Date:         Fri, 2 Jul 2004 08:49:47 +0200
Reply-To:     Marta García-Granero
              <biostatistics@terra.es>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Marta García-Granero
              <biostatistics@terra.es>
Organization: Asesoría Bioestadística
Subject:      Re: LOGISTIC REGRESSION: VARIABLES=...BY [var]
In-Reply-To:  <5.1.0.14.2.20040701145307.00a4bde0@pop.mindspring.com>
Content-Type: text/plain; charset=ISO-8859-1

Hi Richard,

RR>  From the syntax manual entry for LOGISTIC REGRESSION, syntax template:

RR> LOGISTIC REGRESSION [VARIABLES =] dependent var
RR>     [WITH independent varlist [BY var [BY var] ... ]]

RR> "· You can indicate an interaction term on the variable list by using the
RR> keyword BY to separate the individual variables."

RR> I don't quite understand this. If you use "BY var", does that mean
RR> interaction terms between "var" and all preceding variables, or what? Or
RR> can specific interaction variables be included, or excluded, on VARIABLES
RR> or METHOD subcommands?

THe syntax defines specific interaction terms. For example:

LOGISTIC REGRESSION VAR=kyphosis WITH age starvert age BY starvert
  /CRITERIA = PIN(.05) POUT(.10) ITERATE(20) CUT(.5) .

This syntax models the dependent variable Kyphosis as a function of:

- Age
- Starvert
- The interaction between the two above

LOGISTIC REGRESSION VAR=kyphosis WITH age numvert starvert age BY starvert
  /CRITERIA = PIN(.05) POUT(.10) ITERATE(20) CUT(.5) .

This syntax models the dependent variable Kyphosis as a function of:

- Age
- Numvert
- Starvert
- The interaction between age & starvert (numvert is not involved)

I always use "*" to indicate interaction terms, it looks clearer to me:

LOGISTIC REGRESSION VAR=kyphosis WITH age numvert starvert age*starvert
  /CRITERIA = PIN(.05) POUT(.10) ITERATE(20) CUT(.5) .

--
HTH,
 Marta                            mailto:biostatistics@terra.es