In avallecam/powder: A Tidy Extension for Power Analysis

knitr::opts_chunk$set(
  collapse = TRUE,
  warning = FALSE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

powder

Author: Andree Valle Campos
License: MIT

Overview

The goal of powder is to extend power and sample size calculations:

• for more than one set of parameters, and

• create tidy output tables and plots from them.

Installation

You can install the developing version of powder using:

if(!require("devtools")) install.packages("devtools")
devtools::install_github("avallecam/powder")

Quick example

This is a basic example which shows you how to solve a common problem:

library(tidyverse)
library(broom)
library(pwr)
library(powder)

#example("pwr.2p.test")
pwr.2p.test(h=0.3,n=80,sig.level=0.05,alternative="greater")

pwr_grid(h=0.3,n=seq(80,90,5),sig.level=0.05,alternative="greater") %>% 
  pwr_tidy(test_function = pwr.2p.test)

pwr_grid(h=seq(0.3,0.5,0.1),n=seq(80,90,5),sig.level=0.05,alternative="greater") %>% 
  pwr_tidy(test_function = pwr.2p.test)

Core structure

powder consist of three main functions:

• pwr_grid: creates a tibble from all combination of input parameters using tidyr::expand_grid.

• pwr_tidy: creates a broom::tidy() output from the calculations of all input parameters using purrr::pmap.

• pwr_plot: create a ggplot with input parameters and calculated value (sample size, power or effect size)

More examples

One set of parameters

diff <- 10
sigma <- 10
delta <- diff/sigma

pwr.t.test(d = delta, power = 0.8, type = "one.sample")
# sample - power plot
#pwr.t.test(d = delta, power = 0.8, type = "one.sample") %>% plot()

More than one set of parameters

sample size

# stata
# power onemean 20, diff(10 20 30) sd(10 12.5 15)
eg1 <- pwr_grid(n = NULL,
                diff = c(10,20,30),
                sigma = c(10,12.5,15),
                d = NULL,
                sig.level = 0.05,
                power = 0.8,
                type = "one.sample",
                alternative = "two.sided")

#create tidytable
eg1 %>% pwr_tidy(test_function = pwr.t.test)

#create ggplot
eg1 %>%
  pwr_tidy(test_function = pwr.t.test) %>%
  pwr_plot(x = diff,y = n,group = sigma)

power

# stata
# power onemean 20, diff(5 (5) 15) sd(10) n(10 (2) 20)
eg2 <- pwr_grid(n = seq(from = 10,to = 20,by = 2),
                diff = seq(from = 5,to = 15,by = 5),
                sigma = 10,
                d = NULL,
                sig.level = 0.05,
                #power = 0.8,
                type = "one.sample",
                alternative = "two.sided")

#create tidytable
eg2 %>% pwr_tidy(test_function = pwr.t.test)

#create ggplot
eg2 %>%
  pwr_tidy(test_function = pwr.t.test) %>%
  pwr_plot(x = n,y = power,group=diff)

References

Stephane Champely (2018). pwr: Basic Functions for Power Analysis. R package version 1.2-2. https://CRAN.R-project.org/package=pwr

Citation

citation("powder")

avallecam/powder documentation built on March 26, 2021, 4:41 p.m.