Millisecond Forums

Group variables: random and specific presentation

By paughl - 10/29/2014

Hi Folks,

--I have 2 versions of each stim.
--Participants should only ever get 1 of the versions of each stim.
--Participants should get 50% of Version 1 and 50% of Version 2 (ie: half the participants get the alternate version).
--AND THIS IS THE BIT I NEED HELP WITH: For all participants, the first stim they get should always be a randomly selected Version 1 stim.  This stim should not be repeated later.

Here's how I've done everything except the last point above (s=Version1, ns=Version2):

/ group = (1 of 2) (sit1=s1, sit2=ns2, sit3=s3, sit4=ns4, sit5=s5, sit6=ns6, sit7=s7, sit8=ns8, sit9=s9, sit10=ns10,
sit11=s11, sit12=ns12, sit13=s13, sit14=ns14, sit15=s15, sit16=ns16, sit17=s17, sit18=ns18, sit19=s19, sit20=ns20)
/ group = (2 of 2) (sit1=ns1, sit2=s2, sit3=ns3, sit4=s4, sit5=ns5, sit6=s6, sit7=ns7, sit8=s8, sit9=ns9, sit10=s10,
sit11=ns11, sit12=s12, sit13=ns13, sit14=s14, sit15=ns15, sit16=s16, sit17=ns17, sit18=s18, sit19=ns19, sit20=s20)

Many thanks in advance!

By Dave - 10/29/2014

Impossible to answer based on the provided information. For example, it isn't at all clear what e.g. "sit1" is. Is it a <text> or <picture> element? Is it a <trial> element?  Or perhaps a <block>?

Since you do not explain at all how you select stimuli in the first place, etc., the question of how to force the selection of a certain type of stimulus cannot be answered either.
By paughl - 10/29/2014


Attached is the full script.

By Dave - 10/29/2014

This seems like an awkward way to set this up. There should be no reason whatsoever to have separate a separate <block>, <trial>, <video>, etc. elements for each single scenario.

At any rate, the way you designed that procedure, there is no way to force a certain type of stimulus as the 1st trial.

All you need is one ns-<block> element, one ns-<trial> element, one ns-<video> element. Same for "s". Then simply switch <item> elements via the <variables>.

In essence:
<block myblock>
/ trials = [1=uc_trial; 2-4=noreplace(uc_trial,lc_trial,lc_trial)]

/ group = (1 of 2) (ucitems=g1uc; lcitems=g1lc)
/ group = (2 of 2) (ucitems=g2uc; lcitems=g2lc)

<trial uc_trial>
/ stimulusframes = [1=uppercase]
/ validresponse = (57)

<trial lc_trial>
/ stimulusframes = [1=lowercase]
/ validresponse = (57)

<text uppercase>
/ items = ucitems

<text lowercase>
/ items = lcitems

<item g1uc>
/ 1 = "A"
/ 2 = "B"

<item g1lc>
/ 1 = "c"
/ 2 = "d"

<item g2uc>
/ 1 = "C"
/ 2 = "D"

<item g2lc>
/ 1 = "a"
/ 2 = "b"
By paughl - 10/29/2014

Thanks Dave. Makes sense.
By paughl - 10/30/2014

Hi Dave,

In trying to implement this suggestion, I come up with an issue:

My experiment involves this kind of sequence for every stim:

instruction  >> audio >> post audio question page (which includes multiple elements)

As the audio and the question page need to match up, how do i use your suggestion above with the item element and group variables?

Many thanks,

By Dave - 10/30/2014

You would do this the same way as always -- the existence of <variables> doesn't matter.

For pairing items / different stimuli, see the "How to Present Stimulus Pairs" topic in the documentation.

Alternatively, you can /branch based on the stimulus item presented.
By paughl - 11/3/2014

thank you Dave! :)