In article <[EMAIL PROTECTED]> in
sci.stat.edu, Joe D <[EMAIL PROTECTED]> wrote:
>Given a set of sample values like 11.8, 10.9, 12.2, 12.9, 12.3, 13.8,
>13.4, 14.5 and 15, can I compute the CI or P-value for each sample
>relative to a given number 13.5?
My goodness, where do we begin?
1. Most important: You don't have "each sample" there. The nine
numbers are _one_ sample of size 9. The sample mean is 12.98 and the
sample standard deviation is 1.32.
2. You can compute a p-value relative to the hypothesis that this
sample is drawn from a population with population mean 13.5. You
must first decide whether you're doing a one-tailed or two-tailed
test. The two-tailed p-value is 0.271, and the one-tailed p-value is
half that.
3. There is no such thing as "the" confidence interval. You select a
confidence level, and then calculate a confidence interval from that
confidence level. There is no intrinsic reason for choosing one
confidence level over another, though it is true that 95% is the
most common choice.
4. You don't compute a confidence interval relative to a given
number. Rather, the CI concept is the inverse of the concept of a
hypothesis test or significance test. In a HT, you have some
particular value in mind for the population mean and you want to see
if your sample is consistent with that value. In a CI, you simply
follow the data to obtain a confidence interval for what the
population mean might be. The 95% confidence interval for your data
is 11.96 to 14.00, and the 90% confidence interval is 12.16 to
13.80.
Note that the "%" in the confidence interval is your confidence
level. That is not quite the same thing as saying that the true mean
is that % likely to lie within the interval.
--
Stan Brown, Oak Road Systems, Cortland County, New York, USA
http://OakRoadSystems.com/
It's not necessary to send me a copy of anything you post
publicly, but if you do please identify it explicitly to avoid
confusion.
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
