Quad wrote: 
> 
> Statistical significance is a zero/one decision. Either it is
> significant or it is not. You can't tell anything more.
> 

Well, I understand your point, but I respectfully disagree. Without
digging two deeply into experimental design, in my own field we tend to
look at statistical tests of hypotheses and consider things like
parameter estimates from something like an OLS regression equation or
ANOVA, etc. and we want to know the p-value associated with the
parameter estimate rather than only whether the estimate is either
significant or not based on the pre-planned significance level cutoff. 
Say the p-value is 0.01. This is significant under most (all?)
interpretations.  Likewise,  p-value of 0.05 or lower is typically
considered  significant.  But let's say a parameter estimate is
significant at a p-value of 0.07 or 0.11.  This is certainly not the
same as an estimate with a p-value of 0.65.  Bottom line, is that we are
equally interested in the t-stat or F-stat and related p-values as we
are in the labeling of a result as significant or not. 

Obviously, standard practices of reporting research results differ
greatly across fields. And of course there are many other approaches to
analyzing data from experiments other than traditional statistics
approaches.  Note: I haven't read the links noted, so my comments are
generic, not related to these specific tests.


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=97530

_______________________________________________
audiophiles mailing list
[email protected]
http://lists.slimdevices.com/mailman/listinfo/audiophiles

Reply via email to