Another more recent jest asserts that attaching
asterisks to classes of P-values is `a symbolism
more appropriate to an hotel guide-book than
a serious scientific paper' (P. Sprent JRSS A
133: 139-165, 1970; quotation on p.143). Such
disapproval hasn't stopped it becoming
standard in some fields, and more importantly
in some journals.
Maybe the fact that something becomes a standard "by mistake" in an
indication of pent up demand. I understand the purists who would rather
see p-values and confidence intervals. However, there is little
difference between a p-value and a confidence interval and NO
difference between a confidence interval and a significance value
symbol. Economist are also purists in that they will always request the
standard errors... of course once you have that (and N) you can derive
the p-values etc. Thus, if we believe that the estimate and the
standard errors are the fundamental pieces of information, everything
else is just window-dressing/visualization. From that perspective I do
not find it surprising that the asterisks have caught on since they are
both compact and very clear.
