Re: What kind of test can I use?
- From: Richard Ulrich <Rich.Ulrich@xxxxxxxxxxx>
- Date: Wed, 08 Aug 2007 21:41:46 -0400
On Tue, 07 Aug 2007 20:57:46 -0700, ihorganization@xxxxxxxxx wrote:
Hello, and thanks for helping me out.[snip, example]
I need to figure out what kind of test, if any, I can use to reject/
not reject a hypothesis.
The data concerns one individual at two different times. Thus the
data, I assume, is not independent.
The first set of data was obtained through observation of behaviors
outwardly expressed by an individual during sleep in standard
conditions. The second set of data was obtained through observation
again, but after the individual was placed on medication. Again, the
total amount of time over 7 nights a behavior was observed was divided
by the total amount of sleep over 7 nights to get a percent of how
much, on average, per night, an activity was performed.
What test can I run to see if there is a significant difference in the
amounts of each of the motions using only the averages (percentages)
and the total number of instances each behavior was observed over 7
nights.
I guess what I'm looking for is a test that can compare the
differences in averages only (thus there is no standard deviation to
use) but also take into account the number of instances that were used
to come up with the average to see if there's a significant
difference.
Well, that is a bad idea, comparing averages with no s.d.
Your best start might lie in plotting behaviors across 14
nights, or even, across the hours for 14 nights.
I have had experience in modeling drug intervention
and its effect on sleep EEG. This was for patients whose
depression gave them unusual sleep at the start, and
whose drug treatment had large effects. -- There were several
patterns seen for different sleep-variables --
1) No change.
2) A immediate drop/jump, with the next nights showing
a) further change;
b) no further change; or
c) rebound back toward the original;
3) Gradual change after the start of drug (usually, in
a 'normalizing' direction... but that would be interesting).
--
Rich Ulrich,wpilib@xxxxxxxx
http://www.pitt.edu/~wpilib/index.html
.
- References:
- What kind of test can I use?
- From: ihorganization
- What kind of test can I use?
- Prev by Date: Re: Measurement error in longitudinal studies
- Next by Date: Re: Linear regression and Logistic regression
- Previous by thread: What kind of test can I use?
- Next by thread: a question about time series
- Index(es):
Relevant Pages
|
