>>> The use of statistical analysis to draw conclusions from seemingly
>> confused data is one of my biggest concerns about the studies we read and
>> (try to) believe.
> To me, the human eye is so good at pattern detection, that I'd rather see
> the data simply graphed out, with ranges, error bars, or standard
> deviations, than to read complex statistical terms and analyses. If
> results aren't completely obvious from the graph, then they don't resonate
> with me.
I also like to see ALL the data presented this way since often the
exceptions and outliers are more interesting and thought-provoking than
the 'normal' points.
Although the trend is interesting and useful, its shape or slope is
highly dependent on how the outliers are treated.
Whereas trends are important and interesting, in the real world where we
live and work, the outliers may be the most significant and present the
real problems. A picket fence with a few boards missing is hardly a
effective fence at all. Missing boards are outliers, but they are
hugely significant. Rabbits will be found on both sides, no matter how
good the fence is on average.
Consider the a case where, on average and for most of the samples,
almost total control of a disease or pest is achieved by a treatment,
but in individual cases _no_ control is achieved. Concentrating on the
trend and average can result in false confidence in that control,
neglecting the real danger of a continued a presence of reservoir of
infection in a sufficient concentration to be capable of triggering a
serious breakdown incident.
This why IMO, strip-type chemical treatments and antibiotic treatments
tend to be quite successful, assuming they are applied correctly,
whereas less powerful fume or biological methods can result in
incomplete control unless used by a scrupulous, competent and observant
operator, and sometimes even then.
***********************************************
The BEE-L mailing list is powered by L-Soft's renowned
LISTSERV(R) list management software. For more information, go to:
http://www.lsoft.com/LISTSERV-powered.html
Guidelines for posting to BEE-L can be found at:
http://honeybeeworld.com/bee-l/guidelines.htm
|