Introduction
To R

Lists

Lists are rather curious things... They are used to bundle several things up into one object. The things can be vectors, matrices, numbers or characters, or even lists themselves. Here's an example:

> person <- list(age=21,name='Fred',score=c(65,78,55))
> person
$age:
[1] 21

$name:
[1] "Fred"

$score:
[1] 65 78 55

The list is constructed with the list() function. The components of the list are introduced using the usual name=arg form of function arguments. When you print the list object you are shown each component with its name and value.

You can access each element by its name, preceded by a $:

> person$name
[1] "Fred"       name
> person$score[2]      second element of $score
[1] 78

You can also access each element by number, using double-square brackets:

> person[[1]]
[1] 21
> person[[3]]
[1] 65 78 55

Many Things Return Lists

A lot of R functions will return a list. For example, the t.test() function will return an object which is a list. Try this - a simple t-test of two sets of random numbers with different means:

> tt <- t.test(rnorm(1000,mean=1),rnorm(1000,mean=1.2),var.equal=T)

Nothing is printed because the result is being stored in tt - so type its name to print it out:

> tt

         Standard Two-Sample t-Test 

data:  rnorm(1000, mean = 1) and rnorm(1000, mean = 1.2) 
t = -4.5493, df = 1998, p-value = 0 
alternative hypothesis: true difference in means is not equal to 0 
95 percent confidence interval:
 -0.2923768 -0.1162318 
sample estimates:
 mean of x mean of y 
 0.9864194  1.190724

Now this is what we saw when we did the t-test earlier. Doesn't look much like the list printout we saw earlier? But it is a list:

> is.list(tt)
[1] TRUE

You can get the names of components of a list with the names() function - for this t-test object we get:

> names(tt)
[1] "statistic"   "parameters"  "p.value"     "conf.int"    "estimate"   
[6] "null.value"  "alternative" "method"      "data.name"

and so we can get at the confidence interval, for example, with:

> tt$conf.int
[1] -0.2923768 -0.1162318
attr(, "conf.level"):
[1] 0.95

This is a vector of length two, which shows us the 95% confidence interval doesn't contain zero. But whats this attr(,"conf.level") stuff? Thats something called an additional attribute of the vector - in this case it tells is what confidence level the interval is calculated at.

Contents