Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.
Showing results for
Search instead for
Did you mean:
Our World Statistics Day conversations have been a great reminder of how much statistics can inform our lives. Do you have an example of how statistics has made a difference in your life? Share your story with the Community!
I'm asking for help on a proper analyzing a dataset
I have a dataset based on a trial where we fed two different diets (S and A) and applied two different treatments (T and C) in 3 blocks animals.
We want to test the effect of treatment in two not normal distributed parameters (Hg and Leuco) in the two different diets and among the time points sampled.
I'm used in performing Mixed Model (with diet, treatment and time point as fix effect; experimental unit nested in the block as random effect), however, with not normal parameters mixed model is not the appropriate approach (I think).
I'm new to general regression and my purpose is check for differences between treatments among the time point sampled or/and among the diets (I have a lot more parameters than these two, Hg and Leuco), but this one is just an example.
Attached an example of my analysis.
I hope somebody can suggest me the right way to analyze it.
Hi, What do you think is the appropriate distribution for Hg and Leuco? Just from looking at those 2 variables in your dataset they look bell-shaped. I would expect that a normal distribution would be an adequate model. I assume that "Hg gr%" is a %age. Therefore, strictly speaking, it can't be normally distributed as it is bounded by 0 and 100. However, unless you have lots of data points at or very near to 0 or 100, a normal distribution will often give you a useful model. Phil
It does sound like generalised linear mixed modeling would be most appropriate. However, they are currently not supported in JMP or JMP Pro. It has been requested in the JMP Wish List. You can add your vote.
In this case I think you may find that the regular linear mixed models might be adequate.