d-prime

[Martijn]

Does anybody happen to know how I can (easily) calculate d-prime in SPSS?

Regards,

Martijn

[Bruce Weaver]

I'm a little rusty on signal detection theory, but I think d' = z(hit) -
z(false alarm), right? So something like this should work:

COMPUTE #zhit = IDF.NORMAL(hit,0,1) .
COMPUTE #zfa = IDF.NORMAL(fa,0,1) .
compute dprime = #zhit - #zfa.
exe.

It assumes variables HIT and FA that give hits and false alarms as
proportions.

--
Bruce Weaver
bwe...@lakeheadu.ca
www.angelfire.com/wv/bwhomedir

[Art Kendall]

In signal detection theory it is a z found by taking the difference
between the z associated with the hit rate as a proportion and the z
associated with the false alarm rate as a proportion.

data list list/id (f3) hitrate(f7.3) falsealarmrate(f7.3).
begin data
1 .50 .05
2 .70 .08
3 .45 .033
4 .51 .01
5 .96 .01
6 .06 .01
end data.
compute dprime = idf.normal(hitrate,0,1) -idf.normal(falsealarmrate,0,1).
formats dprime (f7.2).
list.

Art Kendall
Social Research Consultants

[Martijn]

Hi Art & Bruce,

Thanks for you help!

Martijn

"Art Kendall" <Arthur....@verizon.net> wrote in message
news:45DC9DEC...@verizon.net...

[vna...@uic.edu]

Thank you for asking that question, still helpful 7 years later

[Rich Ulrich]

On Sun, 23 Nov 2014 17:24:02 -0800 (PST), vna...@uic.edu wrote:

>Thank you for asking that question, still helpful 7 years later

A helpful question? - My ISP shows me the original, but I
don't see an answer.

If you want to ask a question, it is far better to ASK it than
to assume that everyone sees it. Or is curious enough to
look for it.

"Does anybody happen to know how I can (easily) calculate d-prime in
SPSS?"

I don't know. Have you looked?

--
Rich Ulrich
(no more responses from me until after Thankgiving some time.)

[David Marso]

Rich,
Looks like both Art K and Bruce W opined on this subject back in 2007 ;-)
https://groups.google.com/forum/#!topic/comp.soft-sys.stat.spss/4cJW8FeFA1E
--