I have a log likelihood function that I was able to optimize using
mle2. I have two years of the data used to fit the function and I would
like to fit both years simultaneously to test if the model parameter
estimates differ between years, using likelihood ratio tests and AIC.
Can anyone give advice on how to do this?
My likelihood functions are long so I'll use the tadpole predation
example from Ben Bolker's book, Ecological Data and Models in R (p.
268-270).
library(emdbook)
data(ReedfrogFuncresp)
attach(ReedfrogFuncresp)
# Holling Type II Equation
holling2.pred = function(N0, a, h, P, T) {
a * N0 * P * T/(1 + a * h * N0)
}
# Negative log likelihood function
NLL.holling2 = function(a, h, P = 1, T = 1) {
-sum(dbinom(Killed, prob = a * T * P/(1 + a * h * Initial),
size = Initial, log = TRUE))
}
# MLE statement
FFR.holling2 = mle2(NLL.holling2, start = list(a = 0.012,
h = 0.84), data = list(T = 14, P = 3))
I have my negative log likelihood function setup similarly to the above
example. Again, my goal is to simultaneously estimate parameters from
the same function for two years, such that I can test if the parameters
from the two years are different. Perhaps an important difference from
the above example is that I am using a multinomial distribution (dmnom)
because my data are trinomially distributed.