Your approach is clearly correct in principle,
but you not surprisingly drew the line at giving
the complete specification of the 22 graphs
superimposed on top of each other needed to
apply your approach to Erik's example dataset.
Fortunately, it can always be done more
easily. With two (non-missing) scores for every student,
sort student time
twoway connected score time, c(L)
is easier to type and the result is much
cleaner.
With a more general structure (panel data,
whether or not named as such), and the
possibility of missing values, or different
times of observation, a simple
-connect(L)- may give rise to spurious connections
between successive panel members, depending on the
precise configuration of data, as has long been
documented.
This can always be avoided, either by declaring
the data to be panel and using -tsline-, or
(as indicated by my earlier mailing in this
thread) by using -linkplot- from SSC.
Although when faced with data on this form
many analysts would think their way
to a graph with time on the horizontal
axis and the changing response on
the vertical axis, for just two times
this usually implies a criss-crossing
pattern of lines which is not always
easy to interpret.
Erik and I had an interesting discussion
off-line about alternatives, centred on
the use of -pairplot- from SSC, following
my earlier posting.
As for cell means in ANOVA, I do not
know what you want to do, but several official
and user-written programs provide post-modelling
graphics after -anova-.
Nick
[email protected]
P.S. -xlabel(1 2)-
Bryan W. Griffin
> I'm interested in something like this too for plotting cell
> means after
> ANOVA. Here is one way to accomplish the task:
>
> Reorganize data such as this (to keep this simple I used data
> from only the
> first 3 of your 22 students):
>
> student score time
> 1 4.5 1
> 1 5.25 2
> 2 3.5 1
> 2 5.25 2
> 3 7.5 1
> 3 7.5 2
>
> Student 1 has scores of 4.5 and 5.25 on occasion 1 and 2,
> student 2 has 3.5
> and 5.25 on time 1 and 2, etc.
>
> . twoway (connected score time if student==1) (connected
> score time if
> student==2) (connected score time if student==3)
>
> This command will give you the graph desired. The only
> problem I have is
> with the x-axis. I want only time 1 and 2 showing on the
> axis, but the
> graph provides points between 1 and 2, such as 1.2, 1.4, 1.6,
> and 1.8 and I
> could not figure how to suppress those numbers. For those
> knowledgeable
> with graphing, how does one suppress these numbers between 1 and 2?
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
