]
/ validresponse = ("e", "i")
/ correctresponse = ("e")
</trial>

<trial b>
/ stimulusframes = [1=b, areminder, breminder]
/ validresponse = ("e", "i")
/ correctresponse = ("i")
</trial>

<trial c>
/ stimulusframes = [1=c, creminder, dreminder]
/ validresponse = ("e", "i")
/ correctresponse = ("e")
</trial>

<trial d>
/ stimulusframes = [1=d, creminder, dreminder]
/ validresponse = ("e", "i")
/ correctresponse = ("i")
</trial>

<trial e>
/ stimulusframes = [1=e, ereminder, freminder]
/ validresponse = ("e", "i")
/ correctresponse = ("e")
</trial>

<trial f>
/ stimulusframes = [1=f, ereminder, freminder]
/ validresponse = ("e", "i")
/ correctresponse = ("i")
</trial>

<text a>
/ items = aitems
</text>

<item aitems>
/ 1 = "A1"
/ 2 = "A2"
/ 3 = "A3"
/ 4 = "A4"
</item>

<text b>
/ items = bitems
</text>

<item bitems>
/ 1 = "B1"
/ 2 = "B2"
/ 3 = "B3"
/ 4 = "B4"
</item>

<text areminder>
/ items = ("Press E for A")
/ position = (20%, 10%)
</text>

<text breminder>
/ items = ("Press I for B")
/ position = (80%, 10%)
</text>

<text c>
/ items = citems
</text>

<item citems>
/ 1 = "C1"
/ 2 = "C2"
/ 3 = "C3"
/ 4 = "C4"
</item>

<text d>
/ items = ditems
</text>

<text creminder>
/ items = ("Press E for C")
/ position = (20%, 10%)
</text>

<text dreminder>
/ items = ("Press I for D")
/ position = (80%, 10%)
</text>

<item ditems>
/ 1 = "D1"
/ 2 = "D2"
/ 3 = "D3"
/ 4 = "D4"
</item>

<text e>
/ items = eitems
</text>

<item eitems>
/ 1 = "E1"
/ 2 = "E2"
/ 3 = "E3"
/ 4 = "E4"
</item>

<text f>
/ items = fitems
</text>

<item fitems>
/ 1 = "F1"
/ 2 = "F2"
/ 3 = "F3"
/ 4 = "F4"
</item>

<text ereminder>
/ items = ("Press E for E")
/ position = (20%, 10%)
</text>

<text freminder>
/ items = ("Press I for F")
/ position = (80%, 10%)
</text>


Haniel_Do
Haniel_Do
New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)New Member (22 reputation)
Group: Forum Members
Posts: 8, Visits: 37
Hi Dave,

Thanks for the reply, this is much clearer! I think I may have been looking at the wrong example format because the way they coded the trials were completely different but this makes a lot more sense. Thanks again, appreciate the help.

Daniel
GO

Merge Selected

Merge into selected topic...



Merge into merge target...



Merge into a specific topic ID...




Reading This Topic

Explore
Messages
Mentions
Search