# Mean, Standard Deviation, Etc ??mean summary(hsb$read) (read.m <- mean(hsb$read)) read.sd <- sd(hsb$read) read.std <- (hsb$read - read.m)/read.sd read.std.2 <- scale(hsb$read) library(lattice) # for graphics functions densityplot(read.std) summary(hsb) mean(hsb) # produces some WARNINGS read.na <- hsb$read[1:10] read.na[c(3,7,8)] <- NA read.na mean(read.na) mean(read.na, na.rm=TRUE) # Means etc. by group tapply(hsb$read, hsb$gender, mean) by(hsb, hsb$gender, mean) # produces some WARNINGS # t-tests ??ttest ??t.test t.test(hsb$read) t.test(hsb$read, mu=50) t.test(read~gender, data=hsb) # grouping variable t.test(read~gender, data=hsb, var.equal=TRUE) # the classic test densityplot(~read |gender, data=hsb) # correlation with(hsb, cor(read, write)) scores <- subset(hsb, select=c("read","write","math","science","socst")) cor(scores) hsb.type <- unlist(lapply(hsb, class)) scores2 <- subset(hsb, select=(hsb.type=="integer")) scores2$id <- NULL cor(scores2) cor.test(~read+write, data=hsb) # one-way ANOVA ?oneway.test oneway.test(socst~race, data=hsb) oneway.test(socst~race, data=hsb, var.equal=TRUE) oneway.test(socst~gender+race, data=hsb) # oneway? oneway.test(socst~gender:race, data=hsb) # better coding, more clear oneway.test(socst~gender:race, data=hsb, var=T) anova(lm(socst~gender:race, data=hsb))