NetNews Usenet Archive 1992 #19/NN_1992_19.iso/spool/bit/listserv/statl/1403

Text File | 1992-08-26 | 2.0 KB | 41 lines

Comments: Gated by [e-mail]
Path: sparky!uunet!paladin.american.edu!auvm!STATS.ADELAIDE.EDU.AU!WVENABLE
Message-ID: <[e-mail]>
Newsgroups: bit.listserv.stat-l
Date:         Thu, 27 Aug 1992 09:31:16 +0930
Sender:       "STATISTICAL CONSULTING" <[e-mail]>
From:         Bill Venables <[e-mail]>
Subject:      References to power calculations.
Lines: 30

I have a question about references somewhat like Meredith Warshaw's
brother-in-law's. A colleague came to me with the following query:

    Every year in our Psychology honours course students come in
    with project proposals.  They know in advance *roughly* both the
    size of the effect they are looking for, and the error variance.
    However when you look at the experiment they are proposing to do
    it's clear that either they are going to have very little chance
    of finding it, or it's a complete overkill.

    What references can I recommend to these students so that they
    learn something about power and sample size calculations in
    experimental design to make their proposals realistic?

I was tempted to say "Send them along to my inference classes", but that's
not likely to go down well with Psychology students.  My second response
was "Dixon and Massey: Intro..." but that was written in 1968, and next the
charts and tables in Cochran and Cox, but these are fairly obscure for stats
beginners.

I did my PhD work in this area, so I generate most of this stuff for myself
when I need it without going to the references.  So does anyone know of
some good, reasonably elementary, modern references for my colleague, and me?

I will summarise email responses if warranted.

Thanks in advance, Bill.
___________________________________________________________________________
Bill Venables, Dept. of Statistics, | Email: [e-mail]
Univ. of Adelaide, South Australia. | Tel: [phone]  Fax: ...[phone]