Re: st: Connecting Data Points with Lines

[Austin Nichols <austinnichols@gmail.com>]

Jielu Lin <jielu.lin@case.edu> :
Can you
tsset id age
before graphing?

Can you show an example of the problem using publicly available data?

clear all
webuse grunfeld
qui xtmixed mvalue time||company:time, mle cov(un)
predict y_fitted, fitted
g c=inrange(company,1,2)
sort c
line y_fitted time, connect(L) name(wrong)
tsset
line y_fitted time, connect(L)

On Fri, Apr 30, 2010 at 1:44 PM, Jielu Lin <jielu.lin@case.edu> wrote:
> Dear all,
>
> I'm modeling age-graded trajectories. Each individual case in my data
> has 4 repeated observations. Then I tried to graph estimated
> individual trajectories. Each individual should have a separate
> trajectory which is not connected to another individual trajectory. In
> the actual graph, within a person, the 4 estimated data points were
> correctly connected. However, some trajectories were connected even
> though these observations did not belong to the same person. This
> happened to some cases, but not all of them. I've been for patterns of
> these odd connections but so far it seems random.
>
> I used the following codes:
>
> xtmixed y age ||id: age, mle cov(un)
> predict y_fitted, fitted
> twoway line y_fitted age, connect(ascending)
>
*
*   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/