Subject: | |
From: | |
Reply To: | |
Date: | Tue, 10 Oct 2006 17:16:29 +0800 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
It seems that there is no handy statistical package to estimate the path model with the discrete dependent variable. I examined a path model in such case and compared it with the binary logisitic model. The results are very similar.
Chia-hung
※ 引述《">Shanna Pearson-Merkowitz" <[log in to unmask]>》之郵件內容:
>I am working on a paper in which the fundamental argument can only
>really be expressed through a path model. Unfortunately, one of my
>endogenous
>variables is dichotomous. While the ultimate dependent variable is
>interval, I know that having a non-linear dependent variable in a path
>model is less than ideal. (I am using Amos which uses FIML and while
>not
>OLS, FIML still assumes a linear relationship.) Does anyone know of a
>way
>of taking care of the likely non-linearity that comes with a
>dichotomous
>dependent variable? If not, does anyone have any suggestions about what
>is the best thing
>to do (for example, should I just give up or should I just estimate it
>as a linear model and
>acknowledge the problems?)?
>
>I greatly appreciate any and all suggestions. Thanks so much.
>
>Shanna Peason-Merkowitz
>University of Maryland
>Department of Government and Politics
>
>**********************************************************
> Political Methodology E-Mail List
> Editor: Karen Long Jusko <[log in to unmask]>
>**********************************************************
> Send messages to [log in to unmask]
> To join the list, cancel your subscription, or modify
> your subscription settings visit:
>
> http://polmeth.wustl.edu/polmeth.php
>
>**********************************************************
>
**********************************************************
Political Methodology E-Mail List
Editor: Karen Long Jusko <[log in to unmask]>
**********************************************************
Send messages to [log in to unmask]
To join the list, cancel your subscription, or modify
your subscription settings visit:
http://polmeth.wustl.edu/polmeth.php
**********************************************************
|
|
|