Expected cluster size precision inconsistent #109
Description
When detections are clusters, such that there is a column labelled size but size is not used as a covariate in the detection function. My understanding is that individual abundance is computed using the mean group size and E(s) is simply the average of groups in the data.
Something odd appeared in this situation when consulting with someone from the list. I created a Minimum Reproducible Example using the ClusterSolution dataset shipped with the Distance package.
I do not understand the reason for the disagreement in se(E(s)) between the reported average group size (in Summary) and the se(E(s)) shown in Clusters.
library(Distance)
data("ClusterExercise")
fake <- ds(data=ClusterExercise, truncation = 1.5,
key="hr", formula=~Region.Label)
result
> fake$dht$individuals$summary[, c(1, 10, 11)]
Region mean.size se.mean
1 North 2.326531 0.3417987
2 South 2.153846 0.2905123
3 Total 2.250000 0.2287180
> fake$dht$Expected.S
Region Expected.S se.Expected.S
1 North 2.326531 0.5357964
2 South 2.153846 0.3009339
3 Total 2.271200 0.3759673
Discrepency of standard error of E(s) persists when Region.Label is not included as a covariate:
nocovariate <- ds(data=ClusterExercise, truncation = 1.5, key="hr")
nocovariate$dht$individuals$summary[, c(1, 10, 11)]
nocovariate$dht$Expected.S
result
> nocovariate$dht$individuals$summary[, c(1, 10, 11)]
Region mean.size se.mean
1 North 2.326531 0.3417987
2 South 2.153846 0.2905123
3 Total 2.250000 0.2287180
> nocovariate$dht$Expected.S
Region Expected.S se.Expected.S
1 North 2.326531 0.5357964
2 South 2.153846 0.3009339
3 Total 2.286689 0.4173391
is disagreement in precision to be expected?