One feature of Statalist is its scope for discussion across quite
different groups.
I find endless minor fascination in the way that the same concepts
acquire different names (and also how different concepts acquire the
same name, although that's not the story here).
As I mentioned in a thread recently, whether zeros are to be believed or
not has attracted a variety of terminology. I give here terms I can
think of encountering, and my question is whether anyone can add to the
list. The point is not to think thesaurus-like of other terms that could
be used, but to report terms you have found repeatedly in some
literature or among some group.
That is, for variables that could be zero or positive, zeros are
commonly divided into two classes.
1. Zero may be observed, or more precisely reported, but that is not to
be taken that the zero is absolutely true if there is no compelling
reason to expect zero under all circumstances.
Suppose I survey 2000 U.S. citizens, and none of them tell me that they
would like Bill Gould to be President. That does not mean that no-one
anywhere wants Bill Gould as President, just that I didn't find anyone.
Or suppose a lab tells me that the concentration of some compound in a
sample of soil or water is zero. That doesn't mean necessarily that
there are definitely no molecules of that compound present, just that
the technology didn't detect any.
Such zeros are known in various groups as
Random zeros
Sampling zeros
Rounded zeros
Non-detects
2. Zero is observed and that's right. A positive value is impossible or
inconceivable.
Pregnant males used to be the canonical example. (No, don't tell me
about the exceptions.) Note that the term "real zeros" is bespoke.
Such zeros are known in various groups as
Fixed zeros
Structural zeros
Essential zeros
Any more?
Nick
[email protected]
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
