Subject: Sampling

# Sampling

From: Karen Brynjolf Pedersen <karen.brynjolf.pedersen>
Date: Thursday, November 22, 2001
```Pierre-Emmanuel Nyeborg <penyeborg [at] noos__fr> writes

>I am having trouble figuring out the size of the sample we should
>examine to get an accurate view of the problems. And I also have
>problems figuring out the time needed for the examination of each
>sample. I would appreciate any suggestion, remarks or advice.

I have used the method of Carl M. Drott to make surveys of different
photographic collections. The statistical method is founded on the
principle of random selection and is described in his article
"Random Sampling: a Tool for Library Research" in College and
Research Libraries, March 1969 p.119-125. In his article is a table
showing different options regarding the size of the samples,
depending on how you want the tolerance and the confidence of the
survey. For instance if you want your survey to have a 90%
confidence and a 7% tolerance, your sample size should be 138,
according to the table. The confidence is the probability that the
answer is right. If the confidence is 90% this means that if you
repeat the survey 100 times, you will get the same result 90 times
out of 100.The tolerance is the accuracy of the survey, for instance
plus/minus 7%. I think the article makes an excellent help for
surveying collections. If you cannot find the article, I will be
happy to Fax: it to you.

As for the time being used on the survey, I think most time is spent
on the planning of the survey and an accurate description of
damages, photographic technique et cetera to describe each sample.
How much time you use of course depends on how much information you
want from each sample and how the collection is organised. If your
collection consists of various housing systems you might want to
take for instance 138 samples of each type of storage system.

I can tell you that I have used about 200 hours to take samples from
a collection with 7 different storage systems each containing
various kinds of negatives and prints, and 138 samples were taken
from each storage system. The time spent includes planning the
survey, taking the samples, and analyzing the results, but not
writing a report which still awaits (alas). I hope this can give you
some ideas for your survey--I am sure there are other ways to do
this.

Karen Brynjolf Pedersen
Conservator of Photographs
Department of Conservation
The National Museum of Denmark
Brede
P O Box 260
DK-2800 Lyngby
Denmark
+45 33 47 35 51
Fax: +45 33 47 33 27

***
Conservation DistList Instance 15:39
Distributed: Tuesday, November 27, 2001
Message Id: cdl-15-39-011
***
```
Received on Thursday, 22 November, 2001

URL: http://
Timestamp:
Retrieved: