f

#### Analysis for binary DV with between & within subjects IV

```Hello,

I am a PhD candidate in psychology and I have some difficulty analyzing my =
data on decision making in patients with Anxiety Disorders.=20
Participants (patients with Anxiety Disorders and matched controls) had to =
complete 8 trials, in which they had to choose between a safe and a risky o=
ption. 4 trials were presented in terms of gains (two trials with a high ri=
sk and two with a low risk) and 4 were presented in terms of loss (two tria=
ls with a high risk and two with a low risk).=20

The aim was to determine if the formulation, the level of risk (and the int=
eraction of these two factors) affects differently choices of patients with=
Anxiety Disorders vs. control.
In sum:
-	2 within subjects factors (formulation + level of risk)
-	1 between subjects factor (Anxiety-Control)
-	Multiple binary dependent variable

In my domain, almost all papers used ANOVAs to analyze these kind of data (=
even one publication in Science) although the DV was binary.=20
As I have raised statistical comments from my reviewers by using ANOVAs on =
this kind of data set, I would perform more adequate analysis. I was advise=
d to use chi-square or logistic regression (but I am not sure of the soundn=
ess of these techniques because each participant has carried out 8 trials).

Thank you in advance for the help!
```
 0
PhD
12/13/2016 7:11:50 PM
comp.soft-sys.stat.spss 5679 articles. 0 followers.

1 Replies
454 Views

Similar Articles

[PageSpeed] 33

```On Tuesday, December 13, 2016 at 2:11:52 PM UTC-5, PhD student wrote:
> Hello,
>=20
> I am a PhD candidate in psychology and I have some difficulty analyzing m=
y data on decision making in patients with Anxiety Disorders.=20
> Participants (patients with Anxiety Disorders and matched controls) had t=
o complete 8 trials, in which they had to choose between a safe and a risky=
option. 4 trials were presented in terms of gains (two trials with a high =
risk and two with a low risk) and 4 were presented in terms of loss (two tr=
ials with a high risk and two with a low risk).=20
>=20
> The aim was to determine if the formulation, the level of risk (and the i=
nteraction of these two factors) affects differently choices of patients wi=
th Anxiety Disorders vs. control.
> In sum:
>       -	2 within subjects factors (formulation + level of risk)
>       -	1 between subjects factor (Anxiety-Control)
>       -	Multiple binary dependent variable
>=20
> In my domain, almost all papers used ANOVAs to analyze these kind of data=
(even one publication in Science) although the DV was binary.=20
> As I have raised statistical comments from my reviewers by using ANOVAs o=
n this kind of data set, I would perform more adequate analysis. I was advi=
sed to use chi-square or logistic regression (but I am not sure of the soun=
dness of these techniques because each participant has carried out 8 trials=
).
>=20
> Thank you in advance for the help!

You may wish to post your question to the SPSSX-L mailing list--it is far m=
ore active these days than this mailing list. =20

An easy way to join the list is via the Nabble interface, which can be view=
ed here:

http://spssx-discussion.1045642.n5.nabble.com/

Click on the "more options" link at the top of the page for instructions on=
how to join the list.

You could also try searching the archives of that list, using search terms =
such as <repeated measures logistic regression>.  The two main approaches y=
ou'll find are:

1. GENLIN with generalized estimating equations (GEE)
2. Multilevel logistic regression via GENLINMIXED

But please post to SPSSX-L, as many more people will see the question there=
..=20

HTH.
```
 0
Bruce
12/13/2016 9:34:00 PM