[R] Predicted values from glm() when linear predictor is NA.

Ebert,Timothy Aaron tebert @end|ng |rom u||@edu
Thu Jul 28 16:19:39 CEST 2022

Sadly, I see that your practical options are limited. In regards to time and thermal units being equivalent if temperature is constant. The practical answer is yes. The technical answer is no because the growth chamber is not able to keep temperature constant and the insect can moderate the experienced temperature by moving to different parts of the plant (assuming this used plants and not artificial diet or leaf disks). Temperature might be more constant if these are mosquito larvae in vats of water. The water would have thermal mass to even out temperature fluctuations in the growth chamber.
   Often people assume that a growth chamber set to 25C is always 25C. Data show this is not the case even in high end or custom made growth chambers.
   Using time as a surrogate for accumulated thermal units under constant temperature will introduce some error in time due to non-uniform and non-constant temperatures within the growth chamber. Mostly the data are not present to document this let alone include this in the model. So we take the biologists view of statistics and ignore the problem that we cannot solve.

  If I measure time in days, then it makes sense that I can have egg hatch at time 0. However, it is not biologically possible to have anything happen in zero time. If my accuracy in time measurement is "days" then maybe I should consider introducing some small time value for egg hatch. Say eggs hatch at time 0.2 days. That is below the resolution of my data but reconciles the biological impossibility of anything happening in zero time. I note that time is integer in the data set.

  Another odd thing in the data. I assume that the values for "alive" represent the number of living individuals at different time intervals. At the end of the "Alive" data I note that there are a few time intervals where none are alive followed by a living individual. I am not a fan of zombies, or raising the dead (insects). Either the definition of dead is more Monty Python (https://www.youtube.com/watch?v=Jdf5EXo6I68), or I don't quite understand the data.


-----Original Message-----
From: Rolf Turner <r.turner using auckland.ac.nz> 
Sent: Wednesday, July 27, 2022 10:10 PM
To: Ebert,Timothy Aaron <tebert using ufl.edu>
Cc: r-help <r-help using r-project.org>
Subject: Re: [R] Predicted values from glm() when linear predictor is NA.

[External Email]

On Thu, 28 Jul 2022 00:42:51 +0000
"Ebert,Timothy Aaron" <tebert using ufl.edu> wrote:

> Time is often used in this sort of problem, but really time is not 
> relevant. A better choice is accumulated thermal units. The insect 
> will molt when X thermal units have been accumulated. This is often 
> expressed as degree days, but could as easily be other units like 
> degree seconds. However, I suspect that fine time units exceeds the 
> accuracy of the measurement system. A growth chamber might maintain
> 28 C, but the temperature the insect experiences might be somewhat 
> different thereby introducing additional variability in the outcome.
> No thermal units accumulated, no development, so that is not an issue. 
> This approach allows one to predict life stage over a large 
> temperature range. Accuracy can be improved if one knows the lower 
> development threshold. At high temperatures development stops, and a 
> mortality function can be added.

Very cogent comments in respect of dealing with the underlying practical problem, but I am not so much concerned with the practical problem at the moment but rather with the workings of the software that I am using.



P.S.  I am at several removes from the data set(s) that I am messing about with.  But if my understanding is correct (always an assumption of which to be sceptical!) these data were collected with the temperature being held *constant*, whence time and accumulated thermal units would be equivalent.  Is it not so?


Honorary Research Fellow
Department of Statistics
University of Auckland
Phone: +64-9-373-7599 ext. 88276

More information about the R-help mailing list