feedback on accuracy

Nolan, Karen Nolan at NKI.RFMH.ORG
Wed Jul 23 13:29:49 UTC 2003


Taryn  - I would only use the Feedback object if you want to give trial by
trial feedback. To display block by block feedback,
I set up a summation object (TC) and a counter (Trials) prior to the block.

Inline 1
Set TC=New Summation
Trials = 0

As part of the trial procedure, immediately after the stimulus display
("Stim"), a second inline keeps track of number of trials and total correct:

Inline 2
If Stim.ACC =1then
TC.AddObservation c.GetAttrib("Stim.ACC")
End if
Trials = Trials+1

Immediately after the List (block)I use a third inline to set the text
("Message") to be displayed in a text display object

Inline 3
acc = CInt((Tc.Total/Trials)*100)
c.SetAttrib "Message", "You scored " & acc & "% correct"

I hope this is helpful
Karen Nolan

 -----Original Message-----
From: 	Taryn Tacosa [mailto:ttacosa at hotmail.com]
Sent:	Monday, July 21, 2003 1:04 PM
To:	eprime at mail.talkbank.org
Subject:	RE: feedback on accuracy

Thank you for the response. I will try to be more specific this time.

What I would like this feedback screen to look like is actually very simple.

Feedback is being presented for the subject's benefit only and will not be
the focus of the analysis, though we are logging all response values
(standard logging values are checked). Specifically, a subject sees a series

of 16 letters one at a time and is required to make (1) response to each
letter. This response is either correct (ACC=1) or incorrect (ACC=0). After
the series ends, I want the subject to see: "You got ##% correct". So, if
the subject answered correctly to 14 out of the 16 letters, he would see
"You got 87.5% correct!".

I have seen that the feedback object has "@ACC.MEAN" as one of its options,
however, when I run the program using this (and citing the correct Input
Object), the screen still gives accuracy for just the last response given
(0% or 100%), rather than across all 16 responses. Is there something
additional I'm required to do to enable this option to look at accuracy for
all 16 responses?

I am thinking it may be easiest to not use the feedback display and to just
use a counter to record accuracies and then write a simple formula for the %

on a TextDisplay screen. Any thoughts on this would also be appreciated.

Thanks in advance for all input!

Taryn


>From: "Tony Zuccolotto" <anthony.zuccolotto at pstnet.com>
>To: <eprime at mail.talkbank.org>
>Subject: RE: feedback on accuracy
>Date: Mon, 21 Jul 2003 12:15:15 -0400
>
>Taryn,
>
>Can you provide some more detail on what you want this accuracy feedback
>to look like in your specific case?   It would also be helpful to have
>an idea of how you intend to analyze these results (i.e. to verify that
>you have everything critical logged in a manner which makes it easy to
>use other tools in or out of E-Prime to perform your analysis).
>
>As you have described your setup, I would expect that each dependent
>measure is being logged as different variables within a single trial
>(which may or may not be what you want).
>
>As per the feedback, it sounds like you will have to do something out of
>the ordinary, but it's definitely do-able with a little script.   An
>depending on what you need you may get by with just setting few
>attributes and then using the feedback object to display the
>information.  If you describe the type of feedback you want under
>whatever scenarios you can think of and someone on the list will likely
>be able to respond with more a more specific approach.
>
>Regards,
>Tony
>
>*** DISCLAIMER: ALL VIEWS EXPRESSED ARE MY OWN AND DO NOT NECESSARILY
>REFLECT THOSE OF PSYCHOLOGY SOFTWARE TOOLS ***
>Anthony P. Zuccolotto
>Vice President
>Psychology Software Tools, Inc.
>2050 Ardmore Boulevard
>Suite 200
>Pittsburgh, PA 15221-4610
>Phone     412-271-5040
>FAX       412-271-7077
>Email     anthony.zuccolotto at pstnet.com
>Internet  http://www.pstnet.com
>
> > -----Original Message-----
> > From: Ben Staab [mailto:staab at Waisman.Wisc.Edu]
> > Sent: Monday, July 21, 2003 10:58 AM
> > To: Taryn Tacosa; eprime at mail.talkbank.org
> > Subject: Re: feedback on accuracy
> >
> > If you just need to give a mean of their accuracy, you should be able
>to
> > use @ACC.MEAN on your feedback display object.  If you need more than
>just
> > a mean, take a look at FeedbackDisplay.ACCStats in the E-Basic Help.
>It
> > returns a Summation object that provides different statistical
>functions
> > like standard deviations and such.  Otherwise, if you are trying to
>show a
> > chart of some sort with the accuracy for each letter in the sequence
> > individually, that could be a little trickier.  It seems from some of
>the
> > things that I've tried doing in my own scripts, that E-Prime writes
>the
> > individual trial results to the log file and then throws them out.  So
>you
> > may have to keep track of them yourself by doing something like this:
> >
> > 1. In the user tab of the main script window, declare an integer array
> > with
> > 16 slots, and a counter:
> >          Dim MyArray(1-16) as Integer
> >          Dim TrialsCounter as Integer
> >          TrialsCounter = 1
> >
> > 2. Put an inline script right after each letter is displayed that will
> > insert the accuracy into  the array:
> >          MyArray(TrialsCounter) = TheNameOfYourTextDisplayObject.ACC
> >          TrialsCounter = TrialsCounter+1
> >
> > 3. When all the trials have been completed (so make sure this is
>outside
> > your trialsproc, whatever you may have called it) add one last text
> > display
> > object (pick your own name, but I'm calling it SumText) with an inline
> > script after it.  The script will look something like this:
> >          ' Declare some position and a counter variables
> >          Dim x, y, i as integer
> >          x = 20
> >          y = 20
> >          i = 1
> >
> >          ' print a simple header
> >          SumText.Text x, y, "Accuracy Summary:"
> >
> >          'Start looping thru the results.
> >          'This displays results in two columns
> >          For i = 1 to 16
> >                  if(i = 9) then          ' if we've already displayed
>8
> > items,
> >                          x = x + 50      ' move right to a second
>column,
> >                          y = 40          ' and move back to the top.
> >                  else                    ' Otherwise,
> >                          y = y + 20      ' move the y position down
> >                  end if
> >
> >                  'instead of showing 0 or 1, use english
> >                  if(MyArray(i)=0)then
> >                          SumText.Text x, y, "Trial " & i & ":
>Incorrect"
> >                  else
> >                          SumText.Text x, y, "Trial " & i & ": Correct"
> >          Next i
> >
> > I think that should do it.  Lastly, a few caveats: I haven't tried
>this
> > myself; this is all off the top of my head.  So it's entirely possible
> > I've
> > got syntax and/or logic errors in there.  Also, those x and y
>positions
> > are
> > just guesses on my part, change them as you see fit.
> >
> > I hope that helps and that I didn't completely miss the point of your
> > question.
> >
> > Ben Staab
> > staab at waisman.wisc.edu
> >
> > At 10:14 AM 7/18/2003 -0500, Taryn Tacosa wrote:
> > >Hello,
> > >
> > >I am having trouble finding a way to present accuracy feedback on a
> > >sequence of observations. Specifically, a subject is shown a sequence
>of
> > >letters, one at a time, and responds YES or NO to each letter
>(typical
> > >n-back task). I would like a screen to give the subject feedback on
>their
> > >accuracy at the
> > >end of each sequence, but am having trouble with this. So far, I can
>only
> > >get accuracy for the last display screen ONLY to appear, rather than
> > >accuracy across all 16 observations (letters).
> > >
> > >Any help would be greatly appreciated!
> > >
> > >Taryn Tacosa
> > >
> > >>RA
> > >>UNC-CH
> > >
> > >_________________________________________________________________
> > >Help STOP SPAM with the new MSN 8 and get 2 months FREE*
> > >http://join.msn.com/?page=features/junkmail
> > >
> >
>
>

_________________________________________________________________
MSN 8 helps eliminate e-mail viruses. Get 2 months FREE*.
http://join.msn.com/?page=features/virus

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/eprime/attachments/20030723/e662dbff/attachment.htm>


More information about the Eprime mailing list