Signal detection theory

[Paul]

I definitely see your point. But I think that the distinction between where
the positive responses are being made has no bearing here. The way I
understand it, C is simply the propensity to offer a 'yes' response. So by
definition, if either false alarm or hit rate increases, C must increase,
since in either case the perceiver is offering more 'yes' responses.  This
is irregardless of the perceptual situation it is offered in. The fact that
those 'yes' responses turn out to be 'hits' and not 'false alarms' is
captured by the differences in dprime across the two conditions- which, in
fact, tells you that Condition A has a higher sensitivity than Condition B.

Paul




On Mon, May 3, 2010 at 11:04 AM, Tobias <tobias.fw at gmail.com> wrote:

> They don not only have higher hit rates but also the same false alarm
> rate. So they are not more often saying "yes" in blank trials but only
> in target trials.
> If they really had a bias shouldn't they also have a higher false
> alarm rate?
>
> On 21 Apr., 01:55, Robert Ariel <rar... at kent.edu> wrote:
> > Well, if you buy the assumptions of SDT you can.  Remember, SDT makes
> > assumption about behavior.  Specifically that decisions are made by
> applying
> > a decision criterion to the evidence extracted from each trial.   So, you
> > could conclude that one condition has a higher hit rate because that
> > condition has a larger bias toward saying yes in your experiment.  I
> guess
> > the question is, does it make theoretical sense to do so?
> >
> >
> >
> > On Mon, Apr 19, 2010 at 7:33 AM, Tobias <tobias... at gmail.com> wrote:
> > > Thanks Robert,
> >
> > > if I am not getting you wrong, this means that C is independent of d'
> > > but not of the hit rate.
> > > The question occurs to me if you can really say that one condition is
> > > more liberal if they are just better obviously.
> >
> > > Cheers,
> > > Tobias
> >
> > > On 16 Apr., 21:25, Robert Ariel <rar... at kent.edu> wrote:
> > > > Tobias,
> >
> > > > Computationally, C is the average of the your transformed hit and
> false
> > > > alarm rates.  You can see this in the equation you presented.  So, no
> > > doubt
> > > > if you have equal false alarm rates across conditions, differences in
> C
> > > are
> > > > resulting because of differences in hit rates.
> >
> > > > Basically with equal false alarm rates, the condition with a higher
> hit
> > > rate
> > > > will always be more liberal.  If hit rates are equal, the condition
> with
> > > > higher false alarm rate will be more liberal.
> >
> > > > Best,
> >
> > > > Robert
> >
> > > > On Fri, Apr 16, 2010 at 9:59 AM, Tobias <tobias... at gmail.com> wrote:
> > > > > Hi together,
> >
> > > > > this might be a bit off topic but as you are all very much into
> > > > > psychological experimental science you might be of great help for
> this
> > > > > issue. Besides, my topic is the outcome of an E-Prime experiment ;)
> >
> > > > > It is about the response bias in signal detection theory (SDT).
> I've
> > > > > heard that C is usually better than Beta as a measure of response
> bias
> > > > > as it is indpendent of d'. Now what I have in my experiment is a
> very
> > > > > high hit rate for condition A and a lower hit rate for condition B.
> > > > > False alarm rates are however the same for A and B. So what I get
> > > > > using the formula for C (C = -0.5*(z(false alarms) + z(hits)) is a
> > > > > liberal criterion C for A and a less liberal criterion for B.
> >
> > > > > So can I actually say that A is more liberal? Apparently this is
> only
> > > > > due to the fact that the hit rate is higher. I am quite puzzled by
> > > > > this... glad for any help!
> >
> > > > > Tobias
> >
> > > > > --
> > > > > You received this message because you are subscribed to the Google
> > > Groups
> > > > > "E-Prime" group.
> > > > > To post to this group, send email to e-prime at googlegroups.com.
> > > > > To unsubscribe from this group, send email to
> > > > > e-prime+unsubscribe at googlegroups.com<e-prime%2Bunsubscribe at googlegroups.com>
> <e-prime%2Bunsubscribe at googlegroups.com<e-prime%252Bunsubscribe at googlegroups.com>
> >
> > > <e-prime%2Bunsubscribe at googlegroups.com<e-prime%252Bunsubscribe at googlegroups.com>
> <e-prime%252Bunsubscribe at googlegroups.com<e-prime%25252Bunsubscribe at googlegroups.com>
> >
> >
> > > > > .
> > > > > For more options, visit this group at
> > > > >http://groups.google.com/group/e-prime?hl=en.
> >
> > > > --
> > > > You received this message because you are subscribed to the Google
> Groups
> > > "E-Prime" group.
> > > > To post to this group, send email to e-prime at googlegroups.com.
> > > > To unsubscribe from this group, send email to
> > > e-prime+unsubscribe at googlegroups.com<e-prime%2Bunsubscribe at googlegroups.com>
> <e-prime%2Bunsubscribe at googlegroups.com<e-prime%252Bunsubscribe at googlegroups.com>
> >
> > > .
> > > > For more options, visit this group athttp://
> > > groups.google.com/group/e-prime?hl=en.
> >
> > > --
> > > You received this message because you are subscribed to the Google
> Groups
> > > "E-Prime" group.
> > > To post to this group, send email to e-prime at googlegroups.com.
> > > To unsubscribe from this group, send email to
> > > e-prime+unsubscribe at googlegroups.com<e-prime%2Bunsubscribe at googlegroups.com>
> <e-prime%2Bunsubscribe at googlegroups.com<e-prime%252Bunsubscribe at googlegroups.com>
> >
> > > .
> > > For more options, visit this group at
> > >http://groups.google.com/group/e-prime?hl=en.
> >
> > --
> > You received this message because you are subscribed to the Google Groups
> "E-Prime" group.
> > To post to this group, send email to e-prime at googlegroups.com.
> > To unsubscribe from this group, send email to
> e-prime+unsubscribe at googlegroups.com<e-prime%2Bunsubscribe at googlegroups.com>
> .
> > For more options, visit this group athttp://
> groups.google.com/group/e-prime?hl=en.
>
> --
> You received this message because you are subscribed to the Google Groups
> "E-Prime" group.
> To post to this group, send email to e-prime at googlegroups.com.
> To unsubscribe from this group, send email to
> e-prime+unsubscribe at googlegroups.com<e-prime%2Bunsubscribe at googlegroups.com>
> .
> For more options, visit this group at
> http://groups.google.com/group/e-prime?hl=en.
>
>

-- 
You received this message because you are subscribed to the Google Groups "E-Prime" group.
To post to this group, send email to e-prime at googlegroups.com.
To unsubscribe from this group, send email to e-prime+unsubscribe at googlegroups.com.
For more options, visit this group at http://groups.google.com/group/e-prime?hl=en.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/eprime/attachments/20100503/f7296c62/attachment.htm>