On 13-Mar-06 Howard Mann wrote:
> Dear List,
>
> When reading a published article I look for the published
> estimate of the treatment effect on the primary outcome
> measure(s), and the associated 95% C.I., and disregrd any
> published p-values.
>
> I have seen the following comment in another Forum :
>
> "A major reason for including p-values with confidence
> intervals (CI) arises when sample data is not normally
> distributed. P-values are based on a normal distribution,
> whereas CIs are calculated from the data itself.
> In certain cases, a CI that does not cross zero (for absolute
> risk comparisons) is actually *not* significant. This is
> easily determined if a p-value for the comparison exceeds
> the significance (alpha) level."
>
> What does this mean ? What do you understand by this comment ?
Ignore that comment! It is a close approximation to nonsense,
or else is a description of bad practice.
> If the C.I. is wide, I might conclude that a claim (for instance)
> that the innovative intervention is superior to the control
> intervention to lack credibility, and that the result is better
> characterized as "indeterminate."
>
> If both ends of the 95% C.I. are on the "benefit" side of a
> pre-specified "minimum clinically important difference," I'll
> conclude that a claim of evidence of efficacy superiority (in
> this trial) is credible.
>
> Why should I bother with p-values ?
If the number published as "p-value" is simply a statement
like "P < 0.05", i.e. the p-value is not stated exactly,
but simply compared with a conventional level like 5%,
then you learn nothing new over and above the information
that the corresponding CI does not embrace the null hypothesis
value. (There is a 1-to-1 correspondence between hypothesis
tests and CIs: e.g. a 95% CI is the set of values not rejected
as NHs by a corresponding test at 5% significance level).
However, if the p-calue is cited exactly, as in "P = 0.0275",
then you do learn a bit more about how far the data are discrepant
from the NH than you do from the fact that the CI does not
embrace the NH.
Hoping this helps,
Ted.
--------------------------------------------------------------------
E-Mail: (Ted Harding) <[log in to unmask]>
Fax-to-email: +44 (0)870 094 0861
Date: 13-Mar-06 Time: 12:06:45
------------------------------ XFMail ------------------------------
|