Hello everyone,
I am building a meta-regression investigating the carbon stock of soil across regions using both individual and aggregate data from several papers. I specified sigma = TRUE in my se() phrase as suggested by Dr. Buerkner in this thread. As is, my model is working, but I am concerned about how the individual values are weighted relative to my aggregate values.
My understanding is that each observation is attributed a weight based on its precision (based on the standard error). To incorporate the individual values, I specified sigma = TRUE which assigns them a standard error equal to the residual standard deviation.
I am worried that individual observations with low error attributed by sigma will have greater weight than my aggregate values that are formed from sample sizes ranging from 2 to 200.
Do any of you have insight into how the weighting of observations will work in this context? Also, if any of you have recommendations for literature that go into greater depth on the se() function I would really appreciate it.
Below is my code.
mod_Region ← brm(Stock | se(StockError, sigma = TRUE) ~ Region + (1|publication), …
I’m new to brms and bayesian statistics so I’m sorry in advance if this is an elementary question. Thank you for your time!