On 21 Aug 2003 12:17:23 -0700, [EMAIL PROTECTED] (Niko
Tiliopoulos) wrote:

> Hopefully my question will be easy for some of you to answer.
> 
> In a logistic regression model (SPSS 11) I have the interaction term
> between the following variables:
> 
> Gender (0 = Males, 1 = Females). 0 is treaded as the indicator code.
> 
> and
> 
> Competence (measured in a continuous scale, with higher numbers
> suggesting higher levels of competence)
> 
> The DV is agreement with something (0 = disagree, 1 = agree). 
> 
> The logistic regression results show an odds ratio (expBeta) of 2 (I
> simplify it for clarity).

There is an ambiguity in how we use certain terms,
especially when it comes to log-linear models.

You might be predicting  'competence' from
Gender, and that is what you mean by Interaction...
or you might have a third variable that you have not named,
so that there is an equation with two Main effects and
the interaction.  In that case,  you would need to look
at all three regression coefficients in order to see what
the prediction is.

On the other hand --
You will get a pretty good notion of what is going on
if you look at the raw tabulation, which is 2x2  or 2x2x2.


-- 
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
"Taxes are the price we pay for civilization."  Justice Holmes.
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to