Dear Tipsters,
Michael wrote:
> What is wrong with "confidence level"? If we can reject the null hypothesis
> at the .05 level, we are 95% confident that a real difference exists.
>
Strictly speaking, if alpha is set at .05 and p < alpha, then we
reject Ho. We mean that in repeated sampling the obtained difference
(or inferential statistic) could have come from the distribution
centred on no difference only 5% of the time.
Another approach is to calculate a confidence interval around the
point estimate (the obtained difference). We have to be careful
stating what this means, because the distribution underlying the
calculation is centred on the population parameter, not the point
estimate. Because we only have the point estimate, we cannot say that
the probability is 95% that the pop. par. lies within the calculated
interval. Nickerson (2000; see previous post for reference) states
that Bhattacharyya and Johnson argue that...."given an appropriately
computed confidence interval of x% it is correct to say that one is
x% confident that the parameter of interest lies within the
interval."
This is a subtle distinction, Nickerson says, but it seems to capture
what is and is not meant by a 95% confidence interval.
Hope this helps.
Stuart
___________________________________________________
Stuart J. McKelvie, Ph.D., Phone: (819)822-9600
Department of Psychology, Extension 2402
Bishop's University, Fax: (819)822-9661
3 Route 108 East,
Lennoxville, e-mail: [EMAIL PROTECTED]
Quebec J1M 1Z7,
Canada.
Bishop's University Psychology Department Web Page:
http://www.ubishops.ca/ccc/div/soc/psy
___________________________________________________
