Learn R Programming

prw2ppl Readme

Chris Aberson April 2, 2021

pwr2ppl

pwr2ppl contains protocols for running a wide range of power analyses. Analyses range from simple approaches such as t-test and correlations to multifactor ANOVA (between and within subjects, linear mixed model approaches) and regression-based approaches (basic multiple regression, moderated regression, mediation, and logistic regression).

These protocols accompany the book Applied Power Analysis for the Behavioral Sciences (2nd ed.)

The book gives lots of illustrative examples of using the code but is not necessary for using the package.

Prerequisites

I built this under R 4.0.2

Authors

Chris Aberson chrisaberson

License

This project is licensed under the MIT License.

Copy Link

Version

Install

install.packages('pwr2ppl')

Monthly Downloads

310

Version

0.5.0

License

GPL (>= 2)

Maintainer

Chris Aberson

Last Published

September 6th, 2022

Functions in pwr2ppl (0.5.0)

Compute Multiple Regression shortcuts with three predictors (will expand to handle two to five) Requires correlations between all variables as sample size. Means and sds are option. Also computes Power(All)

Compute Precision Analyses for R-Squared This approach simply loops a function from MBESS

Compute power for R2 change in Multiple Regression (up to three predictors) Requires correlations between all variables as sample size. Means, sds, and alpha are option. Also computes Power(All) Example code below for three predictors. Expand as needed for four or five

Compute power for a One Factor Between Subjects ANOVA with four levels Takes means, sds, and sample sizes for each group

Compute power for a One Factor ANOVA with four levels. Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Compute power for a One Factor ANOVA with three levels. Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Compute Multiple Regression shortcuts with three predictors for Ind Coefficients Requires correlations between all variables as sample size. Means and sds are option. Also computes Power(All)

Compute Power for One or Two Factor ANCOVA with a single covariate Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user Factor A can have up to four levels, Factor B, if used, can only be two

Compute power for Multiple Regression with Up to Five Predictors Requires correlations between all variables as sample size. Means, sds, and alpha are option. Also computes Power(All)

Compute power for a One Factor ANOVA with three levels and contrasts. Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Compute Power for Comparing Two Dependent Correlations, No Variables in Common Takes correlations and range of values. First variable in each pair is termed predictor, second is DV

Compute power for a Two by Two Between Subjects ANOVA. Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Compute power for Simple Effects in a Two by Two Between Subjects ANOVA with two levels for each factor. Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Power for Comparing Dependent Coefficients in Multiple Regression with Two or Three Predictors Requires correlations between all variables as sample size. Means, sds, and alpha are option. Also computes Power(All)

Compute Power for Comparing Two Dependent Correlations, One Variable in Common Takes correlations and range of values

Power for Comparing Independent R2 in Multiple Regression with Two or Three Predictors Requires correlations between all variables as sample size. Means, sds, and alpha are option. Also computes Power(All)

Compute power for Pearson's Correlation Takes correlation and range of values

Compute Precision Analyses for Standardized Mean Differences

Compute power for a Two Factor Within Subjects Using Linear Mixed Models with up to two by four levels. Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Compute power for a Two Factor Within Subjects Using Linear Mixed Models with up to two by four levels. Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Compute power for a One Factor Within Subjects Linear Mixed Model with up to four levels. Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Compute Power for Mediated (Indirect) Effects Requires correlations between all variables as sample size. This approach calculates power for the Sobel test. The medjs function calculates power based on joint significance (recommended)

Compute power for an Independent Samples t-test Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Compute Precision Analyses for Mean Differences

Compute Power for Mediated (Indirect) Effects Using Joint Significance Requires correlations between all variables as sample size. This is the recommended approach for determining power

Compute power for a One Factor Within Subjects and One Factor Between LMM with up to two by four levels (within). Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Compute power for a One Factor Within Subjects LMM Trends with up to four levels. Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Power for Comparing Independent Coefficients in Multiple Regression with Two or Three Predictors Requires correlations between all variables as sample size. Means, sds, and alpha are option. Also computes Power(All)

Compute Power for Comparing Two Independent Correlations Takes correlations and range of values

Compute Power for Mediated (Indirect) Effects Using Joint Significance Requires paths for all effects (and if 2 mediators, correlation) Standard deviations/variances set to 1.0 so paths are technically standardized

Compute Precision Analyses for Correlations This approach simply loops a function from MBESS

Compute power for Tests of Two Independent Proportions Takes phi, degrees of freedom, and a range of sample sizes. Alpha is .05 by default, alternative values may be entered by user This test uses what is sometimes called the chi-square test for comparing proportions

Compute power for a Paired t-test Takes means, sd, and sample sizes. Alpha is .05 by default, alternative values may be entered by user. correlation (r) defaults to .50.

Compute power for a single sample proportion test Takes phi, degrees of freedom, and a range of sample sizes. Alpha is .05 by default, alternative values may be entered by user

Compute Power for Regression Interaction (Correlation/Coefficient Approach)

Compute Power for Regression Interaction (R2 Change Approach)

Compute Power for Conditional Process Model 14 Joint Significance Requires correlations between all variables as sample size. This is the recommended approach for determining power

Compute Power for Model 7 Conditional Processes Using Joint Significance Requires correlations between all variables as sample size Several values default to zero if no value provided This is the recommended approach for determining power

medserial_paths

Compute Power for Serial Mediation Effects Requires correlations between all variables as sample size. This approach calculates power for the serial mediation using joint significance (recommended) and path coefficients

Compute power for a t test using d statistic Takes d, sample size range, type of test, and tails.

Compute power for a One Factor Within Subjects and One Factor Between ANOVA with up to two by four levels (within). Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Compute power for a Two Factor Within Subjects ANOVA with up to two by four levels. Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Compute power for Simple Effects in Two Factor Within Subjects ANOVA with up to two by four levels. Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Compute power for a One Factor Within Subjects Trends with up to four levels. Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Compute power for a One Factor Within Subjects ANOVA with up to four levels. Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Compute Power for Serial Mediation Effects Requires correlations between all variables as sample size. This approach calculates power for the serial mediation using joint significance (recommended)

Compute power for an Chi Square 2x3 Takes proportions for each group. Alpha is .05 by default, alternative values may be entered by user

Compute power for Chi Square Based on Effect Size Takes phi, degrees of freedom, and a range of sample sizes. Alpha is .05 by default, alternative values may be entered by user

Compute power for an Chi Square Goodness of Fit Takes proportions for up to six group. Alpha is .05 by default, alternative values may be entered by user

Compute power for Multiple Regression with Violated assumptions (Beta)

Compute power for a One Factor MANOVA with up to two levels and up to four measures. Takes means, sds, and sample sizes for each group. Alpha is .05 by default, alternative values may be entered by user

Compute Power for Logistic Regression with a Single Categorical Predictor

Assumptions_resample

Compute power for Multiple Regression with Violated assumptions using Resamples

Compute Power for Logistic Regression with Continuous Predictors

Compute power for an Chi Square 2x2 Takes proportions for each group. Alpha is .05 by default, alternative values may be entered by user

Compute power for Multiple Regression with up to Five Predictors Example code below for three predictors. Expand as needed for four or five