Type: Package
Title: Permutation Tests for Randomization Model
Version: 0.1.3
Description: Perform permutation-based hypothesis testing for randomized experiments as suggested in Ludbrook & Dudley (1998) <doi:10.2307/2685470> and Ernst (2004) <doi:10.1214/088342304000000396>, introduced in Pham et al. (2022) <doi:10.1016/j.chemosphere.2022.136736>.
Depends: R (≥ 4.3.0)
Imports: magrittr (≥ 2.0.3), dplyr (≥ 1.1.2), lme4 (≥ 1.1-33), parameters (≥ 0.21.0), emmeans (≥ 1.8.6), stats, utils
Suggests: multcompView (≥ 0.1-9)
BugReports: https://github.com/phamdn/peramo/issues
License: GPL-3
Encoding: UTF-8
RoxygenNote: 7.2.3
LazyData: true
NeedsCompilation: no
Packaged: 2023-05-28 10:26:10 UTC; nghia
Author: Duy Nghia Pham ORCID iD [aut, cre], Inna M. Sokolova ORCID iD [ths]
Maintainer: Duy Nghia Pham <nghiapham@yandex.com>
Repository: CRAN
Date/Publication: 2023-05-31 16:20:02 UTC

peramo: Permutation Tests for Randomization Model

Description

Perform permutation-based hypothesis testing for randomized experiments as suggested in Ludbrook & Dudley (1998) doi:10.2307/2685470 and Ernst (2004) doi:10.1214/088342304000000396, introduced in Pham et al. (2022) doi:10.1016/j.chemosphere.2022.136736.

Copyright

peramo: Permutation Tests for Randomization Model.
Copyright (C) 2022-2023 Duy Nghia Pham & Inna M. Sokolova

peramo is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

peramo is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with peramo. If not, see https://www.gnu.org/licenses/.

Author(s)

Duy Nghia Pham & Inna M. Sokolova

Two-Group Permutation Test

Description

AB performs A/B testing for two-group experiments.

Usage

AB(a, b, rand = 9999, seed = 1)

Arguments

a

the measurement of responses of the first group.

b

the measurement of responses of the second group.

rand

an integer, the number of randomization samples. The default value is 9999.

seed

an integer, the seed for random number generation. Setting a seed ensures the reproducibility of the result. See set.seed for more details.

Value

AB returns an one-row data frame with 6 columns:

nA

the sample size of the first group.

mean.A

the mean responses of the first group.

nB

the sample size of the second group.

mean.B

the mean responses of the second group.

mean.dif

the difference between two mean responses.

pval

the p-value.

References

Ernst, M. D. (2004). Permutation Methods: A Basis for Exact Inference. Statistical Science, 19(4), 676–685. doi:10.1214/088342304000000396.

Examples

AB(c(19, 22, 25, 26), c (23, 33, 40))

Permutation Test for Correlation Between Paired Samples

Description

XY performs permutation test on correlation coefficients.

Usage

XY(a, b, rand = 9999, seed = 1, method = c("pearson", "kendall", "spearman"))

Arguments

a

a numeric vector, the first variable.

b

a numeric vector, the second variable.

rand

an integer, the number of randomization samples. The default value is 9999.

seed

an integer, the seed for random number generation. Setting a seed ensures the reproducibility of the result. See set.seed for more details.

method

correlation coefficient, "pearson", "kendall", or "spearman".

Value

XY returns an one-row data frame with 2 columns:

cor

the correlation coefficient.

pval

the p-value.

See Also

cor.test

Examples

with(subset(ctm_Cu, run == "Jan"), XY(sediment, porewater))

Biomarker Responses of the Ragworms to Copper and Warming

Description

Biomarker Responses of the Ragworms to Copper and Warming

Usage

ctm_Cu

ctm_worm

Format

An object of class data.frame with 60 rows and 7 columns.

An object of class data.frame with 210 rows and 26 columns.

Calculate the Differences between Means

Description

Calculate the Differences between Means

Usage

diffcalc(vec, control)

Arguments

vec

a numeric vector, the mean responses.

control

a logical, whether the control group exists.

Biomarker Responses of the Blue Mussels to Organic UV Filters

Description

Biomarker Responses of the Blue Mussels to Organic UV Filters

Usage

mussel_SoS

mussel_gill

mussel_digest

Format

An object of class data.frame with 15 rows and 7 columns.

An object of class data.frame with 120 rows and 24 columns.

An object of class data.frame with 120 rows and 24 columns.

Source

Pham, D. N., Sokolov, E. P., Falfushynska, H., & Sokolova, I. M. (2022). Gone with sunscreens: Responses of blue mussels (Mytilus edulis) to a wide concentration range of a UV filter ensulizole. Chemosphere, 309, 136736. doi:10.1016/j.chemosphere.2022.136736.

Compare the Differences with Critical Values

Description

Compare the Differences with Critical Values

Usage

nolesser(obs, cric)

Arguments

obs

a numeric, the observed difference.

cric

a numeric, the critical values of maximum absolute differences.

One-Way Layout Permutation Test

Description

owl performs the global test and multiple comparisons for single factor experiments.

Usage

owl(df, rand = 9999, alpha.post = 0.05, type.post = "control", seed = 1)

Arguments

df

a data frame with the name of experimental groups as the first column and the measurement of responses as the remaining columns.

rand

an integer, the number of randomization samples. The default value is 9999.

alpha.post

a numeric, the Type I error rate for multiple comparisons. The default value is 0.05.

type.post

the way of multiple comparisons, "all" for pairwise comparisons or "control" for only comparisons with the control group.

seed

an integer, the seed for random number generation. Setting a seed ensures the reproducibility of the result. See set.seed for more details.

Details

The first name appearing in the first column will determine the control group. The other names will be treatment groups.

Value

owl returns a list with 9 components:

n.obs

the sample sizes.

avg.obs

the mean responses.

T.obs

the T statistic for global test.

pval

the p-value for global test.

pval.round

the reported form of p-value.

main.test

the strength of evidence against the null hypothesis.

d.multi.obs

the differences in means for multiple comparisons.

mad.cric

the critical value of maximum absolute differences in means.

post.test

TRUE if the differences are significant.

References

Ernst, M. D. (2004). Permutation Methods: A Basis for Exact Inference. Statistical Science, 19(4), 676–685. doi:10.1214/088342304000000396.

Muff, S., Nilsen, E. B., O’Hara, R. B., & Nater, C. R. (2022). Rewriting results sections in the language of evidence. Trends in Ecology & Evolution, 37(3), 203–210. doi:10.1016/j.tree.2021.10.009.

Examples

ernst2004 <- data.frame(
group = factor(rep(c("style1", "style2", "style3"), each = 5 ),
levels = c("style1", "style2", "style3")),
speed = c( 135,91,111,87, 122, 175,130,514,283, NA,105,147,159,107,194))
owl(ernst2004, type.post = "all")

Test Statistics for One-Way Layout Permutation Test

Description

owlStat computes statistics for owl. This is not meant to be called directly.

Usage

owlStat(lov, env = parent.frame())

Arguments

lov

a list of vectors, responses by experimental groups.

env

an environment, to access outer scope variables.

Value

owlStat returns a list with 5 components:

n

the sample sizes.

avg

the mean responses.

T

the T statistic for global test.

d.multi

the differences in means for multiple comparisons.

mad

the maximum absolute differences in means.

.

References

Ernst, M. D. (2004). Permutation Methods: A Basis for Exact Inference. Statistical Science, 19(4), 676–685. doi:10.1214/088342304000000396.

Permutation Test for Two-Way Layout with Extra Factors

Description

tw_complex performs the permutation test for ANOVA of two-factor experiments with complex design.

Usage

tw_complex(df, res, mains, nested, nuis, seed = 1, rand = 1999, emm = FALSE)

Arguments

df

a data frame with at least three columns.

res

a character string, name of response variable.

mains

two character strings, names of two main factors.

nested

(optional) a character string, name of the nested factor.

nuis

(optional) a character string, name of the nuisance factor.

seed

an integer, the seed for random number generation. Setting a seed ensures the reproducibility of the result. See set.seed for more details.

rand

an integer, the number of randomization samples. The default value is 1999.

emm

a logical, whether to compute estimated marginal means.

Details

res, mains, nested, and nuis refer to column names in df. While nuis column must be a numeric vector, mains and nested columns must be factors. res can be a numeric or logical vector.

tw_complex currently support linear models with only mains, generalized linear mixed-effects models with mains and nested, and linear mixed-effects models with mains, nested, and nuis.

Value

tw_complex returns a list with 3 main components:

lm, glmer, or lmer

model results.

anova

anova table.

perm

permutation test results with F-statistics, p-values, and strength of evidence.

References

Manly, B. F. J. (2007). Randomization, bootstrap, and Monte Carlo methods in biology (3rd ed). Chapman & Hall/ CRC.

Ernst, M. D. (2004). Permutation Methods: A Basis for Exact Inference. Statistical Science, 19(4), 676–685. doi:10.1214/088342304000000396.

Anderson, M., & Braak, C. T. (2003). Permutation tests for multi-factorial analysis of variance. Journal of Statistical Computation and Simulation, 73(2), 85–113. doi:10.1080/00949650215733.

See Also

lm, glmer, and lmer.

Examples


tw_complex(df = subset(ctm_Cu, run == "Jan",
select = c("copper", "temp", "sediment")),
res = "sediment",
mains = c("copper", "temp"))


#might take more than 5s in some machines

Two-Way Layout Permutation Test

Description

twl performs the global test and multiple comparisons for two-factor experiments.

Usage

twl(
  df,
  rand = 4999,
  seed = 1,
  mult = FALSE,
  simple = TRUE,
  control = TRUE,
  alpha = 0.05
)

Arguments

df

a data frame with the first and second columns containing the levels of the two main factors and the third column containing the measurement of responses.

rand

an integer, the number of randomization samples. The default value is 4999.

seed

an integer, the seed for random number generation. Setting a seed ensures the reproducibility of the result. See set.seed for more details.

mult

a logical, whether to perform multiple comparisons.

simple

a logical, whether to perform comparisons for simple effects.

control

a logical, whether to perform only comparisons with the control group.

alpha

a numeric, the Type I error rate for multiple comparisons. The default value is 0.05.

Details

The first levels appearing in the first and second columns will determine the control groups (if any). The other levels will be treatment groups.

Value

twl returns a list with possible components:

n, n.main1, and n.main2

the sample sizes.

avg, avg.main1, and avg.main2

the mean responses.

Fs

the F statistics, p-values, reported form of p-value, and strength of evidence against the null hypotheses.

d.main1sub and d.main2sub or d.main1 and d.main2

the differences in means for multiple comparisons.

mad.main1sub.cric and mad.main2sub.cric or mad.main1.cric and mad.main2.cric

the critical value of maximum absolute differences in means.

mult.test.main1sub and mult.test.main2sub or mult.test.main1 and mult.test.main2

TRUE if the differences are significant.

References

Manly, B. F. J. (2007). Randomization, bootstrap, and Monte Carlo methods in biology (3rd ed). Chapman & Hall/ CRC.

Ernst, M. D. (2004). Permutation Methods: A Basis for Exact Inference. Statistical Science, 19(4), 676–685. doi:10.1214/088342304000000396.

Muff, S., Nilsen, E. B., O’Hara, R. B., & Nater, C. R. (2022). Rewriting results sections in the language of evidence. Trends in Ecology & Evolution, 37(3), 203–210. doi:10.1016/j.tree.2021.10.009.

Motulsky, H. (2020). GraphPad Statistics Guide. GraphPad Software Inc. https://www.graphpad.com/guides/prism/latest/statistics/index.htm.

Examples


manly2007 <- data.frame(
month = factor(rep(c("jun", "jul", "aug", "sep"), each = 6 ),
levels = c("jun", "jul", "aug", "sep")),
size = factor(rep(c("small", "large"), each = 3, times = 4),
levels = c("small", "large")),
consume = c( 13,242,105,182,21,7,8,59,20,24,312,68,515,488,88,460,1223,990,18,44,21,140,40,27))
twl(manly2007)
twl(manly2007, mult = TRUE, simple = TRUE, control = FALSE)
 #might take more than 5s in some machines

Test Statistics for Two-Way Layout Permutation Test

Description

twlStat computes statistics for twl. This is not meant to be called directly.

Usage

twlStat(df, env = parent.frame())

Arguments

df

a data frame with the levels of the two main factors as the first and second columns and the measurement of responses as the third column.

env

an environment, to access outer scope variables.

Value

twlStat returns a list with at least 4 components:

Fs

the F statistics for global test.

F.main1 and F.main2

the F statistics for the first main factor and the second main factor.

F.int

the F statistic for the interaction.

In case of multiple comparisons, additional components are:

avg or avg.main1 and avg.main2

the mean responses for multiple comparisons.

d.main1sub and d.main2sub or d.main1 and d.main2

the differences in means.

mad.main1sub and mad.main2sub or mad.main1 and mad.main2

the maximum absolute differences in means.

References

Manly, B. F. J. (2007). Randomization, bootstrap, and Monte Carlo methods in biology (3rd ed). Chapman & Hall/ CRC.