Question on "It's (In) The Water"

[Muttling]

Greetings Doc,

I read your first article concerning the water quality study you have been working working on. Obviously, I am quite interested in your future articles concerning observations and correlations.

I do have a question concerning your data analysis. Did you perform any tests to determine if the data sets or the transformed data sets were normally distributed or were the analyses strictly non-parametric?

Other questions abound, but I suspect they will be addressed in future articles. For example, what was the quality of source water for the tanks? (e.g. how many were RO, RO/DI, tap, etc.) Another example would be where there any possible details or correlations that could explain the outliers and what is the effect of rejecting the outliers from the analysis?

Thanks,
Mutt

[rshimek]

Originally posted by Muttling

Hi Mutt,

I do have a question concerning your data analysis. Did you perform any tests to determine if the data sets or the transformed data sets were normally distributed or were the analyses strictly non-parametric?

The analyses to date and their rationales are described in the materials and methods, I think. So far they are non-parametric. I will be doing some parametric tests (correlation analyses, etc.), but I am not going to be doing any tests for normality. The data are not too dispersed, and I don't think any deviations from normality will materially alter my conclusions.

Other questions abound, but I suspect they will be addressed in future articles. For example, what was the quality of source water for the tanks? (e.g. how many were RO, RO/DI, tap, etc.)

It varied a lot, but most were RO/DI, a couple were tap water, and one or two used NSW. I think you will see that there was really no effect on the samples from the water source used.

Another example would be where there any possible details or correlations that could explain the outliers and what is the effect of rejecting the outliers from the analysis?

There are some, but I am trying not to reject any outliers, the sample size is simply too small.

:frog: