This is a multilevel problem since, potentially, oranges from the same tree
could be more like each other than they are to oranges on other trees.
therefore, there is variance at each level, variability of the oranges on a
tree and variability between trees in the orchard. The problem with the
sample size necessary is that it requires you know the degree to which the
trees differ relative to the variability of the oranges on each tree, or the
intraclass correlation. There is a multilevel list where you could raise
this issue (you can get there by going to
http://www.stattransfer.com/lists.html). They will want to know the
intraclass correlation as well. You may also rreference a multilevel book,
such as Snijders and Bosker (1999) Multilevel analysis available from Sage.

Paul R. Swank, Ph.D.
Professor, Developmental Pediatrics
Medical School
UT Health Science Center at Houston



-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On
Behalf Of Gregg Maggioli
Sent: Wednesday, January 29, 2003 12:09 PM
To: [EMAIL PROTECTED]
Subject: Sample Size - Help!


Problem: What sample size do I need to have a confidence of 90%, 95%,
99% (or close to these percentages) that the sample represents the
population

Background: We are an orange grower and this problem relates to seeds
in oranges.  Our goal is to provide our customers with seedless
oranges but the nature of oranges is that they typically have seeds.
Growing practices, climate changes (too much rain, too little sun�)
effects if a seed forms in an orange.  I envision sampling an orchard
prior to harvesting to get an accurate assessment of the % of oranges
that contain seeds.  A single orchard might have 1000 trees @ 100
oranges per tree (100,000 oranges per orchard).  The grading would be
perform by cutting an orange open and inspecting it for seeds and
placing it into one of two categories � contains seeds or is seedless.

Goal: I want to create a sampling program that will determine at the
confidence levels above that the oranges we are providing are
seedless.

I have a couple of semesters of college statistics but it has been a
long time since I have used any of it.  Could you please describe how
I can calculate a sample size with its corresponding confidence
interval for the sampling goal stated above?

I want to be able to say something similar to - "I am 95% confident
that 2% of the oranges in this orchard have seeds"
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================


.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to