Useful R Packages for Statistic Analysis

A brief summary of R packages (and corresponding functions) that are used in the book “Discovering Statistics using R (2012)” by Andy Field.

Packages and functions (order by alphabetical):

• boot - for bootstrap
  • boot(), see section 6.5.7, such as boot_kendall<-boot(liarData, bootTau, 2000)
  • boot.ci(), confidence interval, such as boot.ci(boot_kendall, conf=0.99), see section 6.5.7
• car - for Levene’s test, Type III sums of squares, and more
  • leveneTest(), such as leveneTest(viagraData$libido, viagraData$dose, center = median)
  • Anova(), such as Anova(modelName, type="III"), see section 11.4.7
  • durbinWatsonTest() or dwt(), Durbin–Watson test for assumption of independent errors, see section 7.9.3
• compute.es - for effect sizes
  • mes(), see section 11.6, calculate effect sizes between all combinations of groups
• effects - for adjusted means
  • effect(), see section 11.4.8
• ez - for ANOVA
  • ezANOVA(), repeated-measures ANOVA, see section 12.4.7
• ggplot2 - for graphs
• ggm - for partial correlation
  • pcor(), partial correlation, see section 6.6.2
  • pcor.test(), significance of partial correlation, see section 6.6.2
• gmodels - for chi-square
  • CrossTable(), see section 18.6.4.
• Hmisc - for correlation
  • rcorr(), for correlation, see section 6.5.3
• MASS - for loglinear analysis
• mlogit - for multinomial logistic regression
• multcomp - for post hoc tests
  • glht(), Tukey tests, see section 11.4.11.
• pastecs - for descriptive statistics
  • stat.desc(), such as by(viagraData$libido, viagraData$dose, stat.desc)
• polycor - for correlation
  • polyserial(), biserial correlation, see section 6.5.8
• psych -
  • describe() - such as describe(dlf$day1), similar to stat.desc() above.
• QuantPsyc - to get standardized regression coefficients
• reshape2 - for reshape
  • melt()
• stats - built-in, auto-loaded
  • wilcox.test()
  • shapiro.test(), Shapiro-Wilk test, such as shapiro.test(variable)
  • cor(), for correlation
  • cor.test(), for correlation
  • anova(), compare models, see section 7.8.4.2; which is different from Anova() from car package
  • confint(), computes confidence interval
• WRS - for robust tests, see section 5.8.4
  • Updated website: http://dornsife.usc.edu/labs/rwilcox/software/
  • source("http://dornsife.usc.edu/assets/sites/239/docs/Rallfun-v34.txt") – new website. Since all functions become available in R environment, there is no need to call library(WRS).
  • ancova() and ancboot(), see section 11.5