Randomization error


Author
Message
dnaber
dnaber
New Member (12 reputation)New Member (12 reputation)New Member (12 reputation)New Member (12 reputation)New Member (12 reputation)New Member (12 reputation)New Member (12 reputation)New Member (12 reputation)New Member (12 reputation)
Group: Forum Members
Posts: 2, Visits: 6
Hi everyone,

I hope anybody can help me with a maybe somewhat complex problem. I am working on a script in which a total number of 28 profile pages have to be shown to the participants. These pages should contain completely randomized (without replacement) information which are: 1) a foto, 2) firstname, 3) final grade, 4) department, 5) room number. However, the foto and the grade have to combined half systematically, half randomized without replacement. What I need is to have 7 blonde persons having good grades, 7 blonde with bad grades, 7 brunette with good grades and 7 brunette with bad grades. Therefore, I created two item lists in Inquisit for blonde and for brunette (in each item list 14 fotos) and four item lists: two for 7 good grades (one list for brunette and one list for blonde) and two for 7 bad grades (also one list for brunette and one list for blonde). The rest (firstnames, department and room number) is in one list each with 28 stimuli. To give an example, in the list I have written the following (here for blonde fotos. The same holds for brunette and for the final grades):

<item hairbl1>
/ 1 = "blond1.jpg"
/ 2 = "blond2.jpg"

/ 14 = "blond14.jpg"
</item>


<list hairbl2>
/ poolsize = 14
/ selectionmode = random
/ replace = false
/ selectionrate = always
</list>

These lists are converted into profile elements as following:

<picture hairbl3>
/ items = hairbl1
/ select = list.hairbl2.nextindex
/ position = (30,45)
/ size = (70%, 70%)
</picture>


Then, my block looks like this:

<block task2>
/ preinstructions = (main_intro)
/ trials = [1-28=noreplace(trial.blondesm,trial.blondedb,trial.brunsm,trial.brundb)]
</block>


As you can see, I have four different trials (blonde good grades, blonde bad grades, brunette good grades, brunette bad grades) which are randomized without replacement in the block. My trials therefore look like this (this is the one for blonde good grades):

<trial blondesm>
/ ontrialbegin = [
    values.trialblsm = values.trialblsm + 1;
]
/ numframes = 9
/ stimulusframes = [1=firstname4; 2=grade4; 3=depart4; 4=room4; 5=noreplace(text.firstname3); 6=noreplace(text.depart3); 7=noreplace(text.room3); 8=noreplace(text.gradeblsm3); 9=noreplace(picture.hairbl3)]
/ timeout = 1000
/ ontrialend = [
    values.target = picture.hairbl3.currentitem;
    values.firstname = text.firstname3.currentitem;
    values.grade = text.gradeblsm3.currentitem;
    values.depart = text.depart3.currentitem;
    values.room = text.room3.currentitem;
    ]
/ branch=[if (trial.blondesm.latency > 1) surveypage.licoco]
</trial>


And finally, as values I defined at the beginning of the script:

<values >
/ trialblsm = 0
/ trialbldb = 0
/ trialbrsm = 0
/ trialbrdb = 0
/ target = 0
/ firstname = 0
/ grade = 0
/ depart = 0
/ room = 0
</values>


Having collected some test data, I noticed that there are some fotos that are shown more than once so the randomization without replacement apparently did not work. Furthermore, I am not sure if the correct stimulus name is actually written into the values column. For example, I doubt that the name of the foto (e.g. „blond01“) that can be found in the value column (here „target“) is really the foto that the participant has seen.

Has anybody an idea what went wrong? I would be happy if somebody can help me. Thank you in advance!!

Best wishes

Dave
Dave
Supreme Being (855K reputation)Supreme Being (855K reputation)Supreme Being (855K reputation)Supreme Being (855K reputation)Supreme Being (855K reputation)Supreme Being (855K reputation)Supreme Being (855K reputation)Supreme Being (855K reputation)Supreme Being (855K reputation)
Group: Administrators
Posts: 10K, Visits: 59K
dnaber - 9/10/2020
Hi everyone,

I hope anybody can help me with a maybe somewhat complex problem. I am working on a script in which a total number of 28 profile pages have to be shown to the participants. These pages should contain completely randomized (without replacement) information which are: 1) a foto, 2) firstname, 3) final grade, 4) department, 5) room number. However, the foto and the grade have to combined half systematically, half randomized without replacement. What I need is to have 7 blonde persons having good grades, 7 blonde with bad grades, 7 brunette with good grades and 7 brunette with bad grades. Therefore, I created two item lists in Inquisit for blonde and for brunette (in each item list 14 fotos) and four item lists: two for 7 good grades (one list for brunette and one list for blonde) and two for 7 bad grades (also one list for brunette and one list for blonde). The rest (firstnames, department and room number) is in one list each with 28 stimuli. To give an example, in the list I have written the following (here for blonde fotos. The same holds for brunette and for the final grades):

<item hairbl1>
/ 1 = "blond1.jpg"
/ 2 = "blond2.jpg"

/ 14 = "blond14.jpg"
</item>


<list hairbl2>
/ poolsize = 14
/ selectionmode = random
/ replace = false
/ selectionrate = always
</list>

These lists are converted into profile elements as following:

<picture hairbl3>
/ items = hairbl1
/ select = list.hairbl2.nextindex
/ position = (30,45)
/ size = (70%, 70%)
</picture>


Then, my block looks like this:

<block task2>
/ preinstructions = (main_intro)
/ trials = [1-28=noreplace(trial.blondesm,trial.blondedb,trial.brunsm,trial.brundb)]
</block>


As you can see, I have four different trials (blonde good grades, blonde bad grades, brunette good grades, brunette bad grades) which are randomized without replacement in the block. My trials therefore look like this (this is the one for blonde good grades):

<trial blondesm>
/ ontrialbegin = [
    values.trialblsm = values.trialblsm + 1;
]
/ numframes = 9
/ stimulusframes = [1=firstname4; 2=grade4; 3=depart4; 4=room4; 5=noreplace(text.firstname3); 6=noreplace(text.depart3); 7=noreplace(text.room3); 8=noreplace(text.gradeblsm3); 9=noreplace(picture.hairbl3)]
/ timeout = 1000
/ ontrialend = [
    values.target = picture.hairbl3.currentitem;
    values.firstname = text.firstname3.currentitem;
    values.grade = text.gradeblsm3.currentitem;
    values.depart = text.depart3.currentitem;
    values.room = text.room3.currentitem;
    ]
/ branch=[if (trial.blondesm.latency > 1) surveypage.licoco]
</trial>


And finally, as values I defined at the beginning of the script:

<values >
/ trialblsm = 0
/ trialbldb = 0
/ trialbrsm = 0
/ trialbrdb = 0
/ target = 0
/ firstname = 0
/ grade = 0
/ depart = 0
/ room = 0
</values>


Having collected some test data, I noticed that there are some fotos that are shown more than once so the randomization without replacement apparently did not work. Furthermore, I am not sure if the correct stimulus name is actually written into the values column. For example, I doubt that the name of the foto (e.g. „blond01“) that can be found in the value column (here „target“) is really the foto that the participant has seen.

Has anybody an idea what went wrong? I would be happy if somebody can help me. Thank you in advance!!

Best wishes

Please provide the full script and all files it requires to run.
GO

Merge Selected

Merge into selected topic...



Merge into merge target...



Merge into a specific topic ID...




Reading This Topic

Explore
Messages
Mentions
Search