Re: st: Overlaying several regression lines from level-2 units in a single graph

[Nick Cox <njcoxstata@gmail.com>]

I'd get Stata to do the fitting outside -graph-. For example,

webuse grunfeld, clear
gen ln_invest = ln(invest)
gen predict = .

qui forval i = 1/10 {
	regress ln_invest year if company == `i'
	predict work
	replace predict = work if company == `i'
	drop work
}

xtline predict, overlay

Nick

On Tue, May 22, 2012 at 12:06 PM, Love Hansson <love.hansson@gmail.com> wrote:

> I am currently working with a 2-level dataset containing level-1
> dependent and independent variables and level-2 independent variables,
> connected with a level-2 group variable. I have found that one way of
> visualizing the difference between level-2 units is to plot a
> regression line for the level-1 independent and dependent variables in
> the following fashion (similar to a Trellis plot):
>
> graph twoway (lfit depvar indepvar), by(grpvar)
>
> Obviously, this results in a number of graphs being plotted next to
> each other. However, I would prefer it if *all* of the regression
> lines could be plotted in an overlaying fashion into the same graph.
> Is there any convenient way of doing this? One way of solving the
> problem might be to do a "graph twoway" and then add as many "lfit"
> parts as there are level-2 units, but since there are fairly many
> level-2 units in my dataset I would prefer it if I could find a more
> convenient solution.
*
*   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/