<!doctype html public "-//w3c//dtd html 4.0 transitional//en">
<html>
Dear Statisticians,
<p>I need to compare the accuracy and precision of two sampling methods
of marine invertebrates.
<br>The first one involved sampling of 31 stations regularly distributed
over a large area. Each station was sampled with 3 small sampling units.
<br>The second one involved sampling of only 2 stations, each sampled with
3 large sampling units, the large sampling unit being 3 times larger than
the small one.
<br>The second method gave the greatest estimate of abundance with the
highest precision (i.e. SE/mean). However I would like to statistically
test this.
<p>To test the relative accuracy of the sampling methods, I have first
looked for a probability distribution that fit the frequency distribution
of the <b>sum</b> of the 3 values per station obtained with the small sampling
unit over the 31 stations in the first method. The distribution of the
log of the sums was well fitted by a normal distribution. From this normal
distribution, I have then calculated the occurrence probability of the
log of the values obtained from each of the large sampling unit in the
second method (i. e. sampling size of the large sampling unit equal sampling
size of three small ones) and considered that the second method was statistically
more accurate if the probability was less than 0.05.
<p>To test the precision of the methods, I have similarly looked for a
probability distribution that fit the frequency distribution of the precision
(SE/mean) in the first method. Again, the distribution of the log of the
precision was well fitted by a normal distribution. I have then calculated
the occurrence probability of the log of the precision obtained from the
two sampling stations in the second method and considered that the second
method gave a statistically higher precision if the probability was less
than 0.05.
<p>Could you please tell me if these tests are relevant?
<p>Many thanks in advance.
<p>Lenaick Menot
<br>&nbsp;</html>

.
.
=================================================================
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