[R] Puzzled in utilising summary.lm() to obtain Var(x)

[Prof Brian Ripley]

On Wed, 15 Jun 2005, Ajay Narottam Shah wrote:

>>> I have a program which is doing a few thousand runs of lm(). Suppose
>>> it is a simple model
>>>   y = a + bx1 + cx2 + e
>>>
>>> I have the R object "d" where
>>>   d <- summary(lm(y ~ x1 + x2))
>>>
>>> I would like to obtain Var(x2) out of "d". How might I do it?
>>>
>>> I can, of course, always do sd(x2). But it would be much more
>>> convenient if I could snoop around the contents of summary.lm and
>>> extract Var() out of it. I couldn't readily see how. Would you know
>>> what would click?
>>
>> Is the question how to get the variance of a column of the
>> model matrix for a model that is the sum of terms given only
>> summary output and the column name but not the name of the
>> data frame?  If that is it then try this:
>>
>> d <- summary(lm(Sepal.Length ~ Sepal.Width, iris)) # test data
>> var(model.matrix(eval(d$call))[,"Sepal.Width"])
>
> Yes, this is indeed exactly what I was looking for :-) Thanks,
>
> The eval() pays the full cost of running d$call?

A better way is

> d0 <- lm(Sepal.Length ~ Sepal.Width, iris)
> var(model.matrix(d0)[,"Sepal.Width"])

but if you really only have the summary  (it's wasteful to keep the 
summary here, as it is larger than the fit)

> var(model.matrix(structure(d, class="lm"))[,"Sepal.Width"])


-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595