CODEWITHSUNDEEP
rcorrelation
Mikko
Mikko

Reputation: 7755

Substitute results to an error with cor.test

I am trying to write a loop, which returns linear regression and correlation parameters. When trying to pass substitute through cor.test function, I encounter an unexpected error

data(iris)
i <- 1
vars <- list(par = as.name(colnames(iris)[1]), expl = as.name(colnames(iris)[2:4][i]))
lm(substitute(par ~ expl, vars), data = iris) # works
lm(Sepal.Length ~ Sepal.Width, data = iris) # works. Result equal to the statement above

cor.test(~Sepal.Length + Sepal.Width, data = iris) # works
cor.test(substitute(~par + expl, vars), data = iris) # does not work
## Error in cor.test.default(substitute(~par + expl, vars), data = iris) :  
## argument "y" is missing, with no default

To my understanding the cor.test statement should be the same than the manually inputted one.

What is the reason for the error? How can I write a substitute statement for cor.test that works?

Upvotes: 0

Views: 770

Answers (1)

Eldioo
Eldioo

Reputation: 522

The error stems from the fact that the first version is of formula type and the second one is language:

str(substitute(~par + expl, vars))
# language ~Sepal.Length + Sepal.Width
str(~Sepal.Length + Sepal.Width)
# Class 'formula'  language ~Sepal.Length + Sepal.Width
# ..- attr(*, ".Environment")=<environment: R_GlobalEnv>

If you use as.formula on the second version it works:

cor.test(as.formula(substitute(~par + expl, vars)), data = iris)
# Pearson's product-moment correlation
# 
# data:  Sepal.Length and Sepal.Width
# t = -1.4403, df = 148, p-value = 0.1519
# alternative hypothesis: true correlation is not equal to 0
# 95 percent confidence interval:
# -0.27269325  0.04351158
# sample estimates:
# cor 
# -0.1175698

Upvotes: 1

Related Questions