Similar to this question, but not exactly the same.
result test1 test2 test3 test4
a 13 12 12 8
b 5 9 10 8
c 39 24 30 29
How can I use purr for row and columnwise prop.tests comparing test1 to test2:test4 by result?
Manually I guess it would look like this for each test:
result: a, test1, test2: prop.test(c(13, 12), c(57, 45))
result: c, test1, test4: prop.test(c(30, 29), c(57, 45))
# example data
dt = read.table(text = "
result test1 test2 test3 test4
a 13 12 12 8
b 5 9 10 8
c 39 24 30 29
", header=T, stringsAsFactors=F)
# function to get prop.test
# based on a row and two columns
GetTest = function(r, t1, t2) {
prop.test(unlist(dt[dt$result == r, c(t1, t2)]),
colSums(dt[,c(t1, t2)]))
}
# apply function
GetTest("a","test1","test2")
# 2-sample test for equality of proportions with continuity correction
#
# data: unlist(dt[dt$result == r, c(t1, t2)]) out of colSums(dt[, c(t1, t2)])
# X-squared = 0.047595, df = 1, p-value = 0.8273
# alternative hypothesis: two.sided
# 95 percent confidence interval:
# -0.2274729 0.1502799
# sample estimates:
# prop 1 prop 2
# 0.2280702 0.2666667