When doing lognormal method of moments, we would want to set expected value of lognormal = x bar. Let's say we have observations X1..X10. Would x bar be Sum(Xi) / 10, or Sum(LN(Xi)) / 10 ? I'm guessing since X is distributed lognormally, we would just sum the X and divide by 10. Sorry if this is obvious, but I saw a question earlier today that somehow confused my memory of this topic. Is there any situation where we would use Sum(LN(Xi)?
Lognormal Method of Moments