Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: combined linear and quadratic trend in marginsplot

From   Lloyd Dumont <[email protected]>
To   "[email protected]" <[email protected]>
Subject   st: combined linear and quadratic trend in marginsplot
Date   Fri, 30 Sep 2011 06:06:17 -0700 (PDT)

Hello, everyone.  I am trying to do something very simple with �rgins-, but I�stumbling� 
I am running a logit that includes a linear trend (starts at 0 and goes up to 5), a quadratic trend (starts at 0 and goes up to 25, obviously), and a bunch of other variables, some of which are continuous, some of which are binary, some factor variables, some two-way interactions, etc.
But, at least initially, I just want to generate a –marginsplot- that holds all the variables at their sample means EXCEPT the linear and the quadratic trend.
Say the logit looks like…
- logit DEPVAR LINTREND QUADTREND {and a bunch of other indep vars}
-margins, at(LINTREND=(0(1)5))
-marginsplot, xlabel(0(1)5)
This, of course, generates a straight line.  But, what I really want is the curvilinear function that is captured by the linear and the quadratic term together.
How do I do that?
And, am I right to assume that all of the other independent variables—are being held at their sample means?  If I wanted to hold one or more at a specific value, how would I do that?
Thank you very, very much for your help.  Lloyd Dumont
PS~It happens to be complex survey data, but I don't think that makes any difference here.

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index