Herman
Herman

Reputation: 25

R How to make working t.test command line into function?

I have a code line which works independently, but I am trying to make it into a function which does not work. Data set:

cooper <- data.frame(preDist=c(2454, 2666, 2153, 2144, 2957, 2407, 2167, 2259,
                               1993, 2351, 1642, 2121, 2603, 2669, 2064),
                     postDist=c(2763, 2710, 2272, 2342, 3256, 2617, 2515, 2469,
                                2257, 2637, 1597, 2331, 2616, 2679, 2114),
                     group=factor(c(1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3),
                                  labels=c("Group1", "Group2", "Cont")))

Working code:

t.test(cooper$postDist[cooper$group == "Group1"], 
       cooper$preDist[cooper$group == "Group1"], 
       alternative = "greater", 
       paired = TRUE)$p.value

This returns correct value for my chosen group (Group1)

Not-working function:

pairtest <- function(grp) {
  pvalue <- t.test(cooper$postDist[cooper$group == "grp"], 
                   cooper$preDist[cooper$group == "grp"], 
                   alternative = "greater", paired = TRUE)$p.value
  return(pvalue)
      }
pairtest(Group1)

Reports "not enough 'x' observations".

Upvotes: 0

Views: 46

Answers (1)

e.matt
e.matt

Reputation: 886

pairtest <- function(grp,df) { # add data frame to your input

with(df[df$group == grp,], # filter data frame on input
t.test(preDist,postDist,alternative="greater",paired = T)$p.value)
#changed pre to preDist
#changed post to postDist
}
pairtest("Group1",cooper)

Upvotes: 1

Related Questions