Discussion: View Thread

I would be interested to hear from the community how people refer to the analysis of data using Generalized Linear Mixed Models & General Linear Mixed Models when consulting. I feel vaguely uncomfortable calling them "ANOVA's" because of the implication that you are refering to an OLS solution but its such an entrenched term that I usually end up using it. Using terms like GLMM often has no meaning to the client?

-------------------------------------------
Neil Paton
Statistician
Provimi
-------------------------------------------

It would depend on the audience. But typically I would refer to GLMM as a "Linear Model".
"Generalized" doesn't provide any information to most audiences, nor does "Mixed".

-------------------------------------------
David Bristol
Statistical Consulting Services
-------------------------------------------

If you don't use fancy words that the client has never heard before, how can you justify the huge consulting fee you charge?

Seriously, I always use the appropriate language and explain that it is similar to regression (or ANOVA depending on the context) but it better matches the correlated structure of the data set. I expect that this causes some short term confusion and consternation, but in the long run it puts them in a better position for future consultations and for reading the medical literature.

-------------------------------------------
Stephen Simon
Independent Statistical Consultant
P. Mean Consulting
-------------------------------------------

I agree.  The client is intelligent and just not highly trained in statistics.  It is our job to give them little tutorials.  I would even venture further to explain the link function.  It is simply a transformation that improves the model for the data being fit. It is not all that hard to explain.
-------------------------------------------
Michael Chernick
Director of Biostatistical Services
Lankenau Institute for Medical Research
-------------------------------------------







Original Message:
Sent: 01-05-2012 17:46
From: Stephen Simon
Subject: How to refer to Generalized Linear Mixed Models and the like when consulting.

If you don't use fancy words that the client has never heard before, how can you justify the huge consulting fee you charge?

Seriously, I always use the appropriate language and explain that it is similar to regression (or ANOVA depending on the context) but it better matches the correlated structure of the data set. I expect that this causes some short term confusion and consternation, but in the long run it puts them in a better position for future consultations and for reading the medical literature.

-------------------------------------------
Stephen Simon
Independent Statistical Consultant
P. Mean Consulting
-------------------------------------------