Download Cheatsheet R Users ANOVA and more Summaries Design in PDF only on Docsity! Introduction to R 2016-‐2017 Cheatsheet – Analysis of Variance ….. 0. 2016-‐2017\Cheatsheet R users ANOVA.docx Page 1 of 3 Introduction to R 2016-‐2017 Cheat Sheet – Analysis of Variance (2 way factorial anova, actually) Dear R learner, This is a work in progress, so please do not consider this complete. All suggestions and additions are very welcome – cb. Cheat sheet by example Response Y = yvar Factor I predictor is drug (1=control, 2=tx, 3=other) Predictor II predictor is season(1=winter, 2=spring, 3=summer, 4=fall) Get your data into R (3 ways) setwd(“/Users/cbigelow/Desktop/”) install.packages(“openxlsx”) # Excel data (.xlsx) library(openxlsx) dat <- read.xlsx(“myexceldata.xlsx.dta”) install.packages(“readstata13”) library(readstata13) dat <- read.dta13(“mystatadata.dta”, convert.factors=FALSE) # Stata data (.dta) install.packages(“haven”) library(haven) dat <- read_sas(“mysasdata.sas7bdat”) # SAS data (.sas7bdat) Label factor levels. levels(dat$drug) # List levels levels(dat$drug) = c(“control”, “tx”, “other”) # Give levels names Tell R that the predictors are FACTORS dat$drug <- as.factor(dat$drug) dat$season <- as.factor(dat$season) Introduction to R 2016-‐2017 Cheatsheet – Analysis of Variance ….. 0. 2016-‐2017\Cheatsheet R users ANOVA.docx Page 2 of 3 Produce descriptives, separately for groups defined by Factor I x Factor II require(stats) tapply(dat$yvar, list(dat$drug, dat$season), mean) # 2 way table of means (handy) install.packages(“doBy”) library(doBy) options(digits=6) summaryBy(yvar ~ drug + season, data=dat, FUN=c(length, mean, sd), fun.names=c(“n”, “mean”, “sd”)) # n, mean, sd within all groups One way side-‐by-‐side box plots using ggplot( ) install.packages(“ggplot2”) library(ggplot2) blank <- ggplot(data=dat, aes(x=season, y=yvar)) # Initialize. Nothing plotted yet blank2 <- blank + labs(title=”line1\nline2”) # Note \n to obtain 2 line title points <- blank2 + geom_point( ) # Plot scatterplot pointsbox <- points + geom_boxplot(aes(color=season)) # Overlay box plot Graph the Main Effects (not something you’d publish but handy) plot.design(yvar ~ drug*season, data=dat, main=”Main Effects Plot”) Graph the Interaction (also not pretty enough for publishing but handy) install.packages(“sciplot”) library(sciplot) lineplot.CI(x.factor=drug, response=yvar, group=season, data=dat, trace.label=”Season”, xlab=”Drug”, ylab=”Mean of yvar”, main=”Interaction Plot”) Fit 2 Way anova (main effects + interaction) fit <- aov(yvar ~ drug*season, data=dat) # Option 1 fit <- aov(yvar ~ drug + season + drug:season, data-dat) # Option 2 (I prefer this) anova(fit) # show anova table summary(fit) # show more stuff