Re: st: Regression Analysis - dependent and independent variables

[David Hoaglin <dchoaglin@gmail.com>]

Dear Lil,

If you have no problems in your data, you can fit those two models.

Since you are considering both X1 and Y1 as dependent variables, you
may also want to consider the partial correlation between X1 and Y1,
adjusting for the contributions of the other variables.  You can
obtain the partial correlation by regressing X1 on those other
variables (not including Y1) and saving the residuals, regressing Y1
on those other variables (not including X1) and saving those
residuals, and calculating the usual correlation between the X1
residuals and the Y1 residuals.

If it is appropriate for your data, fitting the two regression models
is a straightforward application of regression analysis, not a special
method.  A variety of textbooks discuss the theory of simple linear
regression when the two variables (say X and Y) have a bivariate
normal distribution.  In that setting, Y has a linear regression on X,
X has a linear regression on Y, and the two regressions are generally
different.

David Hoaglin

On Thu, Mar 13, 2014 at 5:12 AM, liliana <lilly-90@hotmail.it> wrote:
> Hello,
>
> I am looking for some papers or lectures on the regression analysis and on
> the possibility of regressing two different models (1 and 2), where in the
> first model the dependent variable is Y1 and one of the independent
> variables is X1. In the second model, instead, one of the independent
> variables is Y1 (the dependent variable in model 1) and the dependent
> variable is X1 (one of the independent variables in model 1).
> Do you think it is possible to use this method?
>
> Can you suggest me any ref? Thank you in advance.
>
> Best,
> Lil
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/