NEWS.md
In Superpower: Simulation-Based Power Analysis for Factorial Designs

NEWS

Updated Tue Dec 07 2021

Added power_oneway_ancova & power_con_ancova to allow for a basic power analysis of an analysis of covariance (ANCOVA) for one-way, between group designs.
Added ANCOVA_analytic and ANCOVA_contrast which allow for power analyses for factorial designs and user specified contrasts.
Added the label_list argument to ANOVA_design and ANCOVA_analytic functions. Now labels can be assigned to factors and levels in a more sane fashion using named lists.

Added morey_plot functions.
- Plot the effect size (x-axis) at different sample sizes (facets) and at different alpha levels (color).
- These plots are helpful in determining the sensitivity of statistical tests (t-test and F-test) across a range of effect sizes.
Added confint method for ANOVA_power produced objects
- Calculates confidence level for binomial proportion (# of results that are below alpha level) confidence intervals (Wilson, 1927)).
Minor changes to Shiny apps to fix glitches.

Added ANOVA_exact2 function as an extension of ANOVA_exact
- Now functional across all sample sizes but does not return a dataframe of afex aov object
liberal_lambda argument added: allows users to specify the type of lambda calculations
- When liberal_lambda = TRUE; lambda = cohen_f^2 * (num_df + den_df + 1)
- When liberal_lambda = FALSE; lambda = cohen_f^2 * den_df
Optimal alpha functions from JustifieR package added
ANOVA_compromise function added which allows a compromise power analysis to be performed for all comparisons in a design
ANOVA_design now returns as a class “design_aov” with specific print and plot methods see ?design_aov-methods
- generate_cor_matrix function is now a non-exported function within the package (no longer contained within ANOVA_design)
All simulation functions ANOVA_power, ANOVA_exact, and ANOVA_exact2 now returns as a class “sim_result” with specific print and plot methods see ?sim_result-methods
plot_power now has reduced sample size limitations -Option to use ANOVA_exact2 (exact2 argument) improves functionality (not limited to product of factors)
Updated vignettes to include updated information on functions
- New vignette “Introduction to Justifying Alpha Levels”
New Shiny App: justify
- Creates a UI for utilizing the ANOVA_compromise function via Shiny

Superpower_options(“plot”) is now set to TRUE. Plots will, by default, be printed -Easily reset with Superpower_options(plot = FALSE)
plot_power has new features -Plots now show desired power -min_n is now limited; smallest min_n allowed is equal to the product of the design (e.g., ’2b*2b’ has a smallest min_n of 4)
Small update to plot_power to fix minor error in original code -Error resulted in power estimates being \~0.1-0.5% off actual power estimate

Unequal sample size in the design is now permitted -Limited to the ANOVA_design and ANOVA_power functions
Added estimated marginal means comparisons using emmeans R package.
- emm = TRUE in the ANOVA_power, ANOVA_exact, and plot_power will result in emmeans being calculated
- Default is all pairwise comparisons but this can be modified with contrast_type and emm_comp options
Added global options
- Options that have crossover between functions can now be set globally for the package
- Includes: verbose, emm, emm_model, contrast_type, alpha_level, and plot
- These global options can be seen with Superpower_options()
Updated Shiny Apps
- Unequal n allowed for ANOVA_power
- Added numeric input for alpha level (no longer slider)
- Now includes emmeans options
- kableExtra, emmeans, magrittr, and dplyr packages now needed to knit markdown file in app.