Bookmark and Share

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


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

st: RE: linear and non-linear regression


From   Nick Cox <n.j.cox@durham.ac.uk>
To   "'statalist@hsphsun2.harvard.edu'" <statalist@hsphsun2.harvard.edu>
Subject   st: RE: linear and non-linear regression
Date   Thu, 14 Jun 2012 18:13:18 +0100

Stata uses whatever you tell it to use. If you are talking about -nl- then it is documented that it uses least squares. The first line of the help file of -nl- is "nonlinear least-squares estimation". 

It is good practice to look at residuals, but for most purposes normality of error terms and even their homoscedasticity is less important than other assumptions, most of all whether the condition mean function is indeed abut right for the data. Sometimes it is obvious in advance that assumptions will be violated, but quantitative checking of residuals does require the model to be fitted first.  

Most of this falls into the territory that it is well documented if not in the manuals, then in basic texts. 

Nick 
n.j.cox@durham.ac.uk 

tashi lama

     The key to fitting a line to a linear looking scatter plot is OLS, abbreviation for Ordinary Least Square. Does Stata use Ordinary Least Square technique to fit non linear line to a non linear looking scatter plot? Does anyone have any clue? Is it necessary that we check whether the residuals are normally distributed, the distribution is homoskedastic etc etc before running any types of regression?

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index