Re: st: linear and quadratic trend analysis

[David Jacobs <jacobs.184@sociology.osu.edu>]

I'm not sure that Martin answered your question although you probably 
already know what I'm about to say.

To control for a linear trend simply construct a counter explanatory 
variable running from 1 to how many years you have; to test a 
quadratic, add the square of this explanatory variable and so on for a cubic.

As analysis of variance allows continuous explanatory variables, this 
can be done in anova although most people probably would use 
regress.  If you use a quadratic or cubic it would be a good idea to 
test the relevant terms for joint significance.  That will help you 
decide if the higher order term is necessary.

Dave Jacobs

At 03:49 PM 3/4/2008, you wrote:

> Hi there,
> I am conducting a study using the National AddHealth data 
> set.  After running factorial anovas, it is clear that I need to 
> follow up with a trend analysis.  I figured out how to run graphs, 
> but does anyone have information on running the statistical tests 
> for trends.  The study has four continuous outcome variables, plus 
> sex, age, and grade level with one  categorical independent 
> variable.  Any help is much appreciated.
> Shelly
>
> Shelly Mahon
> Ph.D. Candidate
> University of Wisconsin-Madison
> Human Development and Family Studies
> 1430 Linden Drive
> Madison, WI 53706-1575
> email: mdmahon@wisc.edu
>
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/