metaplus (metaplus), glht (multcomp), glmm (glmm), manova (stats), crq and crqs (quantreg)model_parameters() (and format_parameters()).df_method can now also be applied to GLMs, to allow calculation of confidence intervals based on Wald-approximation, not profiled confidence intervals. This speeds up computation of CIs for models fit to large data sets.select_parameters() for mixed models, and revised docs and associated vignette.threshold to be passed to efa_to_cfa() when the model is from factor_analysis().factor_analysis().model_parameters() for models with non-estimable parameters or statistics.model_parameters() for plm models with only one parameter.check_heterogeneity() in case no predictor would cause heterogeneity bias.robmixglm (robmixglm), betaor, betamfx, logitor, poissonirr, negbinirr, logitmfx, probitmfx, poissonmfx, negbinmfx (mfx), partial support emmGrid (emmeans)simulate_parameters() and simulate_model()print() method.verbose argument, to show or hide warnings and messages.equivalence_test() or CIs for standardized parameters from model_parameters() when standardization method was "posthoc").averaging (MuMIn), bayesx (R2BayesX), afex_aov (afex)check_heterogeneity() as a small helper to find variables that have a within- and between-effect related to a grouping variable (and thus, may result in heterogeneity bias, see this vignette).equivalence_test()rule argument, so equivalence testing can be based on different approaches.effect argument, to perform equivalence testing on random effects.p_values argument, to calculate p-values for the equivalence test.describe_distribution()ci and iterations arguments, to compute confidence intervals based on bootstrapping.iqr argument, to compute the interquartile range.SE column was removed.model_parameters()model_parameters() for Stan-models (brms, rstanarm) gains a group_level argument to show or hide parameters for group levels of random effects.model_parameters() with standardize = "basic" or standardize = "posthoc".model_parameters.merMod() no longer passes ... down to bootstrap-functions (i.e. when bootstrap = TRUE), as this might conflict with lme4::bootMer().MASS::polr() or ordinal::clm()), a Component column is added, indicating intercept categories ("alpha") and estimates ("beta").select-argument from print.parameters_model() now gets a "minimal"-option as shortcut to print coefficients, confidence intervals and p-values only.parameters_table() and print.parameters_model() now explicitly get arguments to define the digits for decimal places used in output.ci(), standard_error(), p_value() and model_parameters() for glmmTMB models now also works for dispersion models.equivalence_test() for mixed models.model_parameters.anova(..., eta_squared = "partial") when called with non-mixed models.model_parameters() for gam models.model_parameters().model_parameters() now also transforms standard errors when exponentiate = TRUE.model_parameters() for anova() from mixed models can now also compute effect sizes like eta squared.model_parameters() for aov() gains a type-argument to compute type-1, type-2 or type-3 sums of squares.model_parameters() for Bayesian models gains a standardize argument, to return standardized parameters from the posterior distribution.print() method for model_parameters() for nested aov() (repeated measurements).demean() should add attributes to indicate within- and between-effects. This is only relevant for the print()-method of model_parameters().model_parameters() for anova() from lmerTest models.model_bootstrap() was removed, please use bootstrap_model().parameters_bootstrap() was removed, please use bootstrap_parameters().model_simulate() was removed, please use simulate_model().parameters_simulate() was removed, please use simulate_parameters().parameters_selection() was removed, please use select_parameters().parameters_reduction() was removed, please use reduce_parameters().DDR(), ICA() and cmds() are no longer exported, as these were intended to be used internally by reduce_parameters() only.skewness() and kurtosis() always return a data frame.arima (stats), bife (bife), bcplm and zcpglm (cplm)model_parameters.brmsfit().model_parameters.merMod() when fitting REWB-Models (see demean()).model_parameters() (for linear mixed models) when df_method = "kenward".model_parameters() gets a p_adjust-argument, to adjust p-values for multiple comparisons.cluster_analysis() when method = "kmeans" and force = TRUE (factors now also work for kmeans-clustering).p_value_kenward(), se_kenward() etc. now give a warning when model was not fitted by REML.ci(), standard_error() and p_value() for lavaan and blavaan objects.standard_error() for brmsfit and stanreg objects.skewness(), kurtosis() and smoothness() get an iteration argument, to set the numbers of bootstrap replicates for computing standard errors.factor_analysis().demean() now additionally converts factors with more than 2 levels to dummy-variables (binary), to mimic panelr-behaviour.print()-method for model_parameters.befa().model_parameters() (for linear mixed models) with wrong order of degrees of freedom when df_method was different from default.model_parameters() (for linear mixed models) with accuracy of p-values when df_method = "kenward.model_parameters() with wrong test statistic for lmerModLmerTest models.format_parameters() (which is used to format output of model_parameters()) for factors, when variable name was also part of factor levels.degrees_of_freedem() for logistf-models, which unintentionally printed the complete model summary.model_parameters() for mlm models.random_parameters() for uncorrelated random effects.skewness() now uses a different method to calculate the skewness by default. Different methods can be selected using the type-argument.kurtosis() now uses a different method to calculate the skewness by default. Different methods can be selected using the type-argument.cglm (cglm), DirichletRegModel (DirichletReg)model_parameters(). This should now work for more models than before.ci.merMod() for method = "satterthwaite" and method = "kenward".select_parameters() for stanreg models, which was temporarily removed due to the CRAN removal of package projpred, is now re-implemented.dof_betwithin() to compute degrees of freedom based on a between-within approximation method (and related to that, p_value_*() and se_*() for this method were added as well).random_parameters() that returns information about the random effects such as variances, R2 or ICC.closest_component() as a small helper that returns the component index for each variable in a data frame that was used in principal_components().get_scores() as a small helper to extract scales and calculate sum scores from a principal component analysis (PCA, principal_components()).n_clusters() gets the option "M3C" for the package-argument, so you can try to determine the number of cluster by using the M3C::M3C() function.print()-method for model_parameters() gets a select-argument, to print only selected columns of the parameters table.model_parameters() for meta-analysis models has an improved print()-method for subgroups (see examples in ?model_parameters.rma).model_parameters() for mixed models gets a details-argument to additionally print information about the random effects.model_parameters() now accepts the df_method-argument for more (mixed) models.model_parameters() for meta-analysis models was renamed to "Overall".skewness() gets a type-argument, to compute different types of skewness.kurtosis() gets a type-argument, to compute different types of skewness.describe_distribution() now also works on data frames and gets a nicer print-method.model_parameters() when robust = TRUE, which could sometimes mess up order of the statistic column.model_parameters() with wrong df for lme-models.model_parameters.merMod() when df_method was not set to default.model_parameters.merMod() and model_parameters.gee() when robust = TRUE.format_p() when argument digits was "apa".model_parameters() for zeroinfl-models.model_parameters() was renamed from df_residuals to df_error for regression model objects, because these degrees of freedom actually were not always referring to residuals - we consider df_error as a more generic name.model_parameters() for standardized parameters (i.e. standardize is not NULL) only returns standardized coefficients, CI and standard errors (and not both, unstandardized and standardized values).format_ci() was removed and re-implemented in the insight package.model_bootstrap() was renamed to bootstrap_model(). model_bootstrap() will remain as alias.parameters_bootstrap() was renamed to bootstrap_parameters(). parameters_bootstrap() will remain as alias.model_simulate() was renamed to simulate_model(). model_simulate() will remain as alias.parameters_simulate() was renamed to simulate_parameters(). parameters_simulate() will remain as alias.parameters_selection() was renamed to select_parameters(). parameters_selection() will remain as alias.parameters_reduction() was renamed to reduce_parameters(). parameters_reduction() will remain as alias.vgam (VGAM), cgam, cgamm (cgam), complmrob (complmrob), cpglm, cpglmm (cplm), fixest (fixest), feglm (alpaca), glmx (glmx), glmmadmb (glmmADMB), mcmc (coda), mixor (mixor).model_parameters() now supports blavaan models (blavaan).clm2, clmm2 and stanmvreg models.psych::omega models.dof_satterthwaite() and dof_ml1() to compute degrees of freedom based on different approximation methods (and related to that, p_value_*() and se_*() for these methods were added as well).rescale_weights() to rescale design (probability or sampling) weights for use in multilevel-models without survey-design.standard_error_robust() or ci_robust()) can now also compute cluster-robust variance-covariance matrices, using the clubSandwich package.model_parameters() gets a robust-argument, to compute robust standard errors, and confidence intervals and p-values based on robust standard errors.p_method and ci_method in model_parameters.merMod() were replaced by a single argument df_method.model_parameters.principal() includes a MSA column for objects from principal_components().model_parameters() with non-typical ordering of coefficients for mixed models.rlmerMod.model_parameters.BFBayesFactor().Parts of the parameter package are restructured and functions focussing on anything related to effect sizes are now re-implemented in a new package, effectsize. In details, following breaking changes have been made:
cohens_f(), eta_squared() etc.) have been removed and are now re-implemented in the effectsize-package.d_to_odds() etc.) have been removed and are now re-implemented in the effectsize-package.standardize() and normalize() (and hence, also parameters_standardize()) have been removed ;-( and are now re-implemented in the effectsize-package.aareg (survival), bracl, brmultinom (brglm2), rma (metafor) and multinom (nnet) to various functions.model_parameters() for kmeans.p_value(), ci(), standard_error() and model_parameters() now support flexsurvreg models (from package flexsurv).degrees_of_freedom() to get DoFs.p_value_robust(), ci_robust() and standard_error_robust() to compute robust standard errors, and p-values or confidence intervals based on robust standard errors.demean() to calculate de-meaned and group-meaned variables (centering within groups, for panel-data regression).n_parameters() to get number of parameters.n_clusters() to determine the number of clusters to extract.cluster_analysis() to return group indices based on cluster analysis.cluster_discrimination() to determine the goodness of classification of cluster groups.check_clusterstructure() to check the suitability of data for clustering.check_multimodal() to check if a distribution is unimodal or multimodal.plot()-methods for principal_components().n_factors() (Finch, 2019)standard_error() for mixed models gets an effects argument, to return standard errors for random effects.method-argument for ci() gets a new option, "robust", to compute confidence intervals based on robust standard errors. Furthermore, ci_wald() gets a robust-argument to do the same.format_p() gets a digits-argument to set the amount of digits for p-values.model_parameters() now accepts (non-documented) arguments digits, ci_digits and p_digits to change the amount and style of formatting values. See examples in model_parameters.default().print() method for model_parameters() when used with Bayesian models.n_clusters().model_parameters() were denoted as nested interaction when one of the interaction terms was surrounded by a function, e.g. as.factor(), log() or I().parameters_type() when a parameter occured multiple times in a model.model_parameters() for non-estimable GLMs.p_value() for MASS::rlm models.reshape_loadings() when converting loadings from wide to long and back again.format_value() and format_table() have been removed and are now re-implemented in the insight package.parameters() is an alias for model_parameters().p_value(), ci(), standard_error(), standardize() and model_parameters() now support many more model objects, including mixed models from packages nlme, glmmTMB or GLMMadaptive, zero-inflated models from package pscl or other modelling packages. Along with these changes, functions for specific model objects with zero-inflated component get a component-argument to return the requested values for the complete model, the conditional (count) component or the zero-inflation component from the model only.parameters_simulate() and model_simulate(), as computational faster alternatives to parameters_bootstrap() and model_bootstrap().data_partition() to partition data into a test and a training set.standardize_names() to standardize column names from data frames, in particular objects returned from model_parameters().se_kenward() to calculate approximated standard errors for model parameters, based on the Kenward-Roger (1997) approach.format_value() and format_ci() get a width-argument to set the minimum length of the returned formatted string.format_ci() gets a bracket-argument include or remove brackets around the ci-values.eta_squared(), omega_squared(), epsilon_squared() and cohens_f() now support more model objects.print()-method for model_parameters() now better aligns confidence intervals and p-values.normalize() gets a include_bounds-argument, to compress normalized variables so they do not contain zeros or ones.method-argument for ci.merMod() can now also be "kenward" to compute confidence intervals with degrees of freedom based on the Kenward-Roger (1997) approach.p_value_kenward().paramerers_standardize() resp. standardize() for model objects now no longer standardizes log() terms, count or ratio response variables, or variables of class Surv and AsIs.NEWS.md file to track changes to the package