[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: multivariable median regression
I apologise for the format of my last email.
For those who couldn't read it, this is my problem:
Thank you Joe,
Actually, I use the bsqreg. I couldn't find any reference to which test I can use to compare nested models (for ex. the full model with a model without a variable). I can't apply a loglikelihood ratio test, since there's no likelihood estimation. I was wondering about an F test, based on the median regression's pseudo R2, but I couldn't work it out.
So, any help is greatly appreciated.
De: Jhilbe@aol.com [mailto:Jhilbe@aol.com]
Enviado el: vie 22/04/2005 23:19
Asunto: Re: st: multivariable median regression
You asked about how to run a multivariate median regression. Isn't this the
same as Stata's qreg command. If you type
. search median regression
at the Stata prompt, a lists of references appears. One reference is for a
Frequently Asked Questions source related to qreg. Here is the first two
paragraphs from the FAQ. If this is what you're looking for, then qreg is your
"Stata's qreg command fits quantile (including median) regression models,
also known as least-absolute value (LAV) models, minimum absolute deviation
(MAD) models, and L1-norm models.
"Median regression estimates the median of the dependent variable,
conditional on the values of the independent variable. This is similar to
least-squares regression, which estimates the mean of the dependent variable. Said
differently, median regression finds the regression plane that minimizes the sum of
the absolute residuals rather than the sum of the squared residuals."
I have to build a multivariable median regression model but I don't
understand which is the appropriate method to select variables. I can get the pseudo
R2 that is based on the likelihood of a double exponential distribution, but
I don't know how to use this information to decide if a variable has to enter
or not in the equation.
* For searches and help try: