Added coef() methods to extract participation model
coefficients from pw_fit objects and pseudo-weighted means
or category prevalence from pwmean objects.
Added confint() for pwmean objects,
with a choice of confidence level and estimates to include.
Solver diagnostics in print(),
summary(), and verbose output now report why the solver
stopped and whether the final equation residuals satisfy
ftol.
Fixed standard errors and confidence intervals when reference
survey columns collide with internal names such as
h_x.
Fixed cumulative precalibration when auxiliary variables are
shared by only some reference surveys. It also handles factor levels
containing spaces or hyphens and auxiliary variables named
.w.
est_pw() now rejects mismatched categories for
auxiliary variables stored as character in both samples, even when the
category counts match.
pwmean() now treats empty or whitespace-only strings
in character domains as missing before applying
na.action.
Added guidance on standardizing continuous auxiliary variables and harmonizing categorical coding across samples.
Corrected the sp1_bootstrap help example to use
combined.weights = TRUE because the replicate weights
already include the main survey weight.