.-
help for ^aovsum^                                                 (STB-54: sg132)
.-

One-way analysis of variance from summary statistics
----------------------------------------------------

   ^aovsum, n(^nlist^) m^ean^(^mlist^) { sd(^sdlist^)^|^se(^selist^)^ }
           [ ^k^eep ^nam^es^(^yname [grpname [freqname]]^)^  onewayopts ]

   ^aovsum^ ^?^


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

^aovsum^ performs a one-way analysis of variance from the size (^n^), mean (^M^), and
standard deviation [or, standard error] (^s^) of each of a series of independent
groups of observations.  (Data are often summarized by those quantities in aca-
demic and scientific writing.)  This is done by synthesizing a data set having
the correct ^n^, ^M^, and ^s^ within each group.  The synthetic data can be retained
and used to perform a variety of other analyses that, like the analysis of var-
iance, depend on the data only through ^n^, ^M^, and ^s^.

In the first syntax, n() and mean() are required; exactly one of sd() or se()
also must be given.  The three argument lists must contain the same number of
values.  The second syntax produces a terse reminder of the first syntax.


Options
-------

^n(^nlist^)^ provides a list of positive integers, the sample sizes (^n^'s) for the
    groups or factor levels, one per group.

^mean(^mlist^)^ provides a list of group means (^M^'s), one per group, in the same
    order as given in ^n()^.

^sd(^sdlist^)^ provides a list of group standard deviations (^s^'s), one per group,
    in the same order as given in ^n()^.

^se(^selist^)^ is an alternative to ^sd()^, whose elements consist of group standard
    errors, ^s^/sqrt(^n^).

^onewayopts^ is any collection of options for the @oneway@ command; ^aovsum^ invokes
    ^oneway^ to perform a one-way ANOVA on the synthetic data.

^keep^ retains the synthetic data created from ^n^, ^M^, and ^s^.  The data consist of
    a response variable, a group variable (factor), and a frequency count that
    serves as an @fweight@. Their default names are ^y_^, ^cond_^, and ^freq_^, respec-
    tively; the data are placed in observations 1..2K where K is the number of
    groups (factor levels).  By default, those variables are deleted at exit.

^names(^namelist^)^ provides alternative names for the synthetic data. Its argument
    is a list of (up to) three variable names to be used in place of ^y_^, ^cond_^,
    and ^freq_^, in that order.


Examples
--------

 . ^aovsum, n(23 12 14 19) m(11.5 23.0 29.8 38.4) sd(2.11 4.07 3.43 4.62)^
   (One-way ANOVA with four groups)

 . ^aovsum, n(23 12 14 19) m(11.5 23.0 29.8 38.4) sd(2.11 4.07 3.43 4.62)^
           ^keep name(wloss dose)^
   (Same as above, but retain the synthetic data as observations 1..8 of
   new variables named wloss, dose, and freq_)

 . ^anova wloss dose [fw=freq_]^
   (Duplicate last ANOVA using @anova@ rather than @oneway@)

 . ^gen byte Dose = dose^        (duplicate the factor)
 . ^anova wloss dose Dose [fw=freq_], cont(dose) seq^
   (Test for linear trend in dose, and for deviations from linear trend)


Author
------
   John R. Gleason, Syracuse University, Syracuse NY, USA
   (loesljrg@@accucom.net)    [This is Version 1.1.1 (27Sep1999).]


Also see
--------

 Manual:   ^[U] 14.1.6 weight^
           ^[R] anova^
           ^[R] oneway^
 On-line:  help for @anova@, @fweight@, @oneway@