[R] truncated regressor

[Emmanuel Paradis]

At 12:29 15/05/00 +0200, Peter Dalgaard wrote:
>> which is clear to me: doing lm(y ~ a + I(b+c)) is similar to
>> 
>> 	d <- b + c
>> 	lm(y ~ a + d)
>> 
>> But what if the expression in I() is a logical? After playing a moment, I
>> found out that a formula can include a logical variable, and treats it as a
>> factor:
>> 
>> > x <- runif(50, 0, 50)
>> > y <- 2*x + rnorm(50, 0, 5)
>> > lm(y ~ x < 25)
>> 
>> Call:
>> lm(formula = y ~ x < 25)
>> 
>> Coefficients:
>> (Intercept)       x < 25  
>>       78.24       -53.92  
>> 
>
>I should hope that it is being converted to *numeric* (TRUE --> 1),
>not factor, otherwise there'd be all sorts of trouble with contrasts.
>This is a general feature: TRUE + 2 == 3 for instance, and (x>0)-(x<0)
>is the sign function. 
>
>[Easiest way of checking is to compare
>
>lm(y ~ I(x < 25) - 1) 
>
>with 
>
>lm(y ~ factor(x < 25) - 1)   
>
>]

You are right. I was misled by the fact that

  lm(y ~ as.factor(x < 25))

and

  lm(y ~ x < 25)

give the same result.


Emmanuel Paradis
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._