[solved] Selective sampling... with a ratio!?!? Help please!

Comments

• LeeB_86

  June 2013 edited 1:36PM

  ... Just as a postscript: I also realise that I will need to put the solution in a sketchpad item...
• lvanderlinden

  June 2013 edited June 2013

  Hi Lee,

  To randomly draw samples containing only half of the 60 non-rhyming and 40 rhyming word pairs, and to combine the two into one randomised block list, you will need some Python inline scripting.

  For more info on the use Python inline code in OpenSesame, see:

  • http://osdoc.cogsci.nl/python-inline-code/

  More specifically:

  1.) Append an inline_script item at the very beginning of your experiment where you define the complete list of non-rhyming and rhyming word pairs, for example by reading them from a .csv file. The two .csv files have to placed either in the file pool or in the same folder as your experiment.

  Place something like the following code in the Prepare-phase tab of the inline_script item (see comments for more information):

  # Read the lists of the 40 rhyming pairs and the 60
  # non-rhyming pairs from two separate .csv files.
  
  # Specify the paths to the files containing the pairs
  # by using the built-in OpenSesame function exp.get_file.
  # For more info, see:
  # http://osdoc.cogsci.nl/python-inline-code/experiment-functions/#experiment.get_file
  path_rhymes = exp.get_file("rhyme_pairs.csv")
  path_non_rhymes = exp.get_file("non_rhyme_pairs.csv")
  # We'll use the built-in Python module numpy to read our .csv file:
  # Import the module:
  import numpy as np
  
  # And load the text file.
  # For more info, see:
  # http://docs.scipy.org/doc/numpy/reference/generated/numpy.loadtxt.html
  list_rhymes = np.loadtxt(path_rhymes, dtype = str)
  list_non_rhymes = np.loadtxt(path_non_rhymes, dtype = str)
  
  # Make the two lists global for future use:
  global list_rhymes, list_non_rhymes
  

  2.) Append an inline_script item at the very beginning of your block_sequence, where you will randomly draw two samples containing only half of the word-pairs from the original lists. Use something like the following code (in the Prepare-phase tab) to do so:

  # Randomise the from-the-csv-files-read lists
  # and draw two samples containing the necessary numbers of
  # rhymes and non-rhymes, respectively.
  
  # First, we randomise the lists using the function random.shuffle() from 
  # the built-in Python module random:
  # For more info, see:
  # http://docs.python.org/2/library/random.html#random.shuffle
  
  import random
  random.shuffle(list_rhymes)
  random.shuffle(list_non_rhymes)
  
  # We use the function random.sample()t to obtain two lists containing 
  # the desired number of rhyme-pairs and non-rhyme pairs.
  
  # The number of pairs per list list is equal to half of the 
  # number of pairs in the original list (i.e. 20 rhyme-pairs and
  # 30 non-rhyme-pairs).
  
  # For more info, see: 
  # http://docs.python.org/2/library/random.html#random.sample
  
  samples_rhymes = random.sample(list_rhymes,len(list_rhymes)/2)
  samples_non_rhymes = random.sample(list_non_rhymes,len(list_non_rhymes)/2)
  
  # Next, we merge the two lists to obtain one block list:
  block_list = samples_rhymes + samples_non_rhymes
  
  # Shuffle the new list:
  random.shuffle(block_list)
  
  # Finally, we make the block_list global for future use:
  global block_list
  

  3.) Append an inline_script item to the beginning of your trial_sequence to draw one word pair from the block list, without replacement (again, place the code in the Prepare-phase tab):

  # Determine the word pair of the current trial by drawing one word
  # pair from the previously-defined block list.
  
  # To avoid that some pairs are shown twice, and others never, we draw
  # the pairs without replacement by using the built-in Python function
  # pop():
  # For more info, see:
  # http://docs.python.org/2/tutorial/datastructures.html
  
  stim_pair = block_list.pop()
  
  # The pair is separated by a comma (because they come from a .csv
  # file). We split them by using the built-in Python function split():
  # For more info, see:
  # http://docs.python.org/2/library/stdtypes.html#str.split
  word1, word2 = stim_pair.split(",")
  
  # Finally, to make the variables word1 and word2 available in the GUI
  # (e.g. a sketchpad item), we use the built-in OpenSesame function
  # exp.set():
  
  # Now, we can use the square-bracket method to display the values of those
  # two variables in the sketchphad item.
  # For more info, see:
  # http://osdoc.cogsci.nl/usage/variables-and-conditional-qifq-statements/
  
  exp.set("word1", word1)
  exp.set("word2", word2)
  

  4.) Finally, set the appropriate number of cyles (in your case 50) in the block_loop and use the square-bracket method to display word1 and word2 in a sketchpad item.

  After applying this, your overview area should look something like this:

  

  A working example experiment can be downloaded here (rename the extension from .txt into .opensesame.tar.gz and open as normally). Note that two .csv files with the appropriate names are assumed to be placed either in the file pool or in the same folder as your experiment.

  • http://pastebin.com/uhd62pE2

  Please bear in mind I am oblivious to the world of programming languages...

  In order to program experiments like the one you're building now, I would strongly advise you to learn some Python. A good start:

  • http://swaroopch.com/notes/python/

  Best,

  Lotje

  Did you like my answer? Feel free to
• LeeB_86

  June 2013 edited 1:36PM

  Lotje,

  That is amazingly helpful, and I really appreciate you taking the time to reply.
  My Python-less-ness is because I'm quite new to this sort of thing (3rd year undergraduate 'mature' student) - it's actually for a final project for next year.

  Whilst I fully intend to become intimately acquainted with Python at some point, at the moment I find myself with more to do than time to do it in!

  Dank je wel,

  Lee
• LeeB_86

  June 2013 edited 1:36PM

  Hi Lotje,

  I've written the inline scripts and my program is identical to the code in your example.... only difference is that your example file works and mine tells me that I have AttributeError: 'numpy.ndarray' object has not attribute 'split'. Having read Byte I now get the terminology of the error but I have tried to find the root of the problem with no luck (after all the split happens in the trial script!) - the only thing that mine has that your doesn't is the rest of my experiment (i.e. instruction screens, countdown timers, 5 further experimental stages). They are all GUI objects and I don't believe that they should affect the inline script.

  Hopefully you have some thoughts? Or anyone?

  Thanks
  Lee
• lvanderlinden

  June 2013 edited 1:36PM

  Hi Lee,

  Would it be possible to upload the spreadsheet you're using (for example on Dropbox or Filedropper, or simply by making a screenshot of it)?

  Also, could you either upload the full code of your experiment (via pastebin, for example) or paste the Python code from your inline_script items here?

  I'm sure we can figure out what's going on.

  Best,

  Lotje

  Did you like my answer? Feel free to
• LeeB_86

  June 2013 edited 1:36PM

  I have dropped you an email with the files in a .zip (I'm at work so it was the only way!)

  Thanks!

  Lee
• lvanderlinden

  June 2013 edited 1:36PM

  Hi Lee,

  Thanks for your email! I can't see why this is going wrong either. However, when I simply open your .csv files in a text editor, resave them with the extension .txt, and change the corresponding piece of coding accordingly:

  path_rhymes = exp.get_file("rhyming_pairs.txt")
  path_non_rhymes = exp.get_file("non_rhyme_pairs.txt")
  

  everything works.

  If I find out why this is the case, I'll let you know.

  Best,

  Lotje

  Did you like my answer? Feel free to
• LeeB_86

  June 2013 edited 1:36PM

  Interesting! I was worried about a knock-on effect from one of the other elements. Hopefully this means that I can get it all working tonight.
  Just as an off topic query, you don't know where I might find 'spatial rotation resources' for experiments by any chance? I am fully prepared to make some, but it would be easier if there is some kind of cognitive-library-of-resources somewhere!

  Thanks!

  Lee
• lvanderlinden

  June 2013 edited June 2013

  Hi Lee,

  No, sorry, I don't know about such a stimulus set...
  Good luck and let us know if you have more questions!

  Lotje

  Did you like my answer? Feel free to
• LeeB_86

  June 2013 edited 1:36PM

  Thanks Lotje, I certainly will

  Lee
• LeeB_86

  June 2013 edited June 2013

  It didn't take long

  I'm trying to relate the keyboard response 'm' to rhyming pairs and the keyboard response 'z' to non-rhyming pairs (as in the instructions for the block), here's what I have so far:

  http://pastebin.com/JmW4yH3Z

  Full code is here:

  http://pastebin.com/PbeekeyB

  This is probably simple fix but I'm not seeing the wood for the trees!

  Thanks,

  Lee
• lvanderlinden

  June 2013 edited June 2013

  Hi Lee,

  Firstly, you should separate the allowed responses in the keyboard_response with a semicolon instead of a comma. So: m;z.

  Secondly, to determine whether a given stimulus pair comes from the rhyming or the non-rhyming list, and to define the corresponding correct response, you could use something like the following Python inline code:

  # Determine correct response, depending on whether the words come from 
  # the rhyming or the non-rhyming list:
  
  if stim_pair in list_rhymes:
      correct_keyboard_response = 'm'
  else:
      correct_keyboard_response = 'z'
  
  # Set the correct response by giving the built-in OpenSesame variable 
  # 'correct_response' a value:
  exp.set("correct_response", correct_keyboard_response)
  

  Good luck!

  Lotje

  Did you like my answer? Feel free to
• LeeB_86

  June 2013 edited 1:36PM

  It's amazing how clear things become after they are explained by an 'expert other'

  I did actually continue with my own efforts after posting, but it was fruitless as I tried to set 'list_rhymes' as a range and search for the stimulus in it... didn't work....

  However I am learning a lot through your helpful guidance!

  Thanks once more!

  Lee