Where extreme cases (e.g.very early arrivals) would affect the mean
(average date of arrival), one should use the median, which in this
instance would be the middle date, determined by counting up (or down)
from the earliest (or latest) date until you have reached the date
representing 50% of the total number of dates in the data set.

Roger Olstad
Seattle (Lake Forest Park)

On Mon, 2 Feb 1998, Michael Price wrote:

> But this is something I've never quite figured out how to deal with: data
> that are waayyy out of pattern: do you include them knowing they're likely a
> one-off that's gonna distort whatever means and averages, or do you ignore
> them? For example, let's say arbitrarily that this is the pattern of Barn
> Swallow arrival, leading to the derived (and equally arbitrary for purposes
> of this immediate discussion) average arrival date of April 07 above:
>
> 4 12
> ----|----|----|----|----|----|----|----|----|----|----|----
> J F M A M J J A S O N D
>
> then along comes 1998, which leaves the record looking like this:
>
>
> 1 4 12
> ----|----|----|----|----|----|----|----|----|----|----|----
> J F M A M J J A S O N D
>
> which then pulls the average arrival date sometime into mid-March or even
> late February, clearly not reflecting reality. I'm sure there's something
> simple way to absorb that kind of temporary distortion of the 'normal'
> pattern that I'm overlooking here. Anybody with statistics in their blood
> (ie, related to that famous Gaul, Vitalstatistix '-) who knows how to fix this?
>
> Michael Price A brave world, Sir,
> Vancouver BC Canada full of religion, knavery and change;
> mprice at mindlink.net we shall shortly see better days.
> Aphra Behn (1640-1689)
>
>