Millisecond Forums

/ trials = [1-30 = noreplacenorepeat(a,b,c)]
will give you 10 a-trials, 10 b-trials and 10 c-trials in random order, while trying to make sure none repeats (e.g. no a-a).

This is about *trial* selection, not *stimulus* selection. They are not necessarily the same.

Thanks a ton Dave! That absolutely answered my question! My stimuli are in different trials, sorry I didn't explain this well.
Thanks again - G

Hi Dave,

So, in a recent pilot, my colleagues and I had set up a block with trials that had to be presented an equal number of times, but this did not happen. When we looked at the data after running a pilot, we found that the trials were not presented equally, as in the number of presentations varied across participants and was not kept equal.  What did we do wrong? Would you kindly help us understand this? Here is an example of what we did. Thanks a ton for your help!

<block myblock>
/ bgstim = (Alabel, Blabel)
/ trials = [1-40 = noreplacenorepeat(A1_Set1, B1_Set1, A2_Set2, B2_Set2)]
/ recorddata = true
</block>

<trial A1_Set1>
/ validresponse = ("V", "B")
/ correctresponse = ("V", "B")
/ stimulustimes = [0=Fixation; 250=A1; 500=Set1]
/ trialdata = [A1, Set1]
</trial>

<trial A2_Set2>
/ validresponse = ("V", "B")
/ correctresponse = ("V", "B")
/ stimulustimes = [0=Fixation; 250=A2; 500=Set2]
/ trialdata = [A2, Set2]
</trial>

<trial B1_Set1>
/ validresponse = ("V", "B")
/ correctresponse = ("V", "B")
/ stimulustimes = [0=Fixation; 250=B1; 500=Set1]
/ trialdata = [B1, Set1]
</trial>

<trial B2_Set2>
/ validresponse = ("V", "B")
/ correctresponse = ("V", "B")
/ stimulustimes = [0=Fixation; 250=B2; 500=Set2]
/ trialdata = [B2, Set2]
</trial>

<picture Set1>
/ items = Set1
/ select = noreplace
</picture>

<item Set1>
/1 = "red.png"
/2 = "green.png"
/3 = "blue.png"
/4 = "yellow.png"
</item>

<item Set2>
/1 = "black.png"
/2 = "white.png"
/3 = "brown.png"
/4 = "orange.png"
</item>

<picture A1>
/ items = ("A1.png")
/ position = (50, 50)
/ size = (35%, 35%)
</picture>

<picture B1>
/ items = ("B1.png")
/ position = (50, 50)
/ size = (35%, 35%)
</picture>

<picture A2>
/ items = ("A2.png")
/ position = (50, 50)
/ size = (35%, 35%)
</picture>

<picture B2>
/ items = ("B2.png")
/ position = (50, 50)
/ size = (35%, 35%)
</picture>

/ trials = [1-40 = noreplacenorepeat(A1_Set1, B1_Set1, A2_Set2, B2_Set2)]

means you'll get 10 x A1_Set1, 10 x B1_Set1, 10 x A2_Set2 and 10 x B2_Set2. I'm pretty sure that's exactly what your data file will show.

If not, can you provide some example data that shows what you describe?

Thanks Dave,
That is exactly what we expected! In the example block I pasted previously, A1, B1, A2 and B2 are pictures that will be followed immediately by pictures of colors. This was set-up as separate trials. In our actual study, we have done the same thing with 256 trials where we wanted the 4 trials to be equally presented 64 times. So, each trial had 16X4(categories)=64 items ( (as opposed to this example) and we had used 'norepeatnoreplace' command as well. We expected that the recorded data would yield 256 trials of which A1 must have been presented with all 64 items (all 4categoriesX16), A2 is presented with all 64 items (all 4categoriesX16) and same with B1 and B2. Of the 16 items from each category, only 8 are relevant and 8 are irrelavant. Therefore, after coding in SPSS, when we created new variables such as A1 cat1, A1cat2, A1cat3, A1 cat4, for the 32 items from set1. Each of these variables had to have 8 items but they all come up as unequal numbers. I am attaching the excel file (SPSS cannot be uploaded). 

> A1 must have been presented with all 64 items (all 4categoriesX16), A2 is presented with all 64 items (all 4categoriesX16) and
> same with B1 and B2. [...]

No! As I've remarked here https://www.millisecond.com/forums/Topic14239.aspx#bm14241, *trial* selection and *stimulus* selection are *not the same thing*. Look: You have *two* <trial> elements

<trial A1_Set1>
/ validresponse = ("V", "B")
/ correctresponse = ("V", "B")
/ stimulustimes = [0=Fixation; 250=A1; 500=Set1]
/ trialdata = [A1, Set1]
</trial>

<trial B1_Set1>
/ validresponse = ("V", "B")
/ correctresponse = ("V", "B")
/ stimulustimes = [0=Fixation; 250=B1; 500=Set1]
/ trialdata = [B1, Set1]
</trial>

sampling from Set1 (which has 4 items), and *two* <trial> elements

<trial A2_Set2>
/ validresponse = ("V", "B")
/ correctresponse = ("V", "B")
/ stimulustimes = [0=Fixation; 250=A2; 500=Set2]
/ trialdata = [A2, Set2]
</trial>

<trial B2_Set2>
/ validresponse = ("V", "B")
/ correctresponse = ("V", "B")
/ stimulustimes = [0=Fixation; 250=B2; 500=Set2]
/ trialdata = [B2, Set2]
</trial>

sampling from Set2 (also 4 items).


Practical exercise: Take a deck of 4 cards (e.g. the 4 Aces), shuffle the deck and put it in front of you. Draw a card from the deck alternatingly with your left hand and your right hand. Write down the cards you've drawn with your left and your right hand respectively. Once all 4 cards have been drawn, re-shuffle the deck and repeat the process once more. Overall, you will have drawn 4 cards with your left hand and 4 cards with your right hand.

Question: Is there any reason whatsoever that you should have ended up drawing the same 4 cards with your left and your right hand?

Ah! I see what you're saying! Repeating the same set (set1) for A1 and and B1 or set2 for A2 and B2 created this issue! :( So, for example, if we were to change the name of set1 to 'set1a' to be presented with A1 and 'set1b' to be presented with B1, would this issue still occur because the sets are now sort of different but the contents are the same? Or as you imply in your deck of cards analogy, would this still be an issue because inquisit is recognizing the item being presented as it notes down not to repeat? If I am wrong again, would you please direct me to a script where this sort of experimentation has been used? Thanks a ton Dave!

> if we were to change the name of set1 to 'set1a' to be presented with A1 and 'set1b' to be presented with B1, would this issue
> still occur because the sets are now sort of different but the contents are the same?

That would be the equivalent of having *two* identical decks of cards -- one to drawn from with your left hand, the other one to draw from with your right hand. Just to make sure that I've gotten my point across (because it's important; I'm not doing this to annoy you), let me bounce the question back to you: Does this solve your problem? Tip: When in doubt, just try. It's easy to set up a simple test script to verify this for yourself.

Oh! please don't think you're annoying me at all! You are helping me learn and I greatly appreciate it!! To answer your question, I think it would solve the issue because when set1 is used for both A1 and B1, it is mimicking the single deck and creating unequal number of presentations with A1 and B1 drawing from the same set, but by using different names, its like creating two identical decks, one each for A1 and B1. So, in my mind, I am thinking that A1 will be presented with all the items in set1a and B1 will be presented with all the items in set1b. I would like to test this out right away but my home version cannot record data so I cannot tell you for sure with proof that it works or not. Either ways, I think the way our script is set up (or even to rename etc) it seems crude, so please let us know, if its not too much to ask, what would be the best way to set this up (efficiently) such that it is ensured that all set1 items are presented with A1 and B1 but randomly and equally. Thank you very much!