Re: Basic rookie question about standard error


I agree with Paige and say that rarely is it the case that zeroes
should be dropped. Usually when zeroes are dropped, this is done
because someone coded the data in such a way where zero does not mean
zero, e.g. it means missing.

In the example Bob gave, for instance, he assumed that zero means the
shop does not sell live lobsters at all... what if there exists a shop
that failed to sell any lobsters even though it does sell lobsters?
Here, the problem is that zero was coded to mean two things: zero and
missing/NA.


Richard Ulrich wrote:
> On Thu, 18 Aug 2005 12:06:32 GMT, Paige Miller
> <pmiller5NOSPAM@xxxxxxxxxxxxxxxx> wrote:
>
> > On 8/18/2005 4:11 AM, coleman.bill@xxxxxxxxx wrote:
> > > I have a problem and I'm hoping that one of the gurus here will be able
> > > to point me in the right direction!
> > >
> > > I'm attempting to get the standard error for a set of data. Basically,
> > > I'm counting the number of occurances of x per item, y. The problem is
> > > that there is a large number of Zero values in the results - i.e. the
> > > "y" item has no occurances of "x" - a typical set of data would look
> > > like this:
> > >
> > > 0, 20, 34, 0, 0, 0, 46, 15, 0, 29, 0 ,0, 32, 0
> > >
> > > So should I calculate the the standard error including the zero values,
> > > or should I remove the zero values?
> > >
> > > I know it's probably a silly question, but I can't find the answer
> > > anywhere and it's starting to become a problem!!! Thanks to anyone who
> > > can help me.
> >
> > Why would you possibly want to remove the data that is a zero value?
> > What reason is there to think that they do *not* contribute to the
> > variability of the data that you are trying to estimate? They are data
> > -- they tell you something.
>
> On the other hand, sometimes it is useful to know both
> versions of the information -- Here are (say) "the average
> (and SD, and SE) of the number of children per household,
> and here are the similar statistics where there is at least
> one child."
>
> Counting in the zeroes can be misleading. But be sure
> that your audience knows what you are doing, because
> silently dropping zeroes, ones that folks might figure are
> there, would be even more misleading.
>
>
> --
> Rich Ulrich, wpilib@xxxxxxxx
> http://www.pitt.edu/~wpilib/index.html

.

Relevant Pages