thanks, i am afraid that i mis-understand it before
thanks again
To clarify, if by "precise" you mean, is dividing the reported significance
by 2 just an approximation, the answer is no. If by "precise" you mean,
are the results accurate to enough decimal places, it depends on how many
decimal places you want and need. A reported 2-tailed significance of .003
might correspond to a 1-tailed significance of .001 or .002 (e.g. it
depends on whether the .003 is really .0034 or .0026), so if you want to
have accuracy to 3 or more decimal places you should do what Nick or Kit
suggest.