Passing start values and bounds

[LHMarshall]

Things to check and test when passing start values and bounds

• How does changing scaling of adjustment parameters from SIGMA to W affect the start value format and ordering?
• Check the factor ordering.
• What scale are they passed in and what scale does MCDS need them on?
• Why is size field being renamed to Cluster Size?
• In what situation do the covariates affect the shape parameter? Can the same analysis be done in MCDS?
• Is a uniform model with covariates allowed? Scale parameter always 1. Usually adjustments are not allowed with covariates but it has happened in some cases.

[LHMarshall]

Cluster Size appears to be included twice in the datasets passed to MCDS sometimes

library(Distance)
data(book.tee.data)
tee.data <- subset(book.tee.data$book.tee.dataframe, observer==1)

ds.model <- ds(tee.data,
               truncation = 4,
               formula = ~size+sex+exposure,
               initial_values = list(scale = c(-0.13, 0.05, 0.7, 0.3)),
               optimizer = "MCDS")