Empirisoft Support

    Welcome to Empirisoft Support
Results 1 to 12 of 12

Thread: Hot to make RWG with the constraint of equal numbers of stimuli of each type?

  1. #1
    Join Date
    Apr 2008
    Posts
    12

    Hot to make RWG with the constraint of equal numbers of stimuli of each type?

    Hi all,

    Apologies for the naive question, but...

    I'm able to do a RWG (randomization within groups) for my stimuli using different rwg values (I have four different sentence types, so I have about 30 groups of four stimuli having the same rwg value), no problem there.

    however, I need an additional constraint: I need for each subject to see an equal number of each of the four different types of stimuli (I don't want my subjects to happen to have, say, 5 stimuli of type A, 10 of type B, 3 of type C, and 12 of type D or anything like that!

    So, I don't know if we require a metafunction here, or if it's possible at all to have such constraint.

    Any ideas would be much appreciated!

    Giorgos

  2. #2
    Join Date
    Nov 2005
    Posts
    3,328
    You could take the approach of assigning putting the 30 stimuli from each group into four different stim lists. Say list1.txt, list2, list3 and list4. Then randomly draw from them (without replacement) in 30 blocks of 4 trials like this:

    $list1 (wgr=1)
    $list2 (wgr=1)
    $list3 (wgr=1)
    $list4 (wgr=1)
    $list1 (wgr=2)
    $list2 (wgr=2)
    $list3 (wgr=2)
    $list4 (wgr=2)
    ...
    $list1 (wgr=30)
    $list2 (wgr=30)
    $list3 (wgr=30)
    $list4 (wgr=30)

    Or some variation on that. Does that make sense?

  3. #3
    Join Date
    Apr 2008
    Posts
    12

    Thank you

    Thanks very much, I think it does, will try it and let you know,

    Cheers!

    Giorgos

  4. #4
    Join Date
    Apr 2008
    Posts
    12

    Still problems with randomization

    Hi there, sorry to bother you again, but I seem to have mislead you with the description of the stimuli: You see, I have 30 stimuli of four different versions each, but my constraint is not that I have to have every single version of each of the 30 stimuli occurring, but, quite on the contrary actually, that:

    - only one version (a,b,c,d) out of four is selected out of each of the 30 stimuli
    - in the end, I should have an equal number of a's with b's, c's, and d's.

    What you have suggested that I should do makes the all four versions of each of the 30 stimuli appear, that's why I'm saying I wasn't clear enough (profuse apologies for that).

    Thanks a mill for your patience and time

    Giorgos

  5. #5
    Join Date
    Nov 2005
    Posts
    3,328
    Not sure what you mean by:

    one version (a,b,c,d) out of four is selected out of each of the 30 stimuli
    Do you mean that you have 30 a's, 30 b's, 30 c's and 30 d's? And that you want to pick one from each so that you have a set of 4? And then that you want to do this over and over again (10 times? 30 times?) so that the subject is given multiple sets like this? And is the order within each chosen set also to be random (e.g., adcb, bcad, etc.)?

  6. #6
    Join Date
    Apr 2008
    Posts
    12

    Clarifications

    Hi,

    thanks for your time on this.

    So, basically, I have:

    1a, 1b, 1c, 1d
    2a, 2b, 2c, 2d
    ....................
    40a, 40b, 40c, 40d

    Now, for each number (i.e., 1, 2, etc) I want for each session to randomly pick out only one version (a,b,c,d) for each number. So, if 1a appears, I don't want any 1b,1c, or 1d appear. Plus, in the end, I want to have for each session 10 a's, 10 b's 10 c's and 10 d's. But no stimulus appearing in more than one version in one session either. Does that help at all?

    Thanks so much

    Giorgos

  7. #7
    Join Date
    Apr 2008
    Posts
    12
    So, yes, the order would ideally differ as well, I don't need it to be fixed at all.

    Cheers

    G.

  8. #8
    Join Date
    Apr 2008
    Posts
    12
    And ideally I want each subject to see a total of 40 stimuli, different version (b instead of a), and different order (stimulus "3" in the place of stimulus "1") for each subject....

    Cheers

    G.

  9. #9
    Join Date
    Nov 2005
    Posts
    3,328
    So you want to control the pattern of randomization across subjects? So, e.g., if you have 100 subjects, then 25 will have seen 1a, 25 will have seen 1b, 25 will have seen 1c and 25 will have seen 1d?

  10. #10
    Join Date
    Apr 2008
    Posts
    12
    Hello, thanks for your response. Yes, that's right, sorry for not mentioning that.

    Cheers

    G.

  11. #11
    Join Date
    Nov 2005
    Posts
    3,328
    Ahh, ok, then the answer is easy--no, there is no way to do that automatically with DirectRT. Each session is independent. Between subject factors need to be dealt with via different input files.

  12. #12
    Join Date
    Apr 2008
    Posts
    12
    Cheers, sorry for all the fuss

    G

Similar Threads

  1. run time error '13': type mismatch
    By joh in forum MediaLab Older Versions: Troubleshooting
    Replies: 2
    Last Post: 02-28-2008, 05:58 AM
  2. Using slider item type results in <> sorrounding question text
    By epl-israel in forum MediaLab Older Versions: Troubleshooting
    Replies: 0
    Last Post: 11-12-2007, 09:25 AM
  3. Can MediaLab make printed questionnaires?
    By jarvis24 in forum MediaLab Older Versions: General Questions
    Replies: 1
    Last Post: 11-22-2005, 10:18 AM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •