Lognormal stats from normal stats
Message posted by Paul on October 4, 2000 at 12:00 AM (ET)

I have mu and standard deviation for normal distribution of numerous data sets.

I want to calculate mu and standard deviation for lognormal distribution of the data sets.

Are there equations to transform the normal mu and standard deviation to lognormal?

Paul

READERS RESPOND:
(In chronological order. Most recent at the bottom.)

Re: Lognormal stats from normal stats
Message posted by JG on October 4, 2000 at 12:00 AM (ET)

Re: Lognormal stats from normal stats
Message posted by Paul on October 8, 2000 at 12:00 AM (ET)

> Remember, if Z is normal then log(Z) or ln(Z) is the lognormal.

I don’t think I understand this. You seem to be saying that I can calculate Z for lognormal data directly from the Z from the normal data stats via ln(Z).

The following is an example of what I want to do.

Given the following data:
0.03063 0.02779 0.02644 0.02415 0.03050
I calculate mu = 0.02790, and sd = 0.002466

I want to know what value corresponds to Z = 2.0 calculated from ln data stats.

With the ln of the above data,
-3.4858 -3.5830 -3.6330 -3.7234 -3.4900
I calculate mu = -3.5830, and sd = 0.08978

From the ln data, at Z = 2.0, the ln data point = -3.4034,
which corresponds to 0.033259 in the normal data.

However, if I calculate Z for 0.033259, I get Z = 2.172 and ln(Z) = 0.77561, not 2.0

For ln(Z) = 2.0, Z = 7.39138, which corresponds to 0.04613 in the normal data, not 0.033259.

It appears to me that I need to calculate Z using ln data stats. Hence, my motivation for equations to transform normal stats to lognormal stats.

Your thoughts?