Automatic Interaction Detection???

Hi:

I have a set of data (about 10,000 cases) with several dependent variables (DVs). One DV is nominal, one is ordinal, and one is interval. I'd like to analyze each DV separately.

My independent variables (IVs) are a bunch of demographic variables. The IVs represent all scale types. I suspect I'll have to collapse my metric data into categories.

I'd like to use a statistical technique that will examine all possible combinations of the demographic variables and then reveal those cells that have the lowest scores on each DV and the highest scores on each DV. What statistical technique(s) can I use?

I vaguely recall a technique called Automatic Interaction Detection that would do what I want, but I don't know if SPSS has that technique or if it can handle my types of DVs.

Your suggestions, please.

Curt

[ViAnn Beadle]

The TREE procedure provides CHAID (Chi-square Automatic Interaction Detection) and C&RT (Regression Tree) algorithms. Both techniques will automatically collapse interval level predictors into categories to maximize/minimize the DV. C@RT provides binary splits, while CHAID can produce more than two splits for any node. The TREE procedure is not part of the base and I'm not sure which option it is in but I'm sure somebody on this list will know.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of [hidden email]
Sent: Saturday, September 22, 2007 6:31 PM
To: [hidden email]
Subject: Automatic Interaction Detection???

Hi:

I have a set of data (about 10,000 cases) with several dependent variables (DVs). One DV is nominal, one is ordinal, and one is interval. I'd like to analyze each DV separately.

My independent variables (IVs) are a bunch of demographic variables. The IVs represent all scale types. I suspect I'll have to collapse my metric data into categories.

I'd like to use a statistical technique that will examine all possible combinations of the demographic variables and then reveal those cells that have the lowest scores on each DV and the highest scores on each DV. What statistical technique(s) can I use?


I vaguely recall a technique called Automatic Interaction Detection that would do what I want, but I don't know if SPSS has that technique or if it can handle my types of DVs.

Your suggestions, please.

Curt

[zstatman]

These models are available in SPSS via the Tree (Classification) module and
also in AnswerTree



-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
ViAnn Beadle
Sent: Saturday, September 22, 2007 9:53 PM
To: [hidden email]
Subject: Re: Automatic Interaction Detection???

The TREE procedure provides CHAID (Chi-square Automatic Interaction
Detection) and C&RT (Regression Tree) algorithms. Both techniques will
automatically collapse interval level predictors into categories to
maximize/minimize the DV. C@RT provides binary splits, while CHAID can
produce more than two splits for any node. The TREE procedure is not part of
the base and I'm not sure which option it is in but I'm sure somebody on
this list will know.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
[hidden email]
Sent: Saturday, September 22, 2007 6:31 PM
To: [hidden email]
Subject: Automatic Interaction Detection???

Hi:

I have a set of data (about 10,000 cases) with several dependent variables
(DVs). One DV is nominal, one is ordinal, and one is interval. I'd like to
analyze each DV separately.

My independent variables (IVs) are a bunch of demographic variables. The IVs
represent all scale types. I suspect I'll have to collapse my metric data
into categories.

I'd like to use a statistical technique that will examine all possible
combinations of the demographic variables and then reveal those cells that
have the lowest scores on each DV and the highest scores on each DV. What
statistical technique(s) can I use?


I vaguely recall a technique called Automatic Interaction Detection that
would do what I want, but I don't know if SPSS has that technique or if it
can handle my types of DVs.

Your suggestions, please.

Curt