Counterbalancing stimulus positions

Comments

• rrr

  January 2017 edited January 2017

  Hello,

  I made a script without using balanced_latin_square function as below.
  In the script, the positions are just randomized but not counterbalanced.

  ########### EXPERIMENTAL DESIGN ####################
  #import 2 lists of image file#
  #randomizing the two lists#
  listAB = [[i, j] for i, j in zip(listA, listB)]
  
  list_former = [listAB[j - 1] for j in random.sample([i + 1 for i in range(len(listAB))], int(len(listAB) / 2))]
  list_latter = [[i[1], i[0]] for i in tuple(set(map(tuple, list_former)) ^ set(map(tuple, listAB)))]
  list_former.extend(list_latter)
  all = [list_former[i] for i in random.sample(range(len(list_former)), len(list_former))]
  
  listLEFT = [z[0] for z in all]
  listRIGHT = [z[1] for z in all]
  
  t_fix = 1000
  ######### INITIALIZE ##############
  control.initialize(exp)
  
  fixcross = stimuli.FixCross()
  fixcross.preload()
  
  block = design.Block(name="block_all")
  for (l, r) in zip(listLEFT, listRIGHT):
      trial = design.Trial()
      trial.set_factor("Left_im", l)
      trial.set_factor("Right_im", r)
      canvas = stimuli.BlankScreen()
      l_image = stimuli.Picture(l, position=(-200, 0))
      r_image = stimuli.Picture(r, position=(200, 0))
      l_image.plot(canvas)
      r_image.plot(canvas)
      l_image.preload()
      r_image.preload()
      canvas.preload()
      trial.add_stimulus(canvas)
      block.add_trial(trial)
  block.shuffle_trials(block)
  exp.add_block(block)
  
  exp.data_variable_names = ["block", "trial", "Left_im", "Right_im", "Key", "RT"]
  ######### START ##############
  control.start(exp)
  
  for block in exp.blocks:
      for trial in block.trials:
          fixcross.present()
          exp.clock.wait(t_fix)
          canvas.present()
          btn, rt = exp.keyboard.wait([misc.constants.K_LEFT, misc.constants.K_RIGHT])
          exp.data.add([block.name, trial.id, trial.get_factor("Left_im"), trial.get_factor("Right_im"), btn, rt])
  

  In this case, the two images presented in each trial are always same while the exported data shows the images in left and right sides changed correctly every trial.

  Is there any way to counterbalancing the positions across participants, and what is the cause that the same stimuli are presented?

  Thanks
• fladd

  January 2017

  Can you provide a working example?

  Warning There was an error rendering this rich post.
• rrr

  February 2017 edited February 2017

  Hi,

  I found the cause of same stimuli.
  It was caused by canvas.present(). It runs as below by changing canvas.present() to trial.stimuli[0].present().

          block = design.Block(name="block_all")
          for (l, r) in zip(listLEFT, listRIGHT):
              trial = design.Trial()
              trial.set_factor("Left_im", l)
              trial.set_factor("Right_im", r)
              canvas = stimuli.BlankScreen()
              l_image = stimuli.Picture(l, position=(-200, 0))
              r_image = stimuli.Picture(r, position=(200, 0))
              l_image.plot(canvas)
              r_image.plot(canvas)
              l_image.preload()
              r_image.preload()
              canvas.preload()
              trial.add_stimulus(canvas)
              trial.preload_stimuli()
              block.add_trial(trial)
          block.shuffle_trials(block)
          exp.add_block(block)
  
      exp.data_variable_names = ["block", "trial", "Left_im", "Right_im", "Key", "RT"]    
          ######### START ##############
          control.start(exp)
  
          for block in exp.blocks:
              for trial in block.trials:
                  fixcross.present()
                  exp.clock.wait(t_fix)
                  trial.stimuli[0].present()
                  btn, rt = exp.keyboard.wait([misc.constants.K_LEFT, misc.constants.K_RIGHT])
                  exp.data.add([block.name, trial.id, trial.get_factor("Left_im"), trial.get_factor("Right_im"), btn, rt])
  
          ####### END EXPERIMENT ########
          control.end()
  

  I'm still looking for the clear way of counterbalancing the positions across participants.
  Are there any suggestions?

  Thanks!
• fladd

  February 2017

  How about creating a between-subject factor "picture_arrangement" with two levels ("x_is_left", "x_is_right"):

  import expyriment
  
  exp = expyriment.design.Experiment("My Experiment")
  exp.add_bws_factor('picture_arrangement', ['x_is_left', 'x_is_right'])
  

  And then later in the experiment (after calling expyriment.control.start(), i.e. after a subject ID has been entered):

  if exp.get_permuted_bws_factor_condition('picture_arrangement') == "x_is_left":
      # Present arrangement
  else:
     # Present other arrangement
  

  Warning There was an error rendering this rich post.