R help - Jun 2002 - Psychometric curves, two altnerative force choice, glm, and budbworms

Dear R-Listers,

to measure the psychometric curve of pitch discrimination, one sequentially
presents two tones of slightly different pitch to an observer (animal will
do), and asks "which is higher". The pschometric curve is the fraction
of
correct responses plotted against the pitch difference. It passes through
50% (pure guessing) at zero and normally approaches 100% at large
difference.
To compare two psychometric curves, the conventional way is to fit two
logistic curves and compare the 75% correct "threshold" values
(whatever
threshold means).

I want to handle the case similar to the budworm example in MASS
(glm(SF~sex*ldose, family=binomial)). My basic idea is that the 2AFC forced
choice psychometric curve, normally only defined for positive stimuls
differences, could conceptually be continued to negative values by mirroring
the values at (0,0.5) to get the whole binomial/logistic curve. As far as I
can see, the result would be correct if we divide the counts for a stimulus
difference equally between positive and negative branch, with correct/wrong
inverted for the negatives. The solution is not clean if we have uneven
counts.

Is this a valid approach? Alternatively, could I force a
glm/binomial/logistic fit to go through 0/0.5?

Dieter Menne

---------------------------------------
Dr. Dieter Menne
Biomed Software
72074 T?bingen
Tel (49) (7071) 52176
FAX (49) (7071) 55 10 46
dieter.menne at menne-biomed.de
www.menne-biomed.de



-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

"Dieter Menne" <dieter.menne at menne-biomed.de> writes:
> Dear R-Listers,
> 
> to measure the psychometric curve of pitch discrimination, one sequentially
> presents two tones of slightly different pitch to an observer (animal will
> do), and asks "which is higher". The pschometric curve is the
fraction of
> correct responses plotted against the pitch difference. It passes through
> 50% (pure guessing) at zero and normally approaches 100% at large
> difference.
> To compare two psychometric curves, the conventional way is to fit two
> logistic curves and compare the 75% correct "threshold" values
(whatever
> threshold means).
> 
> I want to handle the case similar to the budworm example in MASS
> (glm(SF~sex*ldose, family=binomial)). My basic idea is that the 2AFC forced
> choice psychometric curve, normally only defined for positive stimuls
> differences, could conceptually be continued to negative values by
mirroring
> the values at (0,0.5) to get the whole binomial/logistic curve. As far as I
> can see, the result would be correct if we divide the counts for a stimulus
> difference equally between positive and negative branch, with correct/wrong
> inverted for the negatives. The solution is not clean if we have uneven
> counts.
> 
> Is this a valid approach? Alternatively, could I force a
> glm/binomial/logistic fit to go through 0/0.5?
Sure, just remove the intercept and change the response definition
from "correct" to "chooses A" (i.e. "correct" if
delta is positive,
"wrong" if it is negative). glm(chooseA~delta-1, binomial)

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._