Log transformation of means

Feal

New Member
Hey!

I'm working on a meta-analysis and I want to use the means for moderator analysis. So some journal articles report log transformed values of the mean. In order to be able to conduct the moderator analysis, I need to retransform the logs.

Two questions here:

1) Sometimes studies do not report which type of log (ln or log) they've used. I do have to make an educated guess then, right?

2) Example: firm size is the logarithm of total assets (in millions of dollars)

=> If the mean of this example would be reported as 14,88. What is the exact retransformed value? Do I have to multiply the retransformed value by one million again?

1,251,683.496 * 1,000,000 = 2.899.358.315.628,94 => This number is way to high to make sense.

Some hints or tips on how to deal with that?

Kind regards,

Feal

kiton

Member
I would assume most studies use ln(X1) transformation. Now, to get the original value take the exponential -- exp(X1). E.g., exp(14.88) = 2899358.3