stepwedgepower 0.1.1
Generalized interface
- Generalized the simulation and power interface from
provider/specialty terminology to cluster/sequence terminology.
- Simulated data now carry generic columns (
cluster_id,
sequence, period, intervention,
n, events) alongside the legacy 0.1.0 columns,
and run_stepwedge_analysis() accepts either set.
- Added
icc_to_cluster_sd() and
cluster_sd_to_icc(), and an icc argument as an
alternative parameterisation of cluster_sd.
Monte Carlo uncertainty
and diagnostics
estimate_power() and
estimate_type1_error() return a
stepwedge_power object reporting the estimate, an
exact (Clopper-Pearson) Monte Carlo interval, the Monte
Carlo standard error, fit-failure rate, and singular-fit rate. The exact
interval retains nominal coverage at the boundary, where a Wald interval
collapses to zero width.- Added a
print.stepwedge_power() method.
Deprecation handling
- Legacy argument names are retained and mapped to their canonical
counterparts. Deprecation warnings are now resolved once per call rather
than once per simulated replicate, so a long run emits a single warning
instead of one per iteration.
- Supplying both a canonical argument and its deprecated alias (or
both
cluster_sd and icc) is now an error
rather than being resolved silently.
estimate_type1_error() has an explicit argument list
rather than ....
Other
run_stepwedge_analysis() builds its model formula from
column names instead of transform(), avoiding “no visible
binding for global variable” notes.- Updated the package title in response to CRAN feedback.