Re: Graphics Resolution

From: BIGOS@Darwin.Stanford.EDU
Date: Wed Sep 24 1997 - 08:01:19 EST


Discussions regarding smoothing of graphs can (and have) degenerate into
fanaticism one associates with religion. I congratulate Howard in adding art and
life! 

I tend to favor smoothing of data for graphical presentations. I will give two
justifications, one practical (art?) and one theoretical (religion?). What this
does for our lives is left as an exercise for the reader. 

Practical: 
--------- 
When discussing results, smoothed data tends to paint a "prettier picture." That
is, I believe it generally allows one to see the aspects of flow data which are
relevant (e.g. population positions and relative frequencies) while it ignores
the aspects which tend to be irrelevant (e.g. statistical noise). I hope that
most analysis packages, when doing computations on data, use the "raw"
measurements and not the smoothed values for graphical output! 


Theoretical: 
----------- 
In almost all cases, flow work is used to draw inferential conclusions. We are
trying to classify a tumor, an individual's blood makeup, or a mechanism in a
biological system. We make the assumption that the objects in the tube that we
run through the flow system are representative of the "larger" population, and
use tools of statistical analysis to make the inferences. Inaccurate results
can, and have occurred, when this assumption isn't met. 

The data generated by measuring a tube in flow, is, in this sense, a sample
distribution of the underlying populations. Smoothing techniques, to a large
extent, are designed mathematically to give a better graphical representation of
that underlying population. From this, I believe, the practical point I
mentioned above follows. 

"Raw data" - the actual digitizations from the flow instrument, not only
includes a sample of the underlying distribution, it also includes noise
introduced by the instrument (I am ignoring biological preparation here). When
measuring instrument performance, graphical views of this may have relevance.
For most biological systems I believe it doesn't. 


Philosophical(??) 
------------ 
I suppose we can draw all sorts of psychological conclusions about an
individual's preference for "raw" or "smoothed" graphs. However, we should keep
in mind that it is just "art" and not "life". 


-Marty Bigos Stanford Shared FACS Facility 



This archive was generated by hypermail 2b29 : Wed Apr 03 2002 - 11:50:09 EST