could only do the power analysis for one data-generating scenario (i.e., variation in K and z) and all estimation models, with 4 x 4 x 200 runs. I think its reasonable to do another 4 x 4 x 200 runs, with 4 estimation models, 4 simulation scenarios, and 200 replicates each, with sample size fixed at what we saw in the data. Breaking up factorials is a standard trick, e.g., here: http://www.nrcresearchpress.com/doi/abs/10.1139/cjfas-2012-0330. Anyway, this would double simulation design size, but wouldn't require rerunning the power analysis.
Simulation of data sets is done using the script Simulate_Growth.R
. Simulate_Growth.R
calls
Growth_Model.R
to run the simulations. All simulations are saved as .RData
files.
The model is then fit to each of the simulated data sets using Estimate_Simulations.R
. The
parameter estimates for each simulation are saved along with the simulated data
as .RData
files.
Table showing the simulation design, the number of fits that were positive definite Hessian (pdH), and the directory that contains the 200 simulated/estimated data sets.
WARNING: this takes a long time to run (over a week).
| Random effects | Power |pdH | Directory | | -------------- |:-----:|:----:|:---------:| | None | 50 | 196 | v0/50 | | | 100 | 195 | v0/100 | | | 250 | 194 | v0/250 | | | 500 | 191 | v0/500 | | k | 50 | 183 | v1/50 | | | 100 | 193 | v1/100 | | | 250 | 193 | v1/250 | | | 500 | 192 | v1/500 | | z | 50 | 60 | v2/50 | | | 100 | 86 | v2/100 | | | 250 | 172 | v2/250 | | | 500 | 189 | v2/500 | | k, z | 50 | 138 | v3/50 | | | 100 | 95 | v3/100 | | | 250 | 92 | v3/250 | | | 500 | 158 | v3/500 |
| Random effects | Model fitted | Power |pdH | Directory | | -------------- |:------------:|:-----:|:----:|:---------:| | None | 50 | 196 | v0/50 | | | 100 | 195 | v0/100 | | | 250 | 194 | v0/250 | | | 500 | 191 | v0/500 | | k | 50 | 183 | v1/50 | | | 100 | 193 | v1/100 | | | 250 | 193 | v1/250 | | | 500 | 192 | v1/500 | | z | 50 | 60 | v2/50 | | | 100 | 86 | v2/100 | | | 250 | 172 | v2/250 | | | 500 | 189 | v2/500 | | k, z | 50 | 138 | v3/50 | | | 100 | 95 | v3/100 | | | 250 | 92 | v3/250 | | | 500 | 158 | v3/500 |
Random effects: None
Now has sex specific gamma pars. Fixed psi = 0
Random effects: k
As above but with random effects on k.
Random effects: z
Random effects: k, z
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.