Package ‘powerAnalysis’

February 2, 2017
Title Power Analysis in Experimental Design
Description Basic functions for power analysis and effect size calculation.
Version 0.2.1
Date 2017-02-02
Author Felix Yanhui Fan <[email protected]>
Maintainer Felix Yanhui Fan <[email protected]>
License GPL (>= 2)
RoxygenNote 5.0.1
NeedsCompilation no
Repository CRAN
Date/Publication 2017-02-02 17:38:08

R topics documented:
ES.anova.oneway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
ES.chisq.assoc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
ES.chisq.gof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
ES.proportions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
ES.t.one . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
ES.t.paired . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
ES.t.two . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
power.anova.oneway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
power.chisq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
power.plot.chisq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
power.proportions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
power.t . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Index 14

1
2 ES.chisq.assoc

ES.anova.oneway Calculating effect size (Cohen’s f) of one-way anova for means with
equal observations in each group

Description

Calculating effect size (Cohen’s f) of one-way anova for means with equal observations in each
group

Usage

ES.anova.oneway(data = NULL, sst = NULL, ssb = NULL)

Arguments

data a matrix or data frame

sst total sum of squares
ssb sum of squares between groups

Examples
set.seed(9);x=rnorm(50);y=rnorm(50)
z=rnorm(50);d=data.frame(x,y,z)
ES.anova.oneway(data=d)

ES.anova.oneway(sst=50,ssb=1)

ES.chisq.assoc Compute effect size of chi-squared test of association

Description

Compute effect size of chi-squared test of association

Usage

ES.chisq.assoc(ct = NULL, chisq = NULL, p = NULL, n = NULL, df = NULL,

mindf = NULL)
ES.chisq.gof 3

Arguments

ct a m x n Contingency Table (matrix with m rows and n columes)

chisq the value the chi-squared test statistic
p p value for the chi-squared test
n total number of observations (sample size)
df degree of freedom (e.g., df=(m-1)*(n-1))
mindf the degrees of freedom for the variable with the smaller number of levels, if m
> n, mindf=n-1, otherwise, mindf=m-1

See Also

ES.chisq.gof

Examples
counts <- matrix(c(225,125,85,95),nrow=2,byrow=TRUE)
ES.chisq.assoc(ct=counts)

case <- c(225,85,100)

control <- c(125,95,125)
counts <- cbind(case,control)
ES.chisq.assoc(ct=counts)

p1 <- c(225,85,100)
p2 <- c(125,95,125)
p3 <- c(175,90,113)
counts <- cbind(p1,p2,p3)
ES.chisq.assoc(ct=counts)

ES.chisq.assoc(chisq=13.561,n=530,df=1,mindf=1)

ES.chisq.assoc(p=0.000231,n=530,df=1,mindf=1)

ES.chisq.gof Compute effect size of chi-squared test of goodness of fit

Description

Compute effect size of chi-squared test of goodness of fit

Usage

ES.chisq.gof(p1 = NULL, p0 = rep(1/length(p1), length(p1)))

4 ES.proportions

Arguments
p1 a vector of frequencies or probabilities (alternative hypothesis). Frequencies
will be rescaled to probabilities automatically. An error is given if any entry of
p1 is negative.
p0 a vector of frequencies or probabilities of the same length of p1 (null hypoth-
esis). Frequencies will be rescaled to probabilities automatically. An error is
given if any entry of p0 is negative. Default value of p0 is a vector of 1/n with
length of n. n is the length of p1.

See Also
ES.chisq.assoc

Examples
ES.chisq.gof(p1=c(10,20,30,40))
ES.chisq.gof(p1=c(0.1,0.2,0.3,0.4))

ES.chisq.gof(p1=c(10,20,30,40),p0=c(0.2,0.3,0.3,0.2))
ES.chisq.gof(p1=c(10,20,30,40),p0=c(20,30,30,20))
ES.chisq.gof(p1=c(0.1,0.2,0.3,0.4),p0=c(0.2,0.3,0.3,0.2))
ES.chisq.gof(p1=c(0.1,0.2,0.3,0.4),p0=c(20,30,30,20))

ES.proportions Compute effect size for a difference in proportions

Description
Compute effect size for a difference in proportions

Usage
ES.proportions(p1 = NULL, p2 = NULL, alternative = c("two.sided",
"one.sided"))

Arguments
p1 Proportion of sample one
p2 Proportion of sample two or a constant proportion
alternative The test is two sided or one sided

Examples
ES.proportions(0.65,0.45)

ES.proportions(0.25,0.05)
ES.t.one 5

ES.t.one Calculating effect size (Cohen’s d) of one-sample t test

Description
Calculating effect size (Cohen’s d) of one-sample t test

Usage
ES.t.one(m = NULL, sd = NULL, n = NULL, t = NULL, se = NULL,
df = NULL, mu = NULL, alternative = c("two.sided", "one.sided"))

Arguments
m mean of sample
sd standard deviation of sample
n number of observations
t t statistic
se standard error of sample 1
df degree of freedom
mu population mean
alternative The test is two sided or one sided

See Also
ES.t.two
ES.t.paired

Examples
## mean, sd and mu -> d
ES.t.one(m=-0.0938268,sd=0.9836668,mu=0)

## mean, se, n and mu -> d

ES.t.one(m=-0.0938268,se=0.1391115,n=50,mu=0)

## t and df -> d (df=n-1)

ES.t.one(t = -0.6745,df = 49)

## t and n -> d ((df=n-1))

ES.t.one(t = -0.6745,n = 50)
6 ES.t.paired

ES.t.paired Calculating effect size (Cohen’s d) of paired two-sample t test

Description

Calculating effect size (Cohen’s d) of paired two-sample t test

Usage

ES.t.paired(md = NULL, sd = NULL, n = NULL, t = NULL, se = NULL,

df = NULL, alternative = c("two.sided", "one.sided"))

Arguments

md mean difference (e.g., mean(x-y))

sd standard deviation of mean differences (e.g., sd(x-y))
n number of paires
t t statistic
se standard error of mean differences
df degree of freedom
alternative The test is two sided or one sided

See Also

ES.t.one
ES.t.two

Examples
## md, sd -> d
ES.t.paired(md=-0.08062384,sd=1.401886)

## md,se -> d
ES.t.paired(md=-0.08062384,se=0.1982566,n=50)

## t, df -> d
ES.t.paired(t=-0.4067,df=49)

## t, n -> d
ES.t.paired(t=-0.4067,n=50)
ES.t.two 7

ES.t.two Calculating effect size (Cohen’s d) of independent two-sample t test

Description
Calculating effect size (Cohen’s d) of independent two-sample t test

Usage
ES.t.two(m1 = NULL, m2 = NULL, sd1 = NULL, sd2 = NULL, n1 = NULL,
n2 = NULL, t = NULL, se1 = NULL, se2 = NULL, df = NULL,
alternative = c("two.sided", "one.sided"))

Arguments
m1 mean of sample 1
m2 mean of sample 2
sd1 standard deviation of sample 1
sd2 standard deviation of sample 2
n1 number of observations in sample 1
n2 number of observations in sample 2
t t statistic
se1 standard error of sample 1
se2 standard error of sample 2
df degree of freedom
alternative The test is two sided or one sided

See Also
ES.t.one
ES.t.paired

Examples
## mean, sd, n -> d
ES.t.two(m1=13.5,m2=5.5,sd1=4.1833,sd2=3.02765,n1=14,n2=10)

## mean se, n -> d

ES.t.two(m1=13.5,m2=5.5,se1=1.118034,se2=0.9574271,n1=14,n2=10)

## t and n -> d
ES.t.two(n1=14,n2=10,t=5.4349)

## t, df and n -> d
ES.t.two(t = 5.4349, df = 21.982,n1=14,n2=10)
8 power.chisq

## t and df -> d (assume n1=n2)

ES.t.two(t = 5.4349, df = 21.982)

power.anova.oneway Power calculations for balanced one-way analysis of variance tests

Description
Power calculations for balanced one-way analysis of variance tests

Usage
power.anova.oneway(groups = NULL, n = NULL, f = NULL, power = NULL,
sig.level = 0.05)

Arguments
groups Number of groups
n Number of observations (per group)
f Effect size, Cohen’s f
power power of study
sig.level significance level

See Also
ES.anova.oneway

Examples
power.anova.oneway(groups=4,n=20,f=0.28)

power.chisq Power calculations for chi-squared test

Description
Power calculations for chi-squared test

Usage
power.chisq(es = NULL, df = NULL, n = NULL, power = NULL,
sig.level = NULL)
power.plot.chisq 9

Arguments

es effect size. A numeric value or output of ES.chisq.gof, ES.chisq.assoc

df degree of freedom
n total number of observations
power power of study
sig.level significance level

See Also

ES.chisq.gof
ES.chisq.assoc
power.plot.chisq

Examples

## calculate power
power.chisq(es=0.16,df=1,n=530,sig.level=0.05)

## calculate sig.level
power.chisq(es=0.16,df=1,n=530,power=0.9576)

## calculate sample size

power.chisq(es=0.16,df=1,power=0.9576,sig.level=0.05)

## calculate effect size

power.chisq(df=1,n=530,power=0.9576,sig.level=0.05)

power.plot.chisq Power analysis plot of chi-squared test

Description

Power analysis plot of chi-squared test

Usage

power.plot.chisq(es = NULL, power = NULL, df = NULL, sig.level = NULL,

allele = FALSE, xlab = NULL, ylab = NULL, main = NULL, grid = FALSE,
type = c("np", "ne"))
10 power.proportions

Arguments
es effect size.
power power of study
df degree of freedom
sig.level significance level
allele in genetic association study, whether test allele or genotype
xlab a title for the x axis
ylab a title for the y axis
main an overall title for the plot
grid add grid lines or not
type "np": plot sample size vs. power; "ne": plot effevct size vs. sample size

See Also
power.chisq

Examples
## 'ne' type
### multiple effect size and multiple power
es=seq(from=0.1,to=0.5,by=0.1);
power=seq(from=0.7,to=0.9,by=0.1);
power.plot.chisq(es=es,power=power,df=1,sig.level=0.05,type="ne")
power.plot.chisq(es=es,power=power,df=1,sig.level=0.05,type="np")

### multiple effect size and single power

power.plot.chisq(es=seq(0.05,0.3,0.05),power=0.8,df=1,sig.level=0.05,type="ne")
power.plot.chisq(es=seq(0.05,0.3,0.05),power=0.8,df=1,sig.level=0.05,type="np")

### single effect size and single power

power.plot.chisq(es=0.2,power=0.8,df=1,sig.level=0.05,type="ne")
power.plot.chisq(es=0.2,power=0.8,df=1,sig.level=0.05,type="np")

### single effect size and multiple power

power.plot.chisq(es=0.2,power=seq(0.5,0.9,0.1),df=1,sig.level=0.05,type="ne")
power.plot.chisq(es=0.2,power=seq(0.5,0.9,0.1),df=1,sig.level=0.05,type="np")

power.proportions Power calculations for proportion tests (two-sided)

Description
Power calculations for proportion tests (two-sided)
power.t 11

Usage

power.proportions(n = NULL, h = NULL, power = NULL, sig.level = 0.05,

type = c("two", "one", "unequal"), ratio = 1)

Arguments

n Total number of observations

h Effect size, Cohen’s h
power Power of test
sig.level Significance level
type Type of proportion tests, must be one of "one","two" (default), or "unequal".
"one" means one sample proportion test. "two"/"unequal" means two sample
(equal size/unequal size) proportion test.
ratio The ratio of sample size 1 to sample size 2. Only will be used when ’type’ is
"unequal".

Examples
## one sample
power.proportions(n=600,h=0.3,type="one")

## two sample with same sample size

power.proportions(h=0.2,n=600)

## two sample with different sample size

power.proportions(h=0.2,n=1200,type="unequal",ratio=2)

power.t Power calculations for t-test

Description

Power calculations for t-test

Usage

power.t(es = NULL, n = NULL, power = NULL, sig.level = NULL,

ratio = 1, type = c("two", "paired", "one", "unequal"),
alternative = c("two.sided", "left", "right"))
12 power.t

Arguments
es effect size.
n total number of observations/pairs
power power of study
sig.level significance level
ratio the ratio of sample size 1 to sample size 2. Only will be used when ’type’ is
"unequal".
type type of t test, must be one of "one","two" (default), "paired", or "unequal". "one"
means one sample t test, which test whether the population mean is equal to a
specified value. "two"/"unequal" means two sample (equal size/unequal size) t
test, which is used to ascertain how likely an observed mean difference between
two groups would be to occur by chance alone. "paired" means paired t-test
(also called the correlated t-test and the t-test for dependent means), which is
used to ascertain how likely the difference between two means that contain the
same (or matched) observations is to occur by chance alone.
alternative One- or two-sided test, must be one of "two.sided" (default), "left", "right"

See Also
ES.t.one
ES.t.two
ES.t.paired

Examples
## one sample two sided test, calculate power
power.t(es=0.2,n=60,sig.level=0.10,type="one",alternative="two.sided")

## one sample one sided (left tail) test, calculate power

power.t(es=0.2,n=60,sig.level=0.10,type="one",alternative="left")

## one sample one sided (right tail) test, calculate power

power.t(es=0.2,n=60,sig.level=0.10,type="one",alternative="right")

## one sample two sided test, calculate sampe size

power.t(es=0.2,power=0.8,sig.level=0.05,type="one",alternative="two.sided")

## one sample two sided test, calculate effect size

power.t(n=200,power=0.8,sig.level=0.05,type="one",alternative="two.sided")

## one sample two sided test, calculate sig.level

power.t(es=0.2,n=200,power=0.8,type="one",alternative="two.sided")

## paired sample two sided test, calculate power

power.t(es=0.559,n=40,sig.level=0.05,type="paired",alternative="two.sided")

## paired sample two sided test, calculate sample size

power.t(es=0.15,power=0.8,sig.level=0.05,type="paired",alternative="two.sided")
power.t 13

## paired sample two sided test, calculate effect size

power.t(n=200,power=0.8,sig.level=0.05,type="paired",alternative="two.sided")

## two sample two sided test, calculate power

power.t(es=0.15,n=300,sig.level=0.05,type="two",alternative="two.sided")

## two sample two sided test, calculate sample size

power.t(es=0.15,power=0.8,sig.level=0.05,type="two",alternative="two.sided")

## two sample two sided test, calculate effect size

power.t(n=300,power=0.8,sig.level=0.05,type="two",alternative="two.sided")

## two sample (unequal size), calculate sample size

power.t(es=0.15,power=0.8,sig.level=0.05,type="unequal",ratio=2,alternative="two.sided")

power.t(es=0.1,n=3000,sig.level=0.05,type="unequal",ratio=2,alternative="two.sided")
Index

ES.anova.oneway, 2, 8
ES.chisq.assoc, 2, 4, 9
ES.chisq.gof, 3, 3, 9
ES.proportions, 4
ES.t.one, 5, 6, 7, 12
ES.t.paired, 5, 6, 7, 12
ES.t.two, 5, 6, 7, 12

power.anova.oneway, 8
power.chisq, 8, 10
power.plot.chisq, 9, 9
power.proportions, 10
power.t, 11

14