Being new to the list should increase your
willingness to find out proper procedure. Your
disregard of that may help explain the lack of
answers to your question.
The following are extracts from the FAQ:
"Please do not send a message starting a new thread by replying to
someone else's message on a different topic. Even if you edit out all
the previous content, archiving software will still pick up the fact
that your message is a reply and that will unnecessarily mess up
documentation of mail threads in the list archives."
"Choose an informative subject line for your message."
"Say what command(s) you are using. If they are not part of official
Stata, say where they come from: the STB/SJ, SSC, or other archives."
"Refer to Stata commands and other syntax within ordinary text using
dashes (- -) to flag words that you would type when using Stata.
Example: You could try the -assert- command, which is very useful for
this kind of problem."
Now about your question:
-baplot- is a user-written command (Paul Seed) published in STB-55.
A more up-to-date command in the same territory is -concord-,
last publicly revised in SJ 5(3). -findit concord- points to
downloadable files. The on-line help for -concord- says more
about the Pitman test than does the on-line help for -baplot-.
At this moment, I don't have access to the original STB-55
article by Paul Seed. I am assuming that the Pitman statistic you
refer to is that also calculated by -concord-,
which is the correlation between difference and mean. In one
interpretation this is a test statistic for a null hypothesis of equal
variances given bivariate normality (Pitman 1939: see Snedecor and
Cochran 1989, pp.192-3). Alternatively, it is an exploratory diagnostic.
The original Pitman reference
Pitman, E.J.G. 1939. A note on normal correlation. Biometrika 31: 9-12.
Snedecor, G.W. and W.G. Cochran. 1989. Statistical Methods. Ames: Iowa
State University Press.
In broad terms, rejecting the null at conventional levels is
indeed a signal of genuine difference, but looking at the graph and a
check for non-normality are both prudent.
Nick
n.j.cox@durham.ac.uk
Dr. Todd Garlie
I am new to the list and realize this question may interrupt the thread.
Can anyone explain the interpretation of the Pitman's test of
difference in variance as part of the baplot output. If the r has a
statistically significant p-value does that
indicate that the methods do not agree with each other?