Hi boobee:
"Denoted by H[0], the null hypothesis is a statement about some
specific value of mu [the Greek character which stands for the
population mean] or pi [the Greek character which stands for the
population proportion] that we hope to be able to reject at the
completion of the testing procedure."
"Denoted by H[a], the alternate hypothesis is a statement about a
range of values for mu or pi."
From: Contempory Business Statistics, by Hummelbruner, Rak, and Gray
(ISBN: 0-13-369182-9)
As for a real world example:
A drug treatment center advertises that 85% of the patients it
discharges do not use any harmful drugs in a one year period directly
after leaving the center. However, the Director of the center has
reason to believe that this figure is no longer accurate. A survey of
50 patients who were discharged between one and two years ago is done
and it is found that 38 of them stayed away from drugs during the
prescribed period. Can the success rate claim be maintained with a 95%
confidence level?
To answer the above question, the hypotheses would be set up as
follows:
H[0] = .85
H[a] <> .85
(The "<>" stands for "not equal", which is also sometime denoted by an
equals sign with a diagonal stroke through it.)
I hope this answers your question.
Search Strategy:
None - personal experience.
Some websites to check out:
Statistical Hypothesis Tests
http://www.ganesha.org/spc/hyptest.html
Step 1: The Hypothesis
http://www.studyworksonline.com/cda/content/article/0,,EXP938_NAV2-76_SAR933,00.shtml
websearcher-ga |