Dependencies
- medfit is now available on CRAN (v0.2.0/0.2.1).
Removed
Remotes: data-wise/medfit; medfit now resolves
from CRAN. Pinned Suggests: medfit (>= 0.2.0).
ci() for medfit
MediationData/SerialMediationData now extracts
the path covariance by parameter name and uses the full covariance
matrix (including off-diagonals) for both simple and serial mediation.
Removed the previous positional/value-matching heuristics and the
independence/diagonal fallbacks, which could silently produce incorrect
intervals; unresolved path labels now raise an informative error.- The serial-mediation pipeline is now verified end-to-end against
medfit’s released serial extractor (medfit >= 0.2.0):
medfit::extract_mediation() produces a
SerialMediationData for both lavaan (ordered
mediator vector) and lm/glm (mediator_models)
chains, and ci() consumes it via the documented
d1, d2, ... path-name contract. Added integration tests
that fit a real model, extract, and run ci() end-to-end
(skipped when medfit < 0.2.0).
Workflow
Optimization & Standardization (2025-12-05)
- Optimized R-CMD-check workflow: Reduced from 5 to 3
platforms (macOS, Windows, Ubuntu release)
- Removed failing platforms: Dropped
ubuntu-latest (devel) and
ubuntu-latest (oldrel-1) configurations
- Runtime improvement: Check workflows now complete
in ~2 minutes (down from 6+ minutes, 67% faster)
- Build optimization: Added
--ignore-vignettes flag to skip vignette validation during
CI checks
Bug Fixes
- Fixed pkgdown build errors:
- Removed non-existent
reexports topic from
_pkgdown.yml reference index
- Added missing
tidy topic to _pkgdown.yml
reference index
- pkgdown workflow now passes successfully
Standardization
- Badge order: Added R-hub badge to README,
standardized order to match mediationverse ecosystem: CRAN, Lifecycle,
R-CMD-check, Website, R-hub, Codecov
- .Rbuildignore: Added
^STATUS\.md$ to
match ecosystem standards
- Website configuration: Already compliant with
mediationverse standards:
- Academic blue color scheme (#0054AD)
- Consistent fonts (Inter, Montserrat, Fira Code)
- Ecosystem and Status dropdown menus
Documentation
- Updated README badge order for consistency across ecosystem
- All workflows now passing on main branch
Workflow
Optimization & Standardization (2025-12-05)
- Optimized R-CMD-check workflow: Reduced from 5 to 3
platforms (macOS, Windows, Ubuntu release)
- Removed failing platforms: Dropped
ubuntu-latest (devel) and
ubuntu-latest (oldrel-1) configurations
- Runtime improvement: Check workflows now complete
in ~2 minutes (down from 6+ minutes, 67% faster)
- Build optimization: Added
--ignore-vignettes flag to skip vignette validation during
CI checks
Bug Fixes
- Fixed pkgdown build errors:
- Removed non-existent
reexports topic from
_pkgdown.yml reference index
- Added missing
tidy topic to _pkgdown.yml
reference index
- pkgdown workflow now passes successfully
Standardization
- Badge order: Added R-hub badge to README,
standardized order to match mediationverse ecosystem: CRAN, Lifecycle,
R-CMD-check, Website, R-hub, Codecov
- .Rbuildignore: Added
^STATUS\.md$ to
match ecosystem standards
- Website configuration: Already compliant with
mediationverse standards:
- Academic blue color scheme (#0054AD)
- Consistent fonts (Inter, Montserrat, Fira Code)
- Ecosystem and Status dropdown menus
Documentation
- Updated README badge order for consistency across ecosystem
- All workflows now passing on main branch
Dependency Reduction
- Removed dependencies:
e1071,
modelr, and generics packages are no longer
required.
- Internal implementations: Added internal
.skewness(), .kurtosis(), and
.resample_bootstrap() functions to replace external
dependencies.
- Own tidy generic: Package now defines its own
tidy() generic instead of re-exporting from
generics.
- Total Imports reduced from 10 to 7 packages.
CI/CD Improvements
- Added
R-CMD-check.yaml workflow for multi-platform
testing (macOS, Windows, Ubuntu with R devel/release/oldrel).
- Added
test-coverage.yaml workflow for Codecov
integration.
- Added
VignetteBuilder: knitr to DESCRIPTION for proper
vignette building.
- Standardized README.md to mediationverse format.
Ecosystem Notes
- Part of the mediationverse ecosystem for mediation
analysis in R.
- Integration with medfit planned for v1.5.0 (Q1 2026).
- See Ecosystem
Coordination for version compatibility and development
guidelines.
- See Development
Roadmap for timeline.
Major Changes
S7 Core Architecture
(Complete Refactoring)
- Complete architectural redesign: S7 is now the
computational core, with legacy functions as thin wrappers.
- Zero breaking changes: All existing code continues
to work exactly as before (100% backward compatible).
- Enhanced extensibility: New computational
architecture makes adding CI methods and features straightforward.
S7 OOP Classes
- Introduced S7 object-oriented programming for modern, type-safe
class definitions.
- New S7 Classes:
ProductNormal: Represents the distribution of the
product of normal random variables.MBCOResult: Encapsulates results from MBCO tests.
- New S7 Generics:
cdf(),
dist_quantile(), ci(), print(),
summary(), show().
- All S7 classes include comprehensive display methods
(
print, summary, show) for
meaningful output.
N-Variable Product Support
(NEW)
- Breaking limitation removed: Can now compute
confidence intervals for products of 3 or more variables.
- Works with Monte Carlo method (
type = "mc").
- Example:
ProductNormal(mu = c(0.3, 0.4, 0.5), Sigma = diag(3)) for
three-way products.
- Previous versions limited to 2-variable products only.
Internal Computational Core
- New internal S7 functions:
.compute_ci_dop(), .compute_ci_mc(),
.compute_ci_asymp(), .compute_cdf_dop(),
.compute_cdf_mc(), .compute_quantile_dop(),
.compute_quantile_mc().
- All core algorithms now work with
ProductNormal objects
natively.
- Legacy functions (
medci(), pprodnormal(),
qprodnormal()) convert parameters and dispatch to S7
core.
- Identical numerical results to previous versions (validated with 201
tests).
Enhanced ci() Function
- Migrated
ci() to S7 generic with methods for:
numeric: Legacy support for direct
coefficient/covariance input.lavaan: Automatic extraction and CI computation for
indirect effects.ProductNormal: Native S7 class support.
- Auto-detection: Automatically identifies product
parameters (e.g.,
ab := a*b) in lavaan
models.
- Renamed first argument to
mu for better legacy
compatibility.
MBCO Function Updates
- Migrated
mbco() to S7 generic with method for
MxModel objects.
- Returns
MBCOResult S7 objects with properties:
statistic, df, p_value,
type.
- Legacy Compatibility: Implemented S3 methods
(
$, [[, names) for
backward-compatible list-style access (e.g., result$chisq,
result$p).
- Display methods show significance indicators (, ,
, ns) and interpretation.
- Comprehensive input validation using
checkmate package
across all functions.
- Type checking with
match.arg() for categorical
arguments.
- Improved error messages for invalid inputs.
Display Methods
- All S7 classes have meaningful
print(),
summary(), and show() methods.
ProductNormal displays: distribution type, means,
covariance matrix, SDs, and correlations.MBCOResult displays: test type, statistics, p-values
with significance levels, and interpretation.
Namespace Improvements
- Zero masking warnings: Package now loads cleanly
with no namespace conflicts.
- Renamed
quantile() generic to
dist_quantile() to avoid masking
stats::quantile().
- S7 methods for
print(), summary(),
show() now properly register with base generics.
- Added S4 compatibility via
S7::S4_register() for formal
generic support.
Package Cleanup
- Removed unused dependencies (
doParallel,
foreach, methods).
- Fixed all documentation warnings and notes.
- R CMD check: Now passes with 0 errors, 0 warnings,
0 notes.
- Added 40 new S7 prototype tests + 19 display method tests (total:
201 tests, 100% pass rate).
Bug Fixes
- Fixed argument dispatch issues between
alpha and
level parameters.
- Resolved S7 generic conflicts with base R functions (print, summary,
quantile).
- Fixed recursive dispatch issues in display methods.
- Added case-insensitive handling for
type parameter
(accepts “MC”, “mc”, “DOP”, “dop”, etc.).
- Fixed return value structure for
ci() method (now
returns list with CI, Estimate,
SE components).
- Updated functions and package documentation.
- Updated package dependencies.
- Updated
mbco function documentation.
- Implemented
checkmate for robust input validation.
- Created GitHub pages for package documentation.
- Refactored package structure for better organization.
- Refactored package validation for better input validation.
- Added tests for package functions.
- Fixed an issue where the type=“all” in the medci function would
generate incorrect labels for CIs.
- Fixed issues to meet CRAN submission requirements.
New parametric and non-parametric MBCO test, using
mbco function. Fixed an error with printing medci,
‘(1-alpha)% CI’
- Fixed the issue with ‘medic’ function that printed ‘(1-alpha/2)% CI’
instead of ‘(1-alpha)% CI’
‘medci’ function output structure is changed. For each ‘type’, it
produces a ‘list’ of ‘(1-alpha/2)% CI’, ‘Estimate’, and ‘SE’. See
help(ci) for more information.
Two functions ‘pMC’ and ‘qMC’ are added.
for medci(), type=‘prodclin’ is replaced with ‘dop’.
- New ‘ci’ function produces CIs for any well-defined function of
parameter estimates in an SEM and MSEM using the Monte Carlo or
asymptotic normal theory with the multivariate delta standard error
method.