Applying pre-existing randomisation by participant number

Comments

• Roelof

  July 2017

  Hi Kezzo,

  The idea of a double blind study is a very good one, I like it. One simple way to do it is copy pasting a random list into an inline script and using the subject number as an index to retrieve the random value at that index.

  I guess one potential problem is that if you completely randomize an long list there is a slight chance that you will get a repetition of condition 1 for the first xx subjects.

  A primitive way to avoid that:

  import random   
  nr_subjects = 100
  batch_nr = 10
  total = []
  for i in range(0,nr_subjects/batch_nr):
      nr = [0,1]*batch_nr
      random.shuffle(nr)
      total = (total + nr)
  

  This way you randomize per 'batch' so that after each number of 10 subjects your design is balanced, of course you can reduce or increase this batch number according to your needs, and indeed once you have the random condition variable you can use it in the same manner as described in the example experiments. Hope this helps, good luck
• kezzo2002

  July 2017

  Hi Roelof,

  Thanks for your response, you have come at the problem in a slightly different way to how I had thought about it. Yours is much more elegant I think. Really helpful.

  I am new to open sesame and doing inline scripts so bear with me. So just to make sure I understand how this works, in the most basic example if I added the code below into an inline script:

  random_list = [1,1,1,0,0,0]
  var.condition = random_list[var.subject_nr]

  I would then have a new variable [condition] I can refer to elsewhere in open sesame in which subject numbers 0, 1 and 2 have a value for [condition] of 1. And subject numbers 3, 4 and 5 have a value for [condition] of 0. And any other subject numbers will not have a value for the variable [condition]. Is that correct?

  If so this will almost certainly do what I require - but just as a quick extension... is there anyway the random_list can be retrieved from an external file (i.e., an excel) - that way I would never even have to see the random list (as I would if I copy & paste it into inline script) so there can be no possible way I, for example, might remember the first few in the sequence.
• Roelof

  July 2017 edited July 2017

  Yes, you are completely correct about the description of the assignment of condition number and subject numbers. There is a way to store the codes in an external file and read in the number in an inline script definitely, you have to be careful with running the experiment on multiple platforms though since if the folder structure is different you would have to change the working directory ('log_path') for each device.

  #before the experiment in a python interpreter (e.g. spyder):
  import random   
  nr_subjects = 100
  batch_nr = 10
  total = []
  for i in range(0,nr_subjects/batch_nr):
      nr = [0,1]*batch_nr
      random.shuffle(nr)
      total = (total + nr)
  
  file_name = 'random_numbers'
  log_path = 'C:\your folder here\' 
  f = open('%s\%s.txt' %(log_path, file_name), 'w')
  f.write(str(total))
  f.close()
  

  and then in the inline script:

          #in the inline script in the experiment
          log_path = 'C:\your folder here\' 
          f = open('%s\\%s.txt' %(log_path,file_name),'r+')
          for row in f:
              condition_list = eval(row)
  

  Now you can use the 'condition_list' the same way as before using the subject number as the index
• kezzo2002

  July 2017

  Thats great - thanks Roelof!
• kezzo2002

  August 2017

  @Roelof - I've actually had a bit of trouble implementing this. The bit in spyder works fine (although seems to create twice as many numbers as specified in nr_subjects):

  `import random
  nr_subjects = 100
  batch_nr = 10
  total = []
  for i in range(0,nr_subjects // batch_nr):
  nr = [0,1]*batch_nr
  random.shuffle(nr)
  total = (total + nr)

  file_name = 'random_numbers'
  log_path = '/Users/Kieran/Desktop/'
  f = open('%s\%s.txt' %(log_path, file_name), 'w')
  f.write(str(total))
  f.close()`

  However, in open sesame when I put in the following code:

  `file_name = 'random_numbers'
  log_path = '/Users/Kieran/Desktop/'
  f = open('%s\%s.txt' %(log_path, file_name), 'w')
  for row in f:
  condition_list = eval(row)

  var.condition = condition_list[var.subject_nr]`

  I get this error - IOError: File not open for reading

  Any suggestions?
• Roelof

  August 2017

  Indeed the total number of subjects is twice as large since we are adding two numbers with every execution of the for loop, that is a bit confusing.

  As for the error, if you change the 'w' to an 'r' I think it should work (w= write, r = read)
  f = open('%s\%s.txt' %(log_path, file_name), 'r')

  Let me know if there are any other implementation problems, good luck.