Comparison of the mode of lognormal distributions

Hi. I'm trying to compare the modes of a set of lognormal distributions. I realised I could write a function to fit the data (using lognfit) and bootstrap to estimate CI/variance, but I was wondering if a more computationally simple approach existed (as I'm analysing a lot of samples and the distribution is so similar to the normal). Thanks

Answer 1

[Shashank Prasanna]

If you'd like to compare distribution you can use LILLIETEST or KSTEST

http://www.mathworks.com/help/stats/lillietest.html

http://www.mathworks.com/help/stats/kstest.html

[Will]

Thanks but the lillietest and kstest are tests for the normality of a distribution. I'm comparing whether the mode of samples, which possess the lognormal distribution, are the same.

[Shashank Prasanna]

Those tests can be used for non-normal distributions as well. Specify the distribution as shown in the links.

Is the end goal to understand if the samples come from the given distribution? or just to compare the modes?

[Will]

The end goal is just to compare modes. The difference between distributions is easily shown with a log transformation, to normalise the data, and comparison of mean and variance. However, I'm working with people with little maths/stats background and (probably as they're easier to visualise) I've been asked to perform a comparison of these peaks. Thanks