🚀 go-pugleaf

I am involved in development of a new measurement technique and am planning
to do some statistical analysis.  Will the statistics experts out there
please tell me if I am making any serious mistakes, suggest ways to extract
more information from the data, or point me toward the proper books or
software to use?  Please send E-mail replies, I will summarize if there is
interest.  Thanks.

We measure a quantity which should be related to the mass of an object.  We
want to calculate statistics which indicate how precise the measurement is.

We have a population of objects which is about 90% pure.  We can repeat the
measurement a limited number of times on the same object.  We want to take
a sample of measurements, throw out the measurements of impurity objects,
and calculate the average and standard deviation for the pure objects. 
What critera should be used to throw out impurities?  How many samples are
needed for routine measurements?  How many samples are needed before this
question can be answered the first time?

The purity might not be known in advance.  How many measurements must be
made to conclude that either the population is not pure at all, so that a
measurement is meaningless, or that it is reasonably pure?

In a related experiment, we have a mixture of two objects of different
masses, say 40% of each, with 10% impurities.  We want to take a sample of
measurements and classify each as object A, B, or impurity, and calculate
an average and standard deviation for A and B.

When the mass of A gets close to that of B, so that the tails overlap
seriously, what is the proper way of calculating averages for the two
distributions?

If these questions are answered in some standard text, please recommend one
which is slanted toward this sort of problem.

--
Edward J. Huff   huff@mcclb0.med.nyu.edu   (212)998-8465
Keck Laboratory for Biomolecular Imaging
NYU Chemistry Deptartment, 31 Washington Place, New York NY 10003
Posted from the new Mac NewsWatcher 1.3d5