nparLD 2.3.1
- Fixed the ordering of output rows and cell-effect plots for numeric
factor levels. Output tables and plots now preserve the canonical
factor-level order instead of applying lexicographic ordering to
character representations of factor levels.
nparLD 2.3.0
Major redesign
- Added a general formula interface for crossed factorial longitudinal
designs.
- Added inference for hypotheses in marginal distribution functions
(
hypothesis = "H0F") and unweighted relative marginal
effects (hypothesis = "H0p").
- Added support for missing observations.
- Added support for dependent replicate measurements via the
replicate argument.
- Added subject-level and observation-level cell weighting for
relative marginal effects.
- Added multiple contrast procedures and simultaneous confidence
intervals.
- Added rank- and pseudo-rank-based inference through a unified
interface.
- Added the argument
covariance, which optionally
includes the estimated covariance matrix in the output.
- Added updated documentation, examples, README, vignette, and example
datasets.
- Added the
brdu data set for illustrating dependent
replicate measurements.
- Added term-specific plots of factor-information estimates and
confidence intervals via
plot(fit, term = ...).
Interface changes
- Factor ordering is intended to follow ordinary R factor levels
before calling
nparLD().
- Earlier design-specific ordering arguments are no longer part of the
main interface.
- Classical designs such as
LD-F1, F1-LD-F1,
LD-F2, F1-LD-F2, and F2-LD-F1 are
now specified through the formula interface.
nparLD 2.3.2
- Added accessor functions for fitted-model results.
- Revised print and summary methods.
- Improved p-value formatting.
- Added accessors for covariance, contrast matrices, and permutation
results.
- Corrected Monte Carlo p-value calculation for paired permutation
tests.
- Improved documentation, examples, and plotting reproducibility.
nparLD 2.3.3
- Revised
print() and summary() methods for
fitted nparLD objects. print() now provides a
concise model overview and the ANOVA-type statistic, whereas
summary() provides detailed effect estimates and
inferential results.
- Multiple contrast, factor-specific, permutation, and other detailed
results remain available through
summary() and dedicated
accessor functions.
- Updated documentation and examples accordingly.