"Highly significant" conflates statistical rarity with impact (importance
of the effect, the size of the effect).

On the other hand, I think "approaching significance" can be useful and I
will defend that practice (although I wouldn't push its use in a
publication).

Many statisticians note the arbitrariness of the decision criterion (the
magical .05) and argue that a result that would occur randomly with a
probability of .051 or .052 or .06 (I could go on . . . it is a slippery
slope) deserves closer examination than just deciding that the result is
does not meet the criterion to be declared statistically reliable.  This
rigidness in the decision process seems to reinforce the too-common
treatment of statistical analysis as a ritual of taking out data (our
sacrificial goat, as it were) to the oracle for a decision.  We can be more
thoughtful than this.  (Abelson's excellent book, *Statistics as Principled
Argument*, has some discussion of the thoughtful use of inferential
statistics.)

Failure to reach the criterion can occur for reasons other than absence of
an effect.  The near misses are worth examining.  Similarly, the
just-made-it "successes" deserve replication and questions about Type I
Errors.

Claudia

_____________________________________________

Claudia J. Stanny, Ph.D.
Director
Center for University Teaching, Learning, and Assessment
Associate Professor
NSF UWF Faculty ADVANCE Scholar
School of Psychological and Behavioral Sciences
University of West Florida
11000 University Parkway
Pensacola, FL  32514 – 5751

Phone:   (850) 857-6355 (direct) or  473-7435 (CUTLA)

csta...@uwf.edu

CUTLA Web Site: http://uwf.edu/cutla/
Personal Web Pages: http://uwf.edu/cstanny/website/index.htm

---
You are currently subscribed to tips as: arch...@jab.org.
To unsubscribe click here: 
http://fsulist.frostburg.edu/u?id=13090.68da6e6e5325aa33287ff385b70df5d5&n=T&l=tips&o=25141
or send a blank email to 
leave-25141-13090.68da6e6e5325aa33287ff385b70df...@fsulist.frostburg.edu

Reply via email to