.-
help for ^rmttest^                                               (STB-33: sg49.1)
.-

paired (repeated measures) t-test with scatterplot
--------------------------------------------------

   ^rmttest^ depvar [^if^ exp] [^in^ range] [^,^ ^by(^byvar [val1 val2]^)^
                ^ci^ ^gr^aph ^id(^idvar^)^ ^l^evel^(^cilevel^)^ graph_options] 


Description
-----------

^rmttest^ performs a paired t-test for two samples contained among the 
values of depvar.  The sample pairings are defined by the variable idvar. 
The samples are selected by two values of the variable byvar.  An observation 
will be ignored if it has missing values for any of the byvar, idvar, or 
depvar variables. Subsets of observations can be selected with ^if^ and ^in^ 
clauses, as well as with two values provided in the ^by^ option. In any case, 
the selection must result in observations where the byvar variable assumes 
exactly two values.


Options
-------

^by(byvar)^ specifies a variable that defines the two samples to be compared. 
    Not optional.  What is optional is two byvar values (val1 val2), 
    which request a t-test that compares cases with byvar==val1 vs. those 
    with byvar==val2.

^ci^ displays confidence intervals for the two population means being compared,
    as well as for their difference.  The confidence level is set by ^level^.

^graph^ draws a scatterplot of the paired values of depvar, including a 
    reference line corresponding to ordinate==abscissa.  If ^ci^ is also 
    given, the means and confidence limits from each sample are indicated by 
    tick marks on the top and right axes.

^id(idvar)^ specifies a variable that defines the pairing of the two samples to
    be compared. Not optional.

^level(cilevel)^ sets the confidence level for the confidence intervals 
    requested by ^ci^.  The default cilevel is 95%; values for cilevel may 
    be specified as fractions, level(.975), or as percentages, level(97.5).

graph_options are any of the options allowed with ^graph, twoway^, except for
    connect and, if ^ci^ is given, ttick and rtick.


Examples
--------

 . ^rmttest systolic if group==3 & (week==6|week==12), by(week) id(patient)^
   (paired t-test comparing systolic at week==6 for patients in group==3, with
   the same patients' systolic at week==12)

 . ^rmttest systolic if group==3, by(week 6 12) id(patient)^
   (exactly the same as the previous example)

 . ^rmttest hrate if !trt, by(day 0 7) id(subject) gr ci l(.99) sym([subject])^
   (paired t-test comparing hrate at day==0 and day==7 for all subjects in the
   condition where trt==0, along with 99% confidence intervals. Plot the day==7
   hrate vs. day==0 hrate using the values of the subject variable as the plot
   symbol, with means and confidence limits shown as ticks on the top (day==0)
   and right (day==7) axes, and with a reference line drawn at the position 
   hrate(day==7) == hrate(day==0).


Author
------
	John R. Gleason
        Syracuse University
	73241.717@@compuserve.com


Also see
-------- 

    STB:  STB-33 sg49.1; STB-30 sg49
 Manual:  [R] graph twoway, [R] ttest
On-line:  help for @graph@, @twoway@, @ttest@