From: brian [brian@stat.cmu.edu]
Sent: Monday, April 23, 2007 3:09 PM
To: 'Elise Olson'; 'Nathaniel Anozie'; 'April Galyardt'
Subject: 726 presentation

Thanks for your presentation today, and congrats on having it done.

1. I would like a copy of your presentation slides

2. the following suggestions seemed like they would impact your final paper:

   - check assumptions of t-tests, especially for power analysis.  May wish
to
     do bootstrapping or other simulation to estimate power if the
assumptions
     are badly violated
 
   - additional ways to tease out effects of tutor in fall data

     * fit gain ~ pretest + tutor, and gain ~ pretest + tutor + other
covariates,
       to see if there is an effect for tutor condition after adjusting for
pretest

     * Examine corr(pretest,gain) within tx groups, and examine models like
       gain ~ pretest * tutor, to see whether an argument can be made that
the 
       tutor (partially) counter acts the "rich get richer" hypothesis

     * organize the paper so that the message answering the question "What
can
       we learn from this data" will stand out clearly.

3. please remember your final paper is due the last day of classes.

Best,

-BJ