Geometric standard error of the mean

Dear forum members,

I wonder if anyone can share with me a formula for calculating geometric (multiplicative) standard error of the mean based on original values and (separately) on geometric standard deviation. Unfortunately, the definition of SEM* in the attached article (Table 2) does not shed much light on this point.

Thank you,




Active Member
reeewwiiiind. Today on talk stats were taking a look at old problems that didn't quite hit the interests of the day.

Looking at these formulas, it appears to stem from the notion that CV_hat ~= X_bar/SD = sqrt( exp( var log(X) ) - 1 ). So their 'ohm' is Var( log(X) ), and you can see that by solving for VAR( log(X) ) in the equation. I believe the wikipedia log-normal page gives an equivalent derivation in one section.

Well not winning any awards for editing, i like the manuscript still includes 'insert table 2 about here'? But I think the math is right or so far as i see.