Sorry for my tardy replies.  I'm busy prepping GenEd-QL62.  If this
course works out, you will teach it next year and be busy with it.

On Monday you wrote me

You can download and install Rcmdr.HH_1.3.zip on R-2.4.1 from
    http://astro.ocis.temple.edu/~rmh/HH/Rcmdr.HH_1.3.zip

but you don't say how to do the installation from here.  This may
relate to the e-mail you sent last night to Erich.

Next I react to HH-R.package/Rcmdr.HH/cmts/rmh010507b, numbers
referring to bholland.121606

>Confirm that I have 7 correctly interpreted.
Yes.

>Propose menu appearances for numbers 5 and 6.   

What you are proposing with stuff like number of level(ABCD | group1)
is overly complicated for Stat 22.  Please use as a guide the
appearance of the Minitab Student Edition menus for working the same
test.  For example, consider Keller's Xr13-109.xls.  This example has
2=yes and 1=no and wants a one-sided test comparing the proportion of
1's.  In Minitab and click Stat, Basic Statistics, 2 Proportions,
Options: Alternative = greater than.  Minitab's output pays attention
to which is population 1 and which is population 2, by saying "Event =
2", telling the user that it is comparing the proportion of 2's.  This
is crucial info of the sort that Rcmdr didn't give in F06 and was a
problem using Rcmdr for Chapter 13.  Keller's question is "Can we
infer that New Yorkers are more likely to respond no than
Californians?"  Just flip this to "Can we infer that New Yorkers are
less likely to respond yes than Californians?" and Minitab has
answered it.  The goal is to make Rcmdr work as smoothly as Minitab in
tasks both can do.


